Front. Environ. Sci. Eng.

2023, 17(4): 44
https://doi.org/10.1007/s11783-023-1644-x

REVIEW ARTICLE

State-of-the-art applications of machine learning in the life

cycle of solid waste management

Rui Liang1, Chao Chen1, Akash Kumar1, Junyu Tao (✉)2, Yan Kang2, Dong Han2, Xianjia Jiang2,
Pei Tang2, Beibei Yan1,3, Guanyi Chen2,4
1 School of Environmental Science and Engineering, Tianjin University, Tianjin 300350, China
2 School of Mechanical Engineering, Tianjin University of Commerce, Tianjin 300134, China
3 Tianjin Key Laboratory of Biomass Wastes Utilization/Tianjin Engineering Research Center of Bio Gas/Oil Technology, Tianjin 300072, China
4 School of Science, Tibet University, Lhasa 850012, China

HIGHLIGHTS GRAPHIC ABSTRACT

● State-of-the-art applications of machine
learning (ML) in solid waste (SW) is presented.
● Changes of research field over time, space, and
hot topics were analyzed.
● Detailed application seniors of ML on the life
cycle of SW were summarized.
● Perspectives towards future development of
ML in the field of SW were discussed.

ARTICLE INFO
Article history:
Received 28 May 2022
Revised 18 August 2022
Accepted 6 September 2022 ABSTRACT
Available online 25 October 2022 Due to the superiority of machine learning (ML) data processing, it is widely used in research of solid
waste (SW). This study analyzed the research and developmental progress of the applications of ML
in the life cycle of SW. Statistical analyses were undertaken on the literature published between 1985
Keywords: and 2021 in the Science Citation Index Expanded and Social Sciences Citation Index to provide an
Machine learning (ML) overview of the progress. Based on the articles considered, a rapid upward trend from 1985 to 2021
Solid waste (SW) was found and international cooperatives were found to have strengthened. The three topics of ML,
Bibliometrics namely, SW categories, ML algorithms, and specific applications, as applied to the life cycle of SW
SW management were discussed. ML has been applied during the entire SW process, thereby affecting its life cycle.
Energy utilization ML was used to predict the generation and characteristics of SW, optimize its collection and
Life cycle transportation, and model the processing of its energy utilization. Finally, the current challenges of
applying ML to SW and future perspectives were discussed. The goal is to achieve high economic and
environmental benefits and carbon reduction during the life cycle of SW. ML plays an important role
in the modernization and intellectualization of SW management. It is hoped that this work would be
helpful to provide a constructive overview towards the state-of-the-art development of SW disposal.

© Higher Education Press 2023

1 Introduction
✉ Corresponding author
E-mail: [email protected] Solid waste (SW) disposal is an important international
Special Issue—Artificial Intelligence/Machine Learning on Environ- issue in terms of resources and environmental aspects.
mental Science & Engineering (Responsible Editors: Yongsheng Chen, With respect to resources, the materials and energy
Xiaonan Wang, Joe F. Bozeman III & Shouliang Yi) recovered from SW could contribute to the sustainable
2 Front. Environ. Sci. Eng. 2023, 17(4): 44

and carbon-neutral development of human societies. The predict organic SW treatment and recycling processes.
energy utilization of SW has a significant impact on Although these studies are insightful and informative, it is
energy shortage (Zhang et al., 2021). With respect to the difficult for the readership to obtain comprehensive
environment, mass accumulation and improper disposal knowledge of the entire life cycle of SW because it
of SW threaten the atmosphere, water, ecosystems, and contains too many divisions, making it difficult to have
human health via the food chain (Liu et al., 2021; Shi an objective discussion based on the existing literature.
et al., 2022). Its importance has led to growing concerns Accordingly, a quantitative assessment of the published
worldwide. To enhance the overall performance of SW literature is necessary to provide an overview of ML
recycling and disposal, it is necessary to take action at applied during the life cycle of SW, as this will be helpful
each point during the life cycle of SW, which can be for policymakers and researchers. In addition, quantita-
divided into: 1) generation and characterization, 2) tive assessment of this field is rare, and this study uses the
collection and transportation, and 3) treatment and bibliometric method to facilitate a review on the use of
utilization. Recently, processes related to these three ML during the life cycle of SW. Thereafter, based on the
divisions have experienced an increasing demand for bibliometric analysis, deep perspectives on the research
data-driven techniques. Policymakers urgently need progress and future development were provided in this
instructive guidance from the running data of the chain of study. Bibliometrics is a literature management tool that
SW disposal to reduce the economic cost and obtain more can use mathematical methods to quantitatively analyze
environmental benefits. In contrast, citizens and workers documents in a certain field. It has been applied in many
in the SW industry require more convenient and efficient fields to explore technological progress, such as energy
data-driven techniques, particularly machine learning development (Obileke et al., 2020), business economics
(ML). (Ding and Yang, 2020), and environmental science
Currently, ML is the most popular data-driven techno- (de Sousa, 2021). However, it has rarely been used to
logy. It simulates or recognizes human learning behavior
analyze the application of ML in SW management.
using computers. Due to of its time-saving, strong
In this study, the basic characteristics, including the
learning ability and high accuracy, ML has been applied
total publication, total citations of articles, countries,
in many fields, such as population analysis (Wan et al.,
institutions, keywords, and distribution of research areas,
2009), soil conservation (Qu et al., 2018), and environ-
were investigated using the bibliometric method to obtain
mental science (Erkinay Ozdemir et al., 2021). In recent
an overview of the development trend over the past years.
years, ML has also been used for at various aspects of the
Based on a quantitative and comprehensive analysis,
SW life cycle. For example, Tao et al. (2020) studied the
elemental composition (C, H, O) and lower heating value state-of-the-art advances and future perspectives of ML
of SW and predicted these characteristics according to on different areas of SW disposal were also explored. We
infrared spectra with an accuracy of over 85 %. Coskuner hope that this work will significantly contribute to the
et al. (2021) proposed robust predictive models for annual environmental and resource concerns associated with SW
forest generation rates of domestic, commercial, disposal.
construction, and demolition wastes; R2 values of 0.95,
0.99, and 0.91, respectively were obtained. Chen et al.
(2021a) established a model based on an artificial neural 2 Data and methods
network and decision tree to determine online
measurement and control of polychlorinated dibenzo- 2.1 Bibliometrics
dioxin and dibenzofuran emissions. Idwan et al. (2020)
developed a two-step heuristic algorithm to discover the Bibliometrics is a science that adopts mathematics,
ideal truck route for waste fleet management. Zheng and statistics, and other measurement methods to study the
Gu (2021) proposed an ensemble image-learning model literature system and bibliometric characteristics (Bhatt
based on convolutional neural networks to classify et al., 2020), such as distribution structure, change regul-
domestic SW. ation, and quantitative relationship. Key information,
These attempts to use ML in multiple ways for SW such as the title, author, publication period, keywords,
disposal provide a promising scope for the development and citation information, obtained from literature are
of this field. Meanwhile, a timely overview of relevant analyzed using this method. Bibliometrics can be used to
achievements is urgently needed to help researchers and explore specific aspects of science and technology and
policymakers in this field find potential directions. have become a general method for measuring regulation
Several reviews have been conducted in recent years that (Keramatfar and Amirkhani, 2019). Bibliometrics pro-
focus on ML applications in the field of SW disposal. For vides a useful tool to reflect scenarios and trends in a
example, Wang et al. (2022) analyzed the role of ML in research field. This study concludes with the quantity and
the development of bioenergy and conversion of biofuel. visual process used to identify patterns and characteristics
Guo et al. (2021) summarized the application of ML to of changes in scientific literature.

Need to identify in our current market

 Rui Liang et al. Applications of machine learning in the life cycle of solid waste 3

2.2 VOSviewer Citation Index Expanded (SCI-EXPANDED) and Social

Sciences Citation Index (SSCI). This study was
VOSviewer is a software tool for creating maps based on conducted according to topic and title. The topic search in
network data and visualizing and exploring these maps. It WoS covered each article’s title, abstract, and keywords,
is commonly used for bibliometric analyses. VOSviewer encompassing the maximum amount of information in the
is based on sophisticated clustering and analysis database. To obtain precise results, a title search was also
technology that analyze knowledge units in scientific used. After repeated deliberation (Guo et al., 2021),
literature. Its biggest advantage is that it visualizes specific search formula and keywords in this study were
knowledge units and has good image presentation selected and shown in Fig. 1(a). The search formula
capabilities. A co-occurrence matrix and distance-based stipulates the scope of the literature search and ensures
grouping were used to build maps in VOSviewer (Lima that the obtained results conform to the scope of
et al., 2021). This software can perform co-authorship, application of ML on SW. These three terms of search
co-occurrence, co-citation, citation, and bibliographic formula were linked by Boolean “and” logic, considering
analyses by exploring the density between documents. articles on this subject published between January 1985
VOSviewer includes three browsing formats: network, and June 2021. The search was continued until June 6,
overlay, and density formats. Various browsing formats 2021.
reveal the flow and transfer of knowledge between After the preliminary search, the search results were
scientific documents based on color, size, width, and refined to include journal articles, which were the
other factors. This also reflects the similarity, research objects of this article. Next, these refined results
consistency, and citation relationships between scientific were exported in the format of “Full Record and Cited
documents. References” from WoS. The final exported information
was analyzed using data processing, VOSviewer, and
2.3 Literature analysis methods summaries.

The steps of this study were illustrated in Fig. 1(a). The

first step was to obtain literature on ML applied to SW, 3 Bibliometric analysis
for which the Web of Science (WoS) core collection was
used. WoS is a web-based product developed by 3.1 General introduction
Thomson Scientific in the United States. This is a large-
scale, comprehensive, and multidisciplinary journal A total of 170 documents from 1985 to 2021 related to
citation index database. This study selected two major ML applied to SW were retrieved from the SCI-
citation indices in the WoS core collection: Science EXPANDED and SSCI databases. Of these, journal

Fig. 1 Research route of the article (a) and progress of author keyword analysis in VOSviewer (b).
4 Front. Environ. Sci. Eng. 2023, 17(4): 44

articles accounted for 88.82 % (151). Other types of incineration in 2015 (Wang et al., 2016), the cooperation
documents included reviews, early access, and editorial between international countries has greatly deepened.
materials. Only journal articles were considered in this Against the background of globalization, it is conceivable
study. that international cooperation will be stronger in the
In view of journal articles published yearly (Fig. 2), the future in the context of national cooperation.
development of the application of ML to SW could be Fig. 3(a) showed a comparison of the countries with the
clearly identified. As shown in Fig. 2, journal articles on top eight TPs from 2001 to 2021. Of the eight most
the subject areas of this study were first published in productive countries, five were from Asia (China, Iran,
2001. The total number of publications (TP) showed a Turkey, India, and Malaysia), two were from North
rapid upward trend over time, although there were some America (USA, Canada), and one was from Europe
fluctuations throughout the studied period. From 2001 to (England). China was the most productive country, with
2010, research on ML applied to SW was still in its 37 journal articles (18.41 %), followed by Iran (26), and
infancy worldwide; the global TP did not exceed 3. After Turkey (16). China and Iran have remarkably more
2010, the TP increased significantly, reaching 29 in 2020, articles than other countries, resulting in a significant
an increase of nearly nine times compared to the TP in research gap between other countries.
2010. After 2010, ML and artificial intelligence began to In terms of the number of international collaborative
emerge. The growth of this field can be attributed to the publications (CP), the USA and China had the best
required advanced data processing technologies and the performance compared to other countries. However, the
development of environmentally friendly concepts, and CP of the USA had significantly exceeded the number of
many researchers began to apply ML on SW. For single-country publications (SP). Except the USA, out of
example, many studies have highlighted the importance these eight countries, only England had a higher CP than
of artificial intelligence. This field is now a research hot
SP. Furthermore, articles published in individual
spot and may have better prospects in the future.
countries were more common. This may be because
Change in the total number of times literature cited in
research on ML applied to SW was still in its infancy and
the WoS (TC) was similar to that in the TP. The TC
domestic exploration was still in its initial stages.
showed an overall increasing trend. However, there was a
National cooperation deepened after the initial develop-
sudden decline in 2007 and 2014. In 2007, the TC had a
minimum of 4. TC reached its maximum in 2018, with ment. China had the highest SP (20), accounting for
354 citations, and showed a downward trend thereafter. 18.52 %. Tran was close behind, ranking second. For the
This may be because citations require long-term H-index, which has an important impact on waste
accumulation. The high citation number also proved that management and disposal, China ranked second (256).
ML has great significance in the development of the SW Therefore, based on the comprehensive performances of
life cycle. TP, SP, CP, and FP, China has an important role in the
field of ML applied to SW. However, improvements in
3.2 An analysis of countries scientific technology and further strength in international
cooperation are required.
From 2001 to 2021, 46 countries have published articles Fig. 3(b) showed the TP changes in the countries with
on ML applied to SW. Of the 151 journal articles the top five TPs from 2001 to 2021. The first study in this
investigated in this paper, 43 (28.48 %) were involved in field appeared in USA in 2001. The first article from
international cooperation, and the remaining 108 only had China was published in 2002. However, this field did not
participation from individual countries. Compared to the show great development until 2014. In addition, the
international cooperation (18.01 %) of waste-to-energy development of ML applied to SW in these five countries
fluctuated between 2001 and 2021 and has a similar trend
to TP changes, as discussed under Section 3.1. Prior to
2014, research in this field had appeared almost
exclusively in one country. After 2014, the number of
articles published in this field fluctuated. This may be
because of the higher requirements for sophisticated
equipment worldwide, which has led to technological
progress.
The co-authoring relationships of these 45 countries
were also investigated using VOSviewer. The network
was shown in Fig. S1. This showed that China and the
USA played a key role in international cooperation.
Moreover, these two countries had the closest connec-
Fig. 2 TP and TC changes from 2001 to 2021. tions.
 Rui Liang et al. Applications of machine learning in the life cycle of solid waste 5

Fig. 3 Comparisons of the top 8 productive countries (a) and TP changes by year in the top 5 productive countries (b) during
2001–2021.

3.3 An analysis of institutions between countries may bring vitality to the development
of this new field.
Among the 151 articles on ML applied to SW, 212 were The top 8 productive institutions were shown in Fig. 4.
included. A total of 86 journal articles were completed University of Tehran in Iran had the largest TP (12, 4.08
with the cooperation of multiple organizations, accoun- %), which was much higher than that of other
ting for 56.95 %. This was much higher than the organizations. The TP in subsequent institutions showed
proportion of cooperation (28.48 %) between countries in little difference, all in single digits. These eight organiza-
discussed under Section 3.3, indicating that interchange tions were distributed across Iran, Malaysia, Canada,
within one country was much more common than that Turkey, and Singapore. In particular, Iran, Malaysia,
across multiple countries. Thus, international cooperation Canada, Turkey, and Singapore ranked 2, 7, 5, 3, and 13,
must be strengthened further. More communication respectively, according to their TP from 2001–2021. No
6 Front. Environ. Sci. Eng. 2023, 17(4): 44

Fig. 4 Top 8 productive institutions for ML applied to SW, during 2001–2021.

Chinese institutions appeared in the top 8 productive occurrence would be affected. A synonym dictionary,
institutions, although China had the largest TP. From the thesaurus_terms.txt file, was created in this step. It was
calculated statistical results, it can be seen that the TP of then imported to VOSviewer to perform author keyword
Chinese institutions was 3 or below, but the institutions analysis. In VOSviewer analysis, minimum number of
had TP that fell within a wide range. This may be the occurrences of a keyword was set as four.
reason why China had the highest yield in the field of ML A total of 443 author keywords were identified in 151
applied to SW. journal articles. Fig. 5 showed the results of author
The co-authoring relationship of these 212 institutions keyword co-occurrence in VOSviewer. The nodes
was also investigated using VOSviewer. The network was represent different author keywords. The larger the node,
shown in Fig. S2. Ton Duc Thang University (Vietnam) the more occurrences of author keywords in the 151
and the University of Tehran play key roles in interna- journal articles. The lines between the nodes reflect the
tional cooperation. relationships between the author keywords. If the line
between two nodes was thicker or shorter, the relationship
3.4 An analysis of keywords between the two author keywords was strengthened. The
different node colors represent the different types of
The keywords were the gist of an article and could reflect clusters. Each cluster was determined based on its
the main topic of the relevant research area. This can help weight/significance (Lima et al., 2021). This cluster
researchers better understand the emerging trends in a algorithm of VOSviewer is similar to a probability-based
field. In WoS, keywords contain two parts: author measurement method. The more similar the two author
keywords and keyword plus. The author keywords were keywords, the more related they are.
those we commonly referred. Keyword plus was a The central author keyword of this network was
keyword added by WoS to increase the hit rate of the “municipal solid waste”. The nodes around the center can
article under a relevant topic. Therefore, author keyword be divided into two types: 1) keywords related to
analysis was adopted using VOSviewer to acquire the algorithms, such as artificial neural networks, genetic
hotspots as well as future trend of ML applied to SW. algorithms, and support vector machine and 2) keywords
Fig. 1(b) present author keyword analysis steps. First, related to applications, such as waste management and
Openrefine was used to sort out the duplicate author waste-to-energy. In addition, “municipal solid waste” and
keywords in the search record. If synonymous author “artificial neural network” appeared most frequently. The
keywords with different spellings were directly entered in link strengths of these two keywords were also the
VOSviewer for analysis, the accuracy of author keyword strongest.
 Rui Liang et al. Applications of machine learning in the life cycle of solid waste 7

Fig. 5 Author keywords network of 151 journal articles in VOSviewer.

There were four clusters of different colors, as shown in cooperative relations and their respective research levels
Fig. 5, which also showed the main topics of these four on ML applied to SW. China was the most productive
clusters. The top three keywords with the highest country and played a key role in international
occurrences in these four clusters were summarized in cooperation. More than a quarter of the articles involved
Table S2. Twelve keywords with high occurrences were international cooperation. The TP between institutions
obtained and can be divided into three categories. 1) showed a few differences, all in single digits, except for
Municipal solid waste (with the highest occurrence out of University of Tehran, which had a TP of 12. The
all keywords) and solid waste, which were related to SW cooperation of institutions within one country was much
itself; 2) artificial neural network, support vector more common than in multiple countries. The top 5
machine, genetic algorithm, machine learning, and neural productive journals in the publication of ML applied in
network, which were related to the ML algorithm; and 3) SW during 2001–2021 were provide in Table S1.
optimization, forecasting, recycling, waste management, Keywords reflected the main focus of this research study
and life cycle assessment, which focus on different and can be divided into SW categories, ML algorithms,
aspects of the life cycle of SW. Accordingly, research on and specific applications. Based on this bibliometric
ML applied to the life cycle of SW can be divided into analysis, the development venation of ML applied in SW
three parts: SW categories, ML algorithms, and specific in terms of time, international state, and research topics
applications. These three aspects were discussed in were analyzed. This provided a clear and overall insight
further sections. to researchers and policymakers in this area. It was
We analyzed the literature on ML applied to SW using necessary to summarize the different aspects of ML
the bibliometric methods in provided under Sections application to the lifecycle of SW. Moreover, it can
3.1–3.4. The characteristics of the related literature from significantly contribute to the future prospects of ML and
1985 to 2021, based on the SCI-EXPANDED and SSCI SW.
databases, were examined. This provided a development
venation for the application of ML in SW management.
The first literature on this research area was published in 4 Literature review
2001. The TP and TC of the articles showed rapid upward
trends, with the exception of a few fluctuations. According to the bibliometric analysis of the keywords,
Publications on ML applied in the lifecycle of SW in as mentioned under Section 3.4, ML applied to the life
countries and institutions worldwide were also examined. cycle of SW can be divided into three parts: SW
This provided information on the international categories, ML algorithms, and specific applications.
8 Front. Environ. Sci. Eng. 2023, 17(4): 44

Based on the literature and bibliometric analysis, a generation using conventional statistical methods. Based
detailed introduction, and an in-depth analysis of these on the robust properties of data processing, ML can be
three parts were carried out. used to learn previous data on waste production, monitor
SW was the raw material investigated in this study. SW its generation and predict the evolution regulation. SVM,
refers to the production, consumption, living, and other ANN, and RF were typically used in research.
activities of solid and semisolid waste materials produced Furthermore, prediction models were improved in
by human beings. It consists of a wide range of subsequent studies. They can reflect more information in
substances and can be divided into industrial SW, addition to SW generation in SW management. Dai et al.
municipal SW (MSW), construction SW, agricultural (2011) developed a support vector regression model to
SW, and hazardous SW. MSW was highly prevalent in predict the amount of waste generated in a city. It also
the 151 journal articles (Fig. 5). It is the SW generated in reflects the dynamic, interactive, and uncertain
the daily life of a city or the activities that provide characteristics of SW management systems. These SW
services to the city. With the increase in human progress, generation prediction models were helpful for SW
urbanization has advanced. Thus, human activities have management in MSW plants.
become more abundant. Consequently, the production of As mentioned above, waste generation is a complex
MSW has become aggravated (Noori et al., 2009), and process closely correlated with human social life. It is
the number of scientific studies on MSW has increased. influenced by some uncertain factors and the population
ML was a method used to study the life cycle of SW. and economic aggregates of a city, including population
ML is a multidisciplinary subject that specializes in size, city size, GDP, income level, capita electricity
simulating human learning behaviors in computers. It demand, employment status, urbanization level, education
contains various types of algorithms. In the literature, we level, consumption level, etc. The relationship between
found that artificial neural network (ANN), support SW generation and these factors may be complex and
vector machine (SVM), random forest (RF), and genetic nonlinear. In some cases, ML has a better processing
algorithm (GA) were the most used ML algorithms ability than do traditional methods. Some studies have
applied to the lifecycle of SW. In particular, GA is an investigated the effects of these factors on waste
optimization algorithm that can optimize the parameters generation. Jiang and Liu (2016) analyzed the influences
of other algorithms such as ANN, SVM, and RF. of social and economic changes on SW generation based
ML has been applied to the entire SW process, on a statistical learning approach. Magazzino et al. (2020)
affecting its life cycle. According to the 151 journal investigated the relationship between SW production,
articles obtained in this study, the specific applications of greenhouse gas emissions, and GDP in Switzerland.
ML on SW can be divided into three areas. ML can be Nguyen et al. (2020) explored the relationships between
used to 1) predict the generation characteristics of SW, 2) SW, individual, and socioeconomic factors. The major
collect and transport SW, and 3) treat and utilize SW. The socioeconomic factors that influenced SW composition
status of ML was discussed in detail in the subsequent were safety concerns, economic activities, and lifestyles.
three sections. ML models have also been used to predict the physical
and chemical properties of SW. Physical properties
4.1 Generation and characterization include the wear rate (Nayak and Satapathy, 2020), fiber
content, moisture content, dry unit weight, axial strain
During the life cycle of SW, it is important to first (Falamaki and Shahin, 2019), and compression ratio
consider its properties. The generation of SW and its (Heshmati et al., 2014). Chemical properties include
changes reflect the development of a country. The chemical composition (C, H, O, N) and heating value. In
generation of SW is related to economic, social, popular prediction models, the input variables are
population, and other factors. In addition, SW spectrum data (Tao et al., 2020; Yan et al., 2021) or
characteristics are greatly important for its utilization, physical composition (Lin et al., 2015). However, in
such as elemental composition, heating value, and ash some studies, the input parameters of the characteristic
content. This influenced the treatment method or prediction model could be plant operating (You et al.,
condition of the SW disposal. ML can be used to predict 2017) or seasonal variation (Adeleke et al., 2021).
SW generation and characteristics through data learning. Perhaps the accuracy of predictions is lower than that of
Quantitative prediction of SW production is critical for actual measurements, but it is acceptable in many actual
the design and planning of SW management systems. As downstream treatments (Tao et al., 2020). In addition,
mentioned above, SW is produced by human beings these predictions are obtained faster and are more
through their social activities and is difficult to convenient than those by complex measurements.
understand. Thus, the quantity of SW is influenced by In general, ML contributed to the quantitative
various uncertainty parameters related to human beings. prediction of SW production, its relationship with
Additionally, the amount of SW is highly volatile. economic and social factors, and its characterization. The
Therefore, it is difficult to precisely predict SW current aim of urban management is to promote SW
 Rui Liang et al. Applications of machine learning in the life cycle of solid waste 9

reduction and energy utilization of solid waste. Therefore, allocation were also considered along with the
it is crucial to understand the root cause of SW to control optimization of the transportation path. The locations of
its generation. The SW is a product of human social the collection points were also the input parameters of the
activities. Establishment of the quantitative relationship transport-optimization model. This topic is discussed
between SW generation and socioeconomic factors was later.
helpful to fundamentally clarify the root cause of SW Thus, it is important to consider waste transportation.
production and, consequently, reduce SW generation SW at the waste drop points should be transported by
from economic and social activities. Moreover, ML helps trucks and other vehicles to the central treatment station.
to construct the forecast model of weekly, monthly, or Because of differences in collection point locations,
yearly SW using a variety of linear and non-linear disposal, city construction, SW components, collection
methods. This provided an intuitive understanding of SW frequency, collection type (Vu et al., 2020), etc., the
structure and quantity changes. Additionally, the SW transport route was different for different scenarios.
characterization model constructed using multiple inputs Furthermore, the rational truck route significantly
overcomes the drawbacks of traditional testing methods decreased the transport time, distance, and fuel
and is more suitable for automation technology. ML plays consumption. Therefore, to save energy and costs, it is
an important role in the generation and characterization of necessary to choose an appropriate transport path and
SW. vehicle (including vehicle amount and vehicle capacity).
Using a literature review, heuristic procedures (Beliën
4.2 Collection and transportation et al., 2014) and genetic algorithms (Viotti et al., 2003)
were used to optimize SW transport routes. Geographic
Large amounts of SW are generated annually. Thus, they information systems are also necessary for SW route
require appropriate management. Inappropriate regulation optimization. They are usually studied using a case study.
and disposal of SW have a negative influence on the For example, Vu et al. (2020) developed and solved 48
environment and ecosystems. The open-air storage of SW vehicle routing problem models to optimize travel
causes soil pollution. Toxic and harmful components can distance and time in Austin, Texas, United States. The
harm the soil and enter the food chain, even reaching the input of the models contained geographical locations of
human body (Zhang et al., 2010; Liu et al., 2021). In collection points, location of landfills, road network,
addition, SW seriously destroys aquatic environments. number of collection points, volume of waste at each
SW discharge into rivers causes damage to water quality, collection point, and pick-up time, and others. Thereafter,
river siltation, and blockage (Chandra et al., 2006). In the model can output the optimized truck routes,
addition, dust and SW particles present in the atmosphere collection frequency, truck compartment configuration,
can cause atmospheric pollution (Rabl et al., 1998). and type and capacity of trucks. Vu et al. (2019) explored
Therefore, it is necessary to study different categories of the effect of SW compositional features on optimized
SW, investigate their properties, and identify efficient truck route time, distance, and air emissions. This helped
treatment methods. save between 10.3 % and 16.0 % in travel distance and
As mentioned above, the generation of SW is correlated slightly reduced emissions. These case studies will be of
with the economy, population dynamics, and policies. great significance to waste system designers and
These factors also influence SW management. In policymakers.
addition, SW management is of great significance for A key factor affecting the collection and transportation
waste disposal and environmental improvement. It of SW is the location of SW treatment plants. ML has
contains several topics. The collection and transportation also been used to assist in selecting the location of SW
of SW were investigated in this section. Selection of SW treatment plants and exploring the optimal proportions of
collection sites, proper transportation routes, treatment different types of plants. It is important to consider cost
plant sites, and proper proportions of different plants are and environmental effects when selecting the location of
complex optimization problems that need to consider all plants. The positions of landfills, waste-to-energy plants,
factors, including private and external costs (Korucu and and MSW disposal sites have been studied previously.
Karademir, 2014). In recent years, ML has been Simsek et al. (2006) introduced a new waste site selection
commonly used to solve complex optimization problems. tool based on a geographic information system. Based on
First, SW drop-off or collection points must be five different hydrogeological parameters of groundwater,
selected. If the collection location is suitable, it can the tool developed different disposal area schemes and
contribute to reducing the cost and potential was used in the Torbali Basin to select the optimal
environmental risks. ML has been used to study the project. However, if there are several SW disposal plants
allocation of the SW collection bins. In 1998, Huang et al. in a city, a considerable arrangement is required.
(1998) adjusted trash-flow allocation using a gray linear Otherwise, resources such as land, energy, or costs may
programming model. This was helpful in formulating be wasted. Farzaneh et al. (2021) developed three
local policies. However, studies on collection bin scenarios to consider incineration, composting, recycling,
10 Front. Environ. Sci. Eng. 2023, 17(4): 44

and landfilling, thereby minimizing environmental considered in the treatment processes: reaction monito-
burdens and costs. In my opinion, cloud-based resource- ring, product yield, pollutant control, and assessment of
allocation methods can be applied to optimize different the whole process. ML plays an important role in this
ratios of plants in the future. process. Input parameters of equipment, such as the
In general, ML contributed to optimizing the selection characteristics of SW, structure of equipment, and
of SW collection sites, transportation routes, treatment reaction conditions, can influence the reaction process
plant sites, and the proportions of different plants. and even the yield of the products and pollutants. By
Economic development, social form, topography of an learning previously processed data, ML can be used to
area, and rural-urban differences have important impacts construct complex functions to simulate their correlation.
on SW distribution and SW collection and transportation. Incineration is a common method of SW treatment, and a
Optimization of SW collection and transportation is not detailed analysis of ML under different energy utilization
only an environmental necessity but also an economic methods is discussed below.
necessity. ML is helpful for finding a fast and optimal ML is typically used to simulate SW incineration and
way to solve this problem and reduce energy and predict pollutant emissions of SW incineration. In 2000,
economic consumption. ML plays an important role in Chang and Chen (2000b) designed a fuzzy controller for
SW collection and transportation. municipal incinerators using genetic algorithms. Fuzzy
control technology is helpful for reducing operational
4.3 Treatment and utilization risks. It was also studied subsequently to provide more
reliable control of the combustion process (Chen et al.,
After the collection and transportation of SW, the 2002). ML was also used to test SW combustion (Dong
disposal of SW is the next important issue in SW et al., 2002; Kabugo et al., 2020) and optimize the waste
management. Energy utilization of SW can not only incineration plant using multi-objectives (Anderson et al.,
reduce environmental hazards but also produce energy. 2005). They were helpful in providing clear observations
However, the heterogeneity of SW has a negative and decision support to the human operator. Pollutant
influence on treatment. Thus, it is very important to emission control is of great significance for environme-
classify SW before energy utilization. Incineration, ntal protection. In 2000, Chang and Chen (2000a) found
landfills, anaerobic digestion, pyrolysis, and gasification that artificial intelligence was useful for identifying
are commonly used to treat SW. With the rapid nonlinear structures in relation to polychlorinated
development and good performance of ML, it has been dibenzodioxin and dibenzofuran emissions from SW
used to study SW classification and treatment. incinerators. The emission characteristics of other
Waste classification is of great significance for pollutants (e.g., SO2 (Wen et al., 2006) and HCl (Chi
downstream applications. On the one hand, it ensures that et al., 2005; Zhang et al., 2019)) were studied using a
recyclables can be recycled and reduce resource waste. neural network. The pollutants were caused by
However, as mentioned before, there are different types incomplete combustion of SW. To reduce environmental
of SW and they have heterogeneous properties. problems, it is necessary to find a more optimal
Therefore, different SW require different disposal management strategy. Diagnostic analysis of the Garson
conditions to ensure an optimal energy conversion index is useful solution (Chen and Chen, 2008).
efficiency. Moreover, it is necessary to classify SW Giantomassi et al. (2011) predicted the steam production
before utilization. Sophisticated SW classification of an SW incinerator using fully tuned minimal RBF
methods based on images, spectra, and sounds have been neural networks.
proposed in recent years. ML was used to establish the ML was also used to simulate the SW landfills, predict
classification models and performed well. Hannan et al. the physical properties of SW in landfills, and evaluate
(2014) achieved SW classification according to their the environmental impacts (subsurface temperatures,
images based on a feedforward neural network with the leachate, and methane emissions) and energy
highest accuracy being 0.9875. Junjuri and Gundawar consumption of the landfills. In 2006, Ozcan et al. (2006)
(2020) identified ten post-consumer plastics combined first used artificial neural networks to simulate SW
with laser-induced breakdown spectroscopy and an landfills and predict CH4 production. Moreover, the
artificial neural network, with an accuracy of over 97 %. correlations were 0.983 and 0.806 for programming and
Korucu et al. (2016) developed sound classification testing, respectively. In addition, this year, a neural
systems based on a support vector machine and a hidden network was used to model the leachate flow rate at a SW
Markov model to separate packaging waste with an site (Karaca and Özkaya, 2006). Subsequently, ML was
accuracy of over 88 %. Lai et al. (2017) explored food applied to predict the physical properties of SW in
waste recovery using an electrostatic separation process landfills. The long-term settlement of SW landfills
based on an artificial neural network. occurred in almost every landfill. This was attributed to
Finally, almost all unrecoverable SW is sent for the long-term mechanical or decomposition-based
disposal or energy utilization. Some points should be compression. A genetic algorithm was used to estimate
 Rui Liang et al. Applications of machine learning in the life cycle of solid waste 11

long-term and short-term settlements (Park and Park, ML was also applied during pyrolysis, where it was
2009). The compression ratio of SW is an essential used to model the pyrolysis process and predict the
parameter for evaluating settlement and landfill design. product yield. In 2016, a least-squares support vector
The decision tree method (Heshmati et al., 2014), machine was used to predict the biochar yield from cattle
adaptive fuzzy neural network, and ANN models manure pyrolysis (Cao et al., 2016). ML has also been
(Mokhtari et al., 2015) were used to predict the compre- used to predict the characteristics of products. Li et al.
ssion ratio. In addition, landfill temperature is an (2020a) used an artificial neural network to predict a
important factor for safety. Satellite images and artificial higher heating value for syngas pyrolyzed from sewage
neural networks were used to simulate and predict landfill sludge, and ML was used to simulate the progress. A
surface temperature (Abu Qdais and Shatnawi, 2019). A genetic algorithm and neural fuzzy model were applied to
risk assessment of elevated subsurface temperatures in determine the optimal operating conditions over different
landfills is necessary. Sabrin et al. (2021) proposed a temperature ranges (Pan et al., 2021). ML has also been
comprehensive risk assessment framework specific to used to investigate kinetic parameters (Pan et al., 2022)
subsurface elevated-temperature mitigation. In addition, and waste pyrolysis thermodynamics (Chen et al., 2021b)
leachate and fugitive landfill methane emissions must be and evaluate the potential of pyrolysis (Ye et al., 2018).
considered in actual applications. They are harmful to Moreover, the distribution of special elements during
groundwater and air. ML has been used to predict the pyrolysis can be predicted using ML (Sun et al., 2020).
generation (Abunama et al., 2018), COD load (Azadi ML was applied to model the gasification process and
et al., 2016) and treatment process (Azadi et al., 2021; predict the characteristics and yields of products. In 2016,
Vaz et al., 2021). ML has also been used to simulate and an artificial neural network was used to predict the lower
predict methane emissions (Kormi et al., 2018; Mehrdad heating values of gas, tar, and entrained char (Pandey
et al., 2021) and their effects (Singh et al., 2021). Finally, et al., 2016). Subsequently, several algorithms have been
landfill area estimation is helpful for the better planning used to model gasification progress and predict reaction
and management of landfill sites (Hoque and Rahman, properties (Kardani et al., 2021). ML has also been used
2020). to explore the influence of different variables on its
The energy consumption and environmental impacts of utilization and to identify more suitable conditions (Yan
incineration and landfills were studied using ML. Nabavi- et al., 2018; Kardani et al., 2021).
Pelesaraei et al. developed an ANN for forecasting and Furthermore, ML has been applied in the hydrothermal
modeling the energy and life cycles of incineration and reaction (hydrothermal carbonization, hydrothermal
landfilling of SW. This is helpful for SW management liquefaction, and hydrothermal gasification) of SW to
(Nabavi-Pelesaraei et al., 2017). model the hydrothermal reaction process, predict the
ML was used for the anaerobic digestion of SW. It can yield and characteristics of the products, and optimize the
predict methane production, simulate the process reaction conditions. In 2020–2022, ML will play a more
performance, and evaluate the mass and energy balance. important role in the hydrothermal reaction of SW.
Methane yield prediction and evaluation are important Different ML algorithms, such as the gradient boost
topics in anaerobic research. In 2002, Holubar et al. regressor (Li et al., 2021b) and neural network (Li et al.,
(2002) first used feed-forward backpropagation neural 2021a), were applied to model the process of
networks to simulate methane production in anaerobic hydrothermal gasification of SW. ML has been used to
digesters. Subsequently, biogas production on different predict the characteristics of products, such as yield (Li
substrates using different operations was predicted using et al., 2021d), heating value of hydrochar, energy
different ML algorithms (Turkdogan-Aydınol and recovery efficiency, energy densification (Li et al.,
Yetilmezsoy, 2010; Li et al., 2022b). In addition, SW 2020b), yield, and O and N content in bio-oil (Zhang et al.,
with high biodegradability and high organic and nutrient 2021). In addition, ML has been applied to optimize the
contents is suitable for producing hydrogen through yield of syngas from hydrothermal gasification (Li et al.,
anaerobic digestion. Elsamadony et al. (2015) predicted 2021b), optimize the production of bio-oil using high-
biohydrogen production using an artificial neural energy recovery and low nitrogen content from the
network. In addition, the fates of C and N were predicted hydrothermal liquefaction of biomass (Li et al., 2021c),
based on an artificial neural network (Li et al., 2016). ML and aid in the screening of catalysts (Li et al., 2021a).
can also be used to simulate the process (Saghouri et al., In general, ML has contributed to the establishment of
2020) and predict control parameters (Flores-Asis et al., SW classification models, modeled different processes of
2018). This is helpful for monitoring and optimizing the SW energy utilization (incineration, landfill, anaerobic
reaction process. Moreover, the mass and energy balances digestion, pyrolysis, gasification, and hydrothermal
and economics of anaerobic digestion must be considered reaction), predicted the characteristics of products,
in actual applications. Dahunsi et al. (2017) optimized the optimized the reaction process, and evaluated the
anaerobic co-digestion process based on the response environmental and energy performance of the technology.
surface methodology and an artificial neural network. SW classification based on ML is beneficial for achieving
12 Front. Environ. Sci. Eng. 2023, 17(4): 44

automated and intelligent classification, which is suitable weighted average of its appearance year and times. If the
for development direction of modernization and color of the node is closer to red, this indicates that the
intellectualization. In addition, for SW industrial energy author keyword appeared more frequently around 2020. If
utilization, it is important to recognize efficient automatic it was closer to blue, it indicated that the keyword
monitoring, as well as to save manpower and material appeared more frequently before 2014.
resources. Moreover, ML is helpful for exploring a more The average publication year of “Waste-to-energy” was
efficient energy conversion and breaking the current 2020, followed by machine learning (ML) (2019.64),
upper limit. ML plays an important role in SW treatment recycling (2019.20), prediction (2019), and pyrolysis
and utilization. (2018.75), thereby reflecting the recent research hotspots.
The ongoing severe environmental problems and
energy crises may account for the occurrence of these
5 Future perspectives author keywords. On the one hand, mass accumulation
and improper disposal of SW have damaged ecosystems
Based on the quantitative bibliometric analysis and a and the living environment. On the other hand, severe
comprehensive summary of previous studies, a holistic energy shortage problems have attracted increasing
landscape of ML applied to SW was provided. Perspec- attention worldwide. Waste-to-energy is not only of great
tives on future development opportunities were discussed significance for environmental governance but also for
in this section. solving the current energy crisis. The energy utilization of
SW is an effective method for solving these problems.
5.1 Time evolution of author keywords from 1985 to 2021 Pyrolysis, gasification, and hydrothermal methods have
several potential uses. Through these technologies, SW
The time evolution of author keywords could be used to can achieve high-value transformation and produce oil,
reflect the development of hot topics in a certain research syngas, charcoal, etc. In addition, they produce less
field, even probable future perspectives. The visualization pollution than does waste incineration. Recycling has also
of author keyword evolution (occurrences ≥ 4) has been been emphasized because of its energy-saving and
reduced in VOSviewer, as shown in Fig. 6. The means of environmental friendliness. It aims to recycle valuable
the code size and lines between the nodes were the same objects from a large amount of waste and achieve reuse.
as those mentioned under Section 3.4. The colors of the Waste classification could be helpful in realizing source
codes have slightly changed. They reflect the average recycling, and has been carried out in many countries,
publication year of the author keyword, that is, the especially China, in recent years. This is beneficial for

Fig. 6 Overly visualization of author keyword co-occurrence (> 3).

 Rui Liang et al. Applications of machine learning in the life cycle of solid waste 13

carbon neutrality. projects poses certain risks and limits it application.

In addition, author keywords with lower occurrence Under global intelligent development, an efficient data-
(< 4) may provide new ideas for the future development processing method, ML, will play an important role in the
of ML applied to SW, such as waste collection (2021), lifecycle of SW. The distribution of SW structure and
economic analysis (2021), environmental sustainability characteristics caused by waste classification will
(2020.50), and waste heat recovery (2019.67). The continue to change in China. It requires timely monitoring
economic and environmental assessment of waste and subsequent responses, to which ML can greatly
accumulation and disposal is of increasing concern, contribute. In addition, because of the change in SW
which is consistent with the global call for carbon structure and characteristics, the progress of waste to
reduction and higher benefits. energy would be different, and it needs to be further
explored using ML to find more accurate scenarios.
5.2 Current challenges Furthermore, the SW reactor may require more precise
adjustment. ML can aid in the optimization of reactor
Data collection is the first step in an ML application. system design and monitoring in internal reactors (Li
However, there were some problems during this step. et al., 2022a). In addition, multiple types of ML input
First, some data, including temperature measurements in (e.g., text, number, graph, and image) may be helpful in
multiple areas of the incinerator, were difficult to acquire creating new knowledge, such as new catalysts (Zhong
et al., 2021).
and could not be measured by equipment. Second, data
collection may consume lots of time. These data may
come from numerous repeated experiments or related
published articles. It takes a considerable amount of time
6 Conclusions
to manually repeat experiments and find data from
This study provides an overview of ML applied in the life
published articles. However, there was another problem cycle of SW. Characteristics, including the total number
with the data from the published articles. In articles by of publications and citations, countries, institutions,
different researchers, the experimental operation would keywords, and distribution of research areas, of the
differ, which could affect the uniformity of data. literature on ML applied to SW from 1985 to 2021 based
Nonetheless, this is difficult to avoid. on the SCI-EXPANDED and SSCI databases were
Most algorithms, including four algorithms (ANN, examined. They could be helpful for researchers and
SVM, RF, and GA) commonly used in SW, suffer from policymakers to macroscopically study changes in this
the “black box” effect. The internal procedures of these research field temporally and spatially.
algorithms are unclear. We do not know the detailed The three popular topics on ML applied to the life cycle
processing and calculation routes of the results after SW are SW categories, ML algorithms, and specific
inputting the variables. Further studies exploring these applications. ML is mainly applied throughout the life
routes are, therefore, required. cycle of SW, including generation, characteristics,
The results of the ML models should be critically collection, transportation, and utilization. ML can be used
analyzed. A degree of bias may occur in model to predict the generation and characteristics of SW and to
processing but has not been discovered. Such biases explore the relationship between SW generation, society,
include data bias and algorithm bias (Zhong et al., 2021). and economic development. ML has also been applied to
They should be judged based on a deep understanding of optimize the allocation of SW collection sites, SW
ML, professional knowledge of SW, or a team of transportation, and SW treatment plant arrangement.
environmental researchers. Moreover, ML can be used to classify SW and simulate
the energy utilization process.
5.3 Opportunities of future development Finally, perspectives on the future development of ML
applied to SW were discussed. The goal has been to
Increasing the availability of data is important. Advanced achieve high economic and environmental benefits and
text-processing methods should be explored to obtain data carbon reduction during the lifecycle of SW. Increasing
from networks or published articles. The researcher could the availability of data, internal learning of ML, and
voluntarily share the data on an open platform, such as enhancing the appropriate application of ML in SW may
GitHub. An open and comprehensive database of the life create opportunities. We hope that this work will
cycle of SW is required to promote the application of ML contribute to addressing concerns associated with SW
in this field. More free data, such as those obtained from disposal in the environment and resources.
Phyllis and BIOBIB, are encouraged.
The internal learning of integrated ML packages is
necessary. The transparency of ML is extremely Notations
important when it is used, as this can increase its
SW Solid waste
credibility. Using something unknown in commercial
14 Front. Environ. Sci. Eng. 2023, 17(4): 44

ML Machine learning intelligent approach. Bioresource Technology, 202: 158–164

WoS Web of Science Chandra S, Chauhan L K S, Murthy R C, Gupta S K (2006). In vivo
genotoxic effects of industrial waste leachates in mice following
TP The total number of publications
oral exposure. Environmental and Molecular Mutagenesis, 47(5):
TC The total number of times literature cited in the WoS 325–333
SP The number of single countries publications Chang N B, Chen W C (2000a). Prediction of PCDDs/PCDFs
CP The number of international collaborative publications emissions from municipal incinerators by genetic programming and
FP The number of first country publications neural network modeling. Waste Management & Research, 18(4):
341–351
ANN Artificial neural network
Chang N B, Chen W C (2000b). Fuzzy controller design for municipal
SVM Support vector machine incinerators with the aid of genetic algorithms and genetic
RF Random forest programming techniques. Waste Management & Research, 18(5):
GA Genetic algorithm 429–443
Chen J C, Chen W H (2008). Diagnostic analysis of a small-scale
incinerator by the Garson index. Information Sciences, 178(23):
Acknowledgement This research was supported by the National Natural
Science Foundation of China (No. 52100157). 4560–4570
Chen K, Peng Y, Lu S, Lin B, Li X (2021a). Bagging based ensemble
Electronic Supplementary Material Supplementary material is available learning approaches for modeling the emission of PCDD/Fs from
in the online version of this article at https://doi.org/10.1007/s11783-023- municipal solid waste incinerators. Chemosphere, 274: 129802
1644-x and is accessible for authorized users. Chen R, Zhang D, Xu X, Yuan Y (2021b). Pyrolysis characteristics,
kinetics, thermodynamics and volatile products of waste medical
surgical mask rope by thermogravimetry and online
References thermogravimetry-Fourier transform infrared-mass spectrometry
analysis. Fuel, 295: 120632
Abu Qdais H, Shatnawi N (2019). Assessing and predicting landfill
Chen W C, Chang N B, Chen J C (2002). GA-based fuzzy neural
surface temperature using remote sensing and an artificial neural
controller design for municipal incinerators. Fuzzy Sets and
network. International Journal of Remote Sensing, 40(24):
Systems, 129(3): 343–369
9556–9571
Chi Y, Wen J M, Zhang D P, Yan J H, Ni M J, Cen K F (2005). HCl
Abunama T, Othman F, Younes M K (2018). Predicting sanitary
emission characteristics and BP neural networks prediction in
landfill leachate generation in humid regions using ANFIS
MSW/coal co-fired fluidized beds. Journal of Environmental
modeling. Environmental Monitoring and Assessment, 190(10):
Sciences (China), 17(4): 699–704
597–611
Coskuner G, Jassim M S, Zontul M, Karateke S (2021). Application of
Adeleke O, Akinlabi S A, Jen T C, Dunmade I (2021). Application of artificial intelligence neural network modeling to predict the
artificial neural networks for predicting the physical composition of generation of domestic, commercial and construction wastes. Waste
municipal solid waste: an assessment of the impact of seasonal Management & Research, 39(3): 499–507
variation. Waste Management & Research, 39(8): 1058–1068 Dahunsi S O, Oranusi S, Efeovbokhan V E (2017). Pretreatment
Anderson S R, Kadirkamanathan V, Chipperfield A, Sharifi V, optimization, process control, mass and energy balances and
Swithenbank J (2005). Multi-objective optimization of operational economics of anaerobic co-digestion of Arachis hypogaea (Peanut)
variables in a waste incineration plant. Computers & Chemical hull and poultry manure. Bioresource Technology, 241: 454–464
Engineering, 29(5): 1121–1130 Dai C, Li Y P, Huang G H (2011). A two-stage support-vector-
Azadi S, Amiri H, Rakhshandehroo G R (2016). Evaluating the ability regression optimization model for municipal solid waste
of artificial neural network and PCA-M5P models in predicting management: a case study of Beijing, China. Journal of
leachate COD load in landfills. Waste Management (New York, Environmental Management, 92(12): 3023–3037
N.Y.), 55: 220–230 de Sousa F D B (2021). Management of plastic waste: a bibliometric
Azadi S, Karimi-Jashni A, Javadpour S, Mahmoudian-Boroujerd L mapping and analysis. Waste Management & Research, 39(5):
(2021). Photocatalytic landfill leachate treatment using P-type TiO2 664–678
nanoparticles under visible light irradiation. Environment, Ding X, Yang Z (2020). Knowledge mapping of platform research: a
Development and Sustainability, 23(4): 6047–6065 visual analysis using VOSviewer and CiteSpace. Electronic
Beliën J, De Boeck L, Van Ackere J (2014). Municipal solid waste Commerce Research, 4: 1–23
collection and management problems: a literature review. Dong C, Jin B, Zhong Z, Lan J (2002). Tests on co-firing of municipal
Transportation Science, 48(1): 78–102 solid waste and coal in a circulating fluidized bed. Energy
Bhatt Y, Ghuman K, Dhir A (2020). Sustainable manufacturing. Conversion and Management, 43(16): 2189–2199
Bibliometrics and content analysis. Journal of Cleaner Production, Elsamadony M, Tawfik A, Suzuki M (2015). Surfactant-enhanced
260: 120988 biohydrogen production from organic fraction of municipal solid
Cao H, Xin Y, Yuan Q (2016). Prediction of biochar yield from cattle waste (OFMSW) via dry anaerobic digestion. Applied Energy, 149:
manure pyrolysis via least squares support vector machine 272–282
 Rui Liang et al. Applications of machine learning in the life cycle of solid waste 15

Erkinay Ozdemir M, Ali Z, Subeshan B, Asmatulu E (2021). Applying Kardani N, Zhou A, Nazem M, Lin X (2021). Modelling of municipal
machine learning approach in recycling. Journal of Material Cycles solid waste gasification using an optimised ensemble soft
and Waste Management, 23(3): 855–871 computing model. Fuel, 289: 119903
Falamaki A, Shahin S (2019). Determination of shear strength Keramatfar A, Amirkhani H (2019). Bibliometrics of sentiment
parameters of municipal solid waste from its physical properties. analysis literature. Journal of Information Science, 45(1): 3–15
Civil Engineering (Shiraz), 43(S1): 193–201 Kormi T, Mhadhebi S, Bel Hadj Ali N, Abichou T, Green R (2018).
Farzaneh G, Khorasani N, Ghodousi J, Panahi M (2021). Application Estimation of fugitive landfill methane emissions using surface
of MCAT to provide multi-objective optimization model for emission monitoring and Genetic Algorithms optimization. Waste
municipal waste management system. Journal of Environmental Management (New York, N.Y.), 72: 313–328
Health Science & Engineering, 19(2): 1781–1794 Korucu M K, Kaplan Ö, Büyük O, Güllü M K (2016). An investigation
Flores-Asis R, Méndez-Contreras J M, Juárez-Martínez U, Alvarado- of the usability of sound recognition for source separation of
Lassman A, Villanueva-Vásquez D, Aguilar-Lasserre A A (2018). packaging wastes in reverse vending machines. Waste Management
Use of artificial neuronal networks for prediction of the control (New York, N.Y.), 56: 46–52
parameters in the process of anaerobic digestion with thermal Korucu M K, Karademir A (2014). Siting a municipal solid waste
pretreatment. Journal of Environmental Science and Health. Part A, disposal facility, part II: the effects of external criteria on the final
Toxic/Hazardous Substances & Environmental Engineering, decision. Journal of the Air & Waste Management Association,
53(10): 883–890 64(2): 131–140
Giantomassi A, Ippoliti G, Longhi S, Bertini I, Pizzuti S (2011). On- Lai K C, Lim S K, Teh P C, Yeap K H (2017). An artificial neural
line steam production prediction for a municipal solid waste network approach to predicting electrostatic separation performance
incinerator by fully tuned minimal RBF neural networks. Journal of for food waste recovery. Polish Journal of Environmental Studies,
Process Control, 21(1): 164–172 26(4): 1921–1926
Guo H N, Wu S B, Tian Y J, Zhang J, Liu H T (2021). Application of Li H, Ke L, Chen Z, Feng G, Xia D, Wang Y, Zheng Y, Li Q (2016).
machine learning methods for the prediction of organic solid waste Estimating the fates of C and N in various anaerobic codigestions of
treatment and recycling processes: A review. Bioresource manure and lignocellulosic biomass based on artificial neural
Technology, 319: 124114 networks. Energy & Fuels, 30(11): 9490–9501
Hannan M A, Zaila W A, Arebey M, Begum R A, Basri H (2014). Li H, Xu Q, Xiao K, Yang J, Liang S, Hu J, Hou H, Liu B (2020a).
Feature extraction using Hough transform for solid waste bin level Predicting the higher heating value of syngas pyrolyzed from
detection and classification. Environmental Monitoring and sewage sludge using an artificial neural network. Environmental
Assessment, 186(9): 5381–5391 Science and Pollution Research International, 27(1): 785–797
Heshmati R A A, Mokhtari M, Shakiba Rad S (2014). Prediction of the Li J, Li L, Suvarna M, Pan L, Tabatabaei M, Ok Y S, Wang X (2022a).
compression ratio for municipal solid waste using decision tree. Wet wastes to bioenergy and biochar: a critical review with future
Waste Management & Research, 32(1): 64–69 perspectives. Science of the Total Environment, 817: 152921
Holubar P, Zani L, Hager M, Fröschl W, Radak Z, Braun R (2002). Li J, Pan L, Suvarna M, Tong Y W, Wang X (2020b). Fuel properties
Advanced controlling of anaerobic digestion by means of of hydrochar and pyrochar: prediction and exploration with
hierarchical neural networks. Water Research, 36(10): 2582–2588 machine learning. Applied Energy, 269: 115166
Hoque M M, Rahman M T U (2020). Landfill area estimation based on Li J, Pan L, Suvarna M, Wang X (2021a). Machine learning aided
solid waste collection prediction using ANN model and final waste supercritical water gasification for H2-rich syngas production with
disposal options. Journal of Cleaner Production, 256: 120387 process optimization and catalyst screening. Chemical Engineering
Huang G H, Baetz B W, Patry G G (1998). Trash-flow allocation: Journal, 426: 131285
planning under uncertainty. Interfaces, 28(6): 36–55 Li J, Suvarna M, Pan L, Zhao Y, Wang X (2021b). A hybrid data-
Idwan S, Mahmood I, Zubairi J A, Matar I (2020). Optimal driven and mechanistic modelling approach for hydrothermal
management of solid waste in smart cities using internet of things. gasification. Applied Energy, 304: 117674
Wireless Personal Communications, 110(1): 485–501 Li J, Zhang L, Li C, Tian H, Ning J, Zhang J, Tong Y W, Wang X
Jiang P, Liu X (2016). Hidden Markov model for municipal waste (2022b). Data-driven based in-depth interpretation and inverse
generation forecasting under uncertainties. European Journal of design of anaerobic digestion for CH4-rich biogas production. ACS
Operational Research, 250(2): 639–651 ES&T Engineering, 2(4): 642–652
Junjuri R, Gundawar M K (2020). A low-cost LIBS detection system Li J, Zhang W, Liu T, Yang L, Li H, Peng H, Jiang S, Wang X, Leng L
combined with chemometrics for rapid identification of plastic (2021c). Machine learning aided bio-oil production with high
waste. Waste Management (New York, N.Y.), 117: 48–57 energy recovery and low nitrogen content from hydrothermal
Kabugo J C, Jamsa-Jounela S L, Schiemann R, Binder C (2020). liquefaction of biomass with experiment verification. Chemical
Industry 4.0 based process data analytics platform: a waste-to- Engineering Journal, 425: 130649
energy plant case study. International Journal of Electrical Power & Li J, Zhu X, Li Y, Tong Y W, Ok Y S, Wang X (2021d). Multi-task
Energy Systems, 115: 105508 prediction and optimization of hydrochar properties from high-
Karaca F, Özkaya B (2006). NN-LEAP: a neural network-based model moisture municipal solid waste: application of machine learning on
for controlling leachate flow-rate in a municipal solid waste landfill waste-to-resource. Journal of Cleaner Production, 278: 123928
site. Environmental Modelling & Software, 21(8): 1190–1197 Lima L, Trindade E, Alencar L, Alencar M, Silva L (2021).
16 Front. Environ. Sci. Eng. 2023, 17(4): 44

Sustainability in the construction industry: a systematic review of gasification in a fluidized bed reactor. Waste Management (New
the literature. Journal of Cleaner Production, 289: 125730 York, N.Y.), 58: 202–213
Lin X, Wang F, Chi Y, Huang Q, Yan J (2015). A simple method for Park H I, Park B (2009). Prediction of MSW long-term settlement
predicting the lower heating value of municipal solid waste in induced by mechanical and decomposition-based compressions.
China based on wet physical composition. Waste Management International Journal of Environmental Research, 3(3): 335–348
(New York, N.Y.), 36: 24–32 Qu Y, Qian X, Song H, Xing Y, Li Z, Tan J (2018). Soil moisture
Liu C, Dong H, Cao Y, Geng Y, Li H, Zhang C, Xiao S (2021). investigation utilizing machine learning approach based
Environmental damage cost assessment from municipal solid waste experimental data and Landsat5-TM images: a case study in the
treatment based on LIME3 model. Waste Management (New York, Mega City Beijing. Water, 10(4): 423
N.Y.), 125: 249–256 Rabl A, Spadaro J V, Mcgavran P D (1998). Health risks of air
Magazzino C, Mele M, Schneider N (2020). The relationship between pollution from incinerators: a perspective. Waste Management &
municipal solid waste and greenhouse gas emissions: evidence from Research, 16(4): 365–388
Switzerland. Waste Management (New York, N.Y.), 113: 508–520 Sabrin S, Nazari R, Karimi M, Fahad M G R, Everett J, Peters R
Mehrdad S M, Abbasi M, Yeganeh B, Kamalan H (2021). Prediction of (2021). Development of a conceptual framework for risk
methane emission from landfills using machine learning models. assessment of elevated internal temperatures in landfills. Science of
Environmental Progress & Sustainable Energy, 40(4): 13629 the Total Environment, 782: 146831
Mokhtari M, Heshmati R A A, Shariatmadari N (2015). Compression Saghouri M, Abdi R, Ebrahimi-Nik M, Rohani A, Maysami M (2020)
ratio of municipal solid waste simulation using artificial neural Modeling and optimization of biomethane production from solid-
network and adaptive neurofuzzy system. Earth Sciences Research state anaerobic co-digestion of organic fraction municipal solid
Journal, 18(2): 165–171 waste and other co-substrates. Energy Sources Part A. Recovery
Nabavi-Pelesaraei A, Bayat R, Hosseinzadeh-Bandbafha H, Afrasyabi Utilization and Environmental Effects, 102: 1–17
H, Chau K W (2017). Modeling of energy consumption and Shi M, Wang X, Shao M, Lu L, Ullah H, Zheng H, Li F (2023).
environmental life cycle assessment for incineration and landfill Resource utilization of typical biomass wastes as biochars in
systems of municipal solid waste management: a case study in removing plasticizer diethyl phthalate from water: characterization
Tehran Metropolis of Iran. Journal of Cleaner Production, 148: and adsorption mechanisms. Frontiers of Environmental Science &
427–440 Engineering, 17(1): 5
Nayak S K, Satapathy A (2020). Wear analysis of waste marble dust- Simsek C, Kincal C, Gunduz O (2006). A solid waste disposal site
filled polymer composites with an integrated approach based on selection procedure based on groundwater vulnerability mapping.
design of experiments and neural computation. Journal of Environmental Geology, 49(4): 620–633
Engineering Tribology, 234(12): 1846–1856 Singh D, Chavan D, Pandey A K, Periyaswami L, Kumar S (2021).
Nguyen, KLP, Chuang Y H, Chen H W, Chang C C (2020). Impacts of Determination of landfill gas generation potential from
socioeconomic changes on municipal solid waste characteristics in lignocellulose biomass contents of municipal solid waste. Science
Taiwan, China. Resources, Conservation and Recycling, 161: of the Total Environment, 785: 147243
104931 Sun Y, Tao J, Chen G, Yan B, Cheng Z (2020). Distribution of Hg
Noori R, Abdoli M A, Ghasrodashti A A, Jalili Ghazizade M (2009). during sewage sludge and municipal solid waste Co-pyrolysis:
Prediction of municipal solid waste generation with combination of influence of multiple factors. Waste Management (New York,
support vector machine and principal component analysis: a case N.Y.), 107: 276–284
study of Mashhad. Environmental Progress & Sustainable Energy, Tao J, Liang R, Li J, Yan B, Chen G, Cheng Z, Li W, Lin F, Hou L
28(2): 249–258 (2020). Fast characterization of biomass and waste by infrared
Obileke K, Onyeaka H, Omoregbe O, Makaka G, Nwokolo N, spectra and machine learning models. Journal of Hazardous
Mukumba P (2020). Bioenergy from bio-waste: a bibliometric Materials, 387: 121723
analysis of the trend in scientific research from 1998–2018. Turkdogan-Aydinol F I, Yetilmezsoy K (2010). A fuzzy-logic-based
Biomass Conversion and Biorefinery, 28(2): 1–16 model to predict biogas and methane production rates in a pilot-
Ozcan H, Ucan O, Sahin U, Borat M, Bayat C (2006). Artificial neural scale mesophilic UASB reactor treating molasses wastewater.
network modeling of methane emissions at Istanbul Kemerburgaz- Journal of Hazardous Materials, 182(1–3): 460–471
Odayeri landfill site. Journal of Scientific and Industrial Research, Vaz C A D, Samanamud G L, Da Silva R S, Franca A B, Quintao C M
65(2): 128–134 F, Urzedo A P, Silva M B, Bosch Neto J C B, Amaral M S, Loures
Pan R, Duque J V F, Debenest G (2021). Investigating waste plastic C C A, et al. (2021). Modeling and optimization of hybrid leachate
pyrolysis kinetic parameters by genetic algorithm coupled with treatment processes and scale-up of the process: review. Journal of
thermogravimetric analysis. Waste and Biomass Valorization, Cleaner Production, 312: 127732
12(5): 2623–2637 Viotti P, Polettini A, Pomi R, Innocenti, C (2003). Genetic algorithms
Pan R, Duque J V F, Debenest G (2022). Waste plastic thermal as a promising tool for optimisation of the MSW collection routes.
pyrolysis analysis by a neural fuzzy model coupled with a genetic Waste Management & Research, 21(4): 292–298
algorithm. Waste and Biomass Valorization, 13(1): 135–148 Vu H L, Bolingbroke D, Ng K T W, Fallah B (2019). Assessment of
Pandey D S, Das S, Pan I, Leahy J J, Kwapinski W (2016). Artificial waste characteristics and their impact on GIS vehicle collection
neural network based modelling approach for municipal solid waste route optimization using ANN waste forecasts. Waste Management
 Rui Liang et al. Applications of machine learning in the life cycle of solid waste 17

(New York, N.Y.), 88: 118–130 Omar U, Gull M, Mehmood M A (2018). Evaluating the bioenergy
Vu H L, Ng K T W, Fallah B, Richter A, Kabir G (2020). Interactions potential of Chinese Liquor-industry waste through pyrolysis,
of residential waste composition and collection truck compartment thermogravimetric, kinetics and evolved gas analyses. Energy
design on GIS route optimization. Waste Management (New York, Conversion and Management, 163: 13–21
N.Y.), 102: 613–623 You H, Ma Z, Tang Y, Wang Y, Yan J, Ni M, Cen K, Huang Q (2017).
Wan Y, Xiao L, Wu C (2009). An Optimum Intelligent Algorithm and Comparison of ANN (MLP), ANFIS, SVM, and RF models for the
its Application in Population Statistic and Forecast, 2009 WRI online classification of heating value of burning municipal solid
Global Congress on Intelligent Systems, pp. 40–44 waste in circulating fluidized bed incinerators. Waste Management
Wang Y, Lai N, Zuo J, Chen G, Du H (2016). Characteristics and (New York, N.Y.), 68: 186–197
trends of research on waste-to-energy incineration: a bibliometric
Zhang H, Yu S, Shao L, He P (2019). Estimating source strengths of
analysis, 1999–2015. Renewable & Sustainable Energy Reviews,
HCl and SO2 emissions in the flue gas from waste incineration.
66: 95–104
Journal of Environmental Sciences (China), 75: 370–377
Wang Z, Peng X, Xia A, Shah A A, Huang Y, Zhu X, Zhu X, Liao Q
Zhang W, Li J, Liu T, Leng S, Yang L, Peng H, Jiang S, Zhou W, Leng
(2022). The role of machine learning to boost the bioenergy and
L, Li H (2021). Machine learning prediction and optimization of
biofuels conversion. Bioresource Technology, 343: 126099
bio-oil production from hydrothermal liquefaction of algae.
Wen J, Yan J, Zhang D, Chi Y, Ni M, Cen K (2006). SO2 emission
Bioresource Technology, 342: 126011
characteristics and BP neural networks prediction in MSW/coal co-
Zheng H, Gu Y (2021). EnCNN-UPMWS: waste classification by a
fired fluidized beds. Journal of Thermal Science, 15(3): 281–288
Yan B, Liang R, Li B, Tao J, Chen G, Cheng Z, Zhu Z, Li X (2021). CNN ensemble using the UPM weighting strategy. Electronics
Fast identification and characterization of residual wastes via laser- (Basel), 10(4): 427
induced breakdown spectroscopy and machine learning. Resources, Zhong S, Zhang K, Bagheri M, Burken J G, Gu A, Li B, Ma X,
Conservation and Recycling, 174: 105851 Marrone B L, Ren Z J, Schrier J, et al. (2021). Machine learning:
Yan L, Li Y, Yang B, Farahani M R, Gao W (2018). Air-steam new ideas and tools in environmental science and engineering.
gasification of municipal solid wastes (MSWs) for hydrogen Environmental Science & Technology, 55(19): 12741–12754
production. Energy Sources, Part A: recovery, Utilization, and Zhang Y, Li J, Liu H, Zhao G, Tian Y, Xie K (2021). Environmental,
Environmental Effects, 40(5): 538–543 social, and economic assessment of energy utilization of crop
Ye G, Luo H, Ren Z, Ahmad M S, Liu C G, Tawab A, Al-Ghafari A B, residue in China. Frontiers in Energy, 15(2): 308–319