Research on distribution automation security situational awareness technology based on risk transmission path and multi-source information fusion

Abstract

It may be difficult for existing methods to make full use of the correlation and complementarity of various kinds of information when processing multi-source information. In order to accurately perceive the security situation of distribution automation and ensure the safe and stable operation of distribution network, the multi-source information fusion distribution automation security situation awareness technology based on risk transmission path is studied. Based on the risk transmission path, the distribution automation security situational awareness factors are analyzed, and the main factors affecting the distribution automation security situation are divided into two dimensions: internal source and external source, and eight main awareness factors; Different types of sensors are set in the main areas of security situational awareness factors to collect data of different awareness factors. Using ant colony algorithm to optimize DS evidence fusion method, data with different perception factors are fused, and data fusion results with different perception factors are obtained. The distribution automation security situational awareness model is constructed, and the security situational awareness results are obtained based on the data fusion results of the awareness factors. If the results are higher than the set threshold, the abnormal signal can be output to determine the area where the distribution automation abnormal equipment is located. The experimental results show that the multi-source data fusion effect of this method is good, and it can accurately perceive the security status of different nodes of the experimental object at different time nodes.

Introduction

Ensuring the safe operation of the distribution network is the main way to guarantee the operation level of the power system (Zhang et al. 2019). During the application of the distribution network, there is a high probability of accidental occurrences due to various factors, which can affect the operation of the distribution network and even the entire power system (Qian et al. 2019). Therefore, the analysis of the distribution network's safety status has received widespread attention in relevant fields (Xinrui et al. 2019).

The distribution automation security situation awareness technology is the application of situation awareness technology in the distribution network's security aspect. Ding et al. (2020) introduced the Bayesian method into the security situation awareness issue, constructed the situation awareness index grading standard based on the situation awareness index data, and obtained the situation awareness results layer by layer based on the Bayesian network. However, this method has poor application efficiency due to the long data interaction time in the practical application process. Jiagen et al. (2019) introduced the RBF neural network into the security situation awareness problem, determined the nonlinear mapping correlation of the security situation awareness value, and used the genetic algorithm to solve the network structure and parameters. However, this method ignored the main factors that affect the target security situation. Risk transmission path analysis is a key step in understanding how risks are transmitted and spread within a distribution system. Through in-depth analysis of the physical structure of the system, equipment performance and operating environment, we can extract the main parameters that affect the safety situation of distribution automation, such as equipment failure rate and risk propagation speed. These parameters not only reflect the current security state of the system, but also provide an important basis for predicting the future situation change. In order to make full use of various sensor data, this paper adopts multi-source information fusion technology to integrate data from different sources. Ant colony algorithm, as an optimization method, can improve the accuracy and reliability of fusion results by intelligently adjusting the parameters in the fusion process. The ant colony algorithm is used to optimize the DS evidence fusion method, which can realize the effective fusion of different perception factors, so as to obtain more comprehensive security situation awareness results. To address these issues, we propose a multi-source information fusion distribution automation security situation awareness technology based on risk transmission paths to reduce the burden on distribution network managers.

Previous work has proposed a variety of methods for locating abnormal devices, these methods improve the accuracy and efficiency of anomaly location to a certain extent, but there are still some limitations, such as poor anomaly detection in complex environments and limited ability to identify unknown anomalies.

The multi-source information fusion technology based on risk transmission path proposed in this paper not only considers the state information of the equipment, but also considers the transmission process of the risk in the power distribution system, so as to reflect the power distribution security situation more comprehensively. In addition, the novelty of multi-source information fusion technology based on risk transmission path is that it integrates risk transmission path analysis and multi-source information fusion technology to realize comprehensive and accurate perception of distribution automation security situation. This technology not only focuses on the monitoring of equipment status, but also focuses on the transmission process of risks in the power distribution system, so that potential security risks can be detected earlier and corresponding preventive measures can be taken.

The fusion method is superior to Bayesian-based perception techniques or RBF neural network perception techniques can realize comprehensive and accurate perception of power distribution security situation through comprehensive processing of multi-source information, so as to find and deal with potential security risks more effectively.

The research contribution of this paper:

• (1) Setting the security situational awareness factor of distribution automation is beneficial to the evaluation of potential risks in distribution automation system. Setting different kinds of sensors in main areas, collecting data of different sensing factors, and obtaining information of power system operation state, load situation, environmental conditions, etc., can perceive the security situation of the system in real time and find out abnormal situations and risks in time.

• (2) The ant colony algorithm is used to optimize the DS evidence fusion method, and the data of different perceptual factors are fused to enhance the ability to judge events or targets. According to the probability of data source selection, the problems of trust modeling and information conflict in data fusion are solved, and the accuracy and reliability of data fusion are improved.

• (3) Construct the security situation awareness model of distribution automation, determine the area where abnormal equipment of distribution automation is located, locate the fault and abnormal equipment, and improve the efficiency of fault diagnosis and maintenance. It is helpful to improve the safety of distribution automation system and reduce energy waste and cost loss.

Distribution automation security situational awareness technology

Factor analysis of distribution automation security situational awareness based on risk transmission path

The data involved in different perception factors include system state, load and environmental conditions. By comprehensively analyzing these factors, we can determine the possible abnormal situation of the system and give early warning so as to take corresponding measures to avoid accidents. Starting from the source of distribution automation risk transmission (Liwen et al. 2020), the distribution automation risk transmission path is divided into two types: risk transmission from the internal source of distribution automation and risk transmission from the external environmental factors of distribution automation:

• (1) The risk transmission from the internal source of distribution automation describes the risk source from the internal side of distribution automation. Due to the uncertainty and error of different equipment and facilities in actual operation of distribution network, it is transmitted and amplified through the operation and benefit chain of distribution network, which causes certain risks to the distribution network itself and the external environment of distribution automation. In the risk transmission from the internal source of distribution automation, it can be divided into five different transmission modes in detail, which are: transmission from the internal source of distribution automation to external natural disaster environmental factors; transmission from the internal source of distribution automation to external foreign object short-circuit environmental factors; transmission from the internal source of distribution automation to external construction damage environmental factors; transmission from the internal source of distribution automation to external traffic damage environmental factors; and transmission between internal sources of distribution automation.

The risk transmission caused by external environmental factors in power distribution automation describes the risk source coming from outside of power distribution automation, which exists in the external environment of power distribution network operation, such as natural disasters, foreign object short circuits, construction damages, and traffic damages. As the interaction between the power distribution network and the external environment continues, the risk gradually propagates into the content of power distribution automation, which brings certain risks to the implementation and application of power distribution automation. Regarding the transmission from external risks to internal risks in power distribution automation, it can be divided into five different transmission modes: transmission from external natural disasters to power distribution automation internal, transmission from external foreign object short circuits to power distribution automation internal, transmission from external construction damages to power distribution automation internal, transmission from external traffic damages to power distribution automation internal, and transmission between external environmental factors of power distribution automation.

Figure 1 shows the risk transmission path diagram of power distribution automation, in which the bidirectionality, direct or indirectness of power distribution automation risk transmission path makes the distribution of power distribution automation risk transmission path present a mesh-like structural characteristic.

Fig. 1

Distribution automation risk transmission path diagram

The various factors shown in Fig. 1 are the main factors that cause changes in the security situation of power distribution automation, which are the main factors of security situation perception. Different types of sensors (Shan et al. 2019) are set up in the main area of security situation perception factors to collect information from sensors in different locations. By utilizing the technology of multi-source information fusion, the data fusion result of security situation perception factors is obtained, and based on this, combined with the actual operating status of power distribution automation, the security situation perception of power distribution automation is achieved. The transmission and evolution process of risk in distribution system is considered comprehensively. The dynamic monitoring and early warning of risks can be realized by combining the risk transmission path with the automatic perception of power distribution security situation. This method can capture the changing trend of risks in real time, discover potential security risks in time, and provide a strong guarantee for the safe operation of the power distribution system. Risk management theory emphasizes the identification, evaluation, control and monitoring of risks. The analysis method based on risk transmission path is precisely based on this theory, by identifying the risk factors in the power distribution system and the transmission path between them, and then assessing the size of the risk and the possible impact, and finally realizing the effective control and monitoring of the risk.

Multi-source information fusion for power distribution automation security situation perception

By fusing information from different sources, more comprehensive and diversified data can be obtained, thus improving the accuracy and reliability of security situation awareness. Different information sources can include sensors, monitoring systems, intelligent devices and other sources. Through the fusion analysis of these information, a more comprehensive system status and operation will be obtained, which will make the detection of potential risks and anomalies more accurate. The technical framework of multi-source information fusion for power distribution automation security situational awareness based on risk transmission path is shown in Fig. 2.

Fig. 2

Structural framework of multi-source information fusion power distribution automation security situational awareness technology based on risk transmission path

Based on security situational awareness factors, the distribution automation security situational awareness is generated layer by layer. The security situational awareness factor describes the main factors that can cause changes in the distribution automation situation, and its data can be obtained through sensors of different categories.

Security situational awareness is implemented by information fusion in two directions: horizontal and vertical. Vertical information fusion represents the multi-source information fusion of equipment security situational awareness factors: the security situational awareness factors of distribution automation equipment operating information are fused into internal sources, and the security situational awareness factors of the external environment information of distribution automation are fused into external sources. On this basis, the equipment security situational value is generated by weighting. Horizontal information fusion represents the state produced by the fusion of different equipment basic operating states during distribution automation operation. Multi-source and multi-level information fusion is constructed through the fusion of vertical and horizontal information.

Multi-source information fusion of distribution automation security situation factors

Considering the diversity of various types of equipment in the operation process of distribution automation (Shuxin et al. 2021; Zhaojun et al. 2020), and the complexity of data, it is crucial to effectively integrate multi-class information from different distribution automation experimental data for distribution automation security situational awareness.

Figure 3 shows the structure of the distribution automation multi-source information fusion model.

Fig. 3

Multi-source information fusion model of power distribution automation

The process of multi-source information fusion in power distribution automation can generally be divided into three stages:

• (1) The main function of the data pre-processing stage is to process the risk source data in the power distribution network through a data normalization process.

• (2) The feature fusion stage uses a Deep Belief Network (DBN) to fuse features.

• (3) In the DS evidence fusion stage, the feature fusion result is processed at the decision level to obtain the data required for the power distribution automation security situation awareness process.

In the DS evidence fusion process, the ant colony algorithm is used to optimize the data fusion process. By organically combining DS evidence fusion algorithm with ant colony algorithm, the credibility of data fusion of different categories can be improved, while effectively achieving the problem of multi-source data fusion in power distribution automation.

Ant colony algorithm is a self-organizing algorithm with strong robustness. In data fusion, this means that the algorithm can automatically adapt to different data environments and fusion needs. Moreover, the Ant colony algorithm has fewer parameters and is simple to set, so it is easy to realize and apply to practical data fusion problems. At the same time, due to its parallelism and adaptivity, the algorithm also has a greater advantage in solving complex data fusion problems.

When applying ant colony algorithm to optimize the parameters of the DS evidence fusion algorithm, there are M ants representing a k-dimensional combination of power automation data sources. Each ant is evaluated using $Amk$ to describe the evaluation of a k-dimensional combination of power automation data sources by the m-th ant, and the fusion priority of power automation data sources is determined using Eq. (1).

$$P_{{X_{t}^{m} }}^{{I_{i} }} \left( t \right) = \left\{ {\begin{array}{*{20}l} {\frac{{T_{{I_{i} }} \left( t \right)}}{{\sum\limits_{{I_{j} \in J_{k\left( t \right)} }} {\left( {T_{{I_{i} }} \left( t \right)} \right)} }},} \hfill & {I_{i} \in J_{{X_{t}^{m} }} \left( t \right)} \hfill \\ {0,} \hfill & {otherwise} \hfill \\ \end{array} } \right.$$

(1)

In Eq. (1), $I_{i}$ and $I_{i} \left( t \right)$ respectively represent the overall combined dataset of a power distribution automation data source and the information pheromone selected during the t-th iteration process and $I_{i}$. $J_{{A_{t}^{m} }} \left( t \right)$ and X represent the set of power automation data sources that have not been used and the security perception situational factors, respectively.

Once a power automation data set is selected, it cannot be used again. Therefore, the power automation data set can be determined based on the local optimum solution state (Zhang Liang and Gang 2021; Yu et al. 2022; Wang et al. 2021; Xu et al. 2022), and Eq. (2) shows the basis for determining the data set.

$$P_{{X_{t}^{m} }}^{{I_{i} }} \left( t \right) = \left\{ {\begin{array}{*{20}l} {1,} \hfill & {I_{j} = rand\left( {I_{{X_{t}^{m} }} \left( t \right)} \right)} \hfill \\ {0,} \hfill & {otherwise} \hfill \\ \end{array} } \right.$$

(2)

Each ant can determine two probability functions according to the corresponding frequency (Li and Jingyi 2021), setting the frequencies of the two different functions as $\theta$ and $1 - \theta$, which can achieve diversified distribution automation information search. Based on this, the overall optimal probability can be determined using relevant mathematical formulas, and thus the corresponding weights can be determined.

The final weight is determined through iterative processing (Leijiao et al. 2021), which obtains the probability of selecting a data source and determines its importance. The expression of importance is as follows:

$$W = (w_{1} ,\;w_{2} , \cdots w_{n} )$$

(3)

Based on the description above, referencing the DS evidence combination theory can optimize the process of multi-source data fusion, and the following steps can be taken:

$$m\left( X \right) = \frac{{\sum\limits_{{X_{1} \cap X_{2} \cap \cdots \cap X_{n} }} {m_{1} \left( {X_{1} } \right)^{{w_{1} }} m_{2} \left( {X_{2} } \right)^{{w_{2} }} \cdots m_{n} \left( {X_{n} } \right)^{{w_{n} }} } }}{1 - G}$$

(4)

In Eq. (4), $G = \sum\limits_{{X_{1} \cap X_{2} \cap \cdots \cap X_{n} = \emptyset }} {\sum\limits_{{X_{1} \cap X_{2} \cap \cdots \cap Xn}} {m_{1} \left( {X_{1} } \right)^{{w_{1} }} m_{2} \left( {X_{2} } \right)^{{w_{2} }} \cdots m_{n} \left( {X_{n} } \right)^{{w_{n} }} } }$, $X \ne \emptyset$.

Construction of security situation awareness model for distribution automation

After obtaining the real-time status data fusion results of all security situation awareness factors $X_{i}$, the probability $P_{ijk} \left( {i = 1,2, \cdots ,N;j = 1,2;k = 0,1,2} \right)$ of different devices $i$ (in the context of distribution automation, device $i$ can be defined as the leaf node) during the distribution automation process is obtained according to the Bayesian network inference theory. Based on the probability $P_{ijk}$ of device $i$, the probability values of internal source and external source generated by the Bayesian network inference at network level are obtained $P_{jk} \left( {j = 1,2;k = 0,1,2} \right)$, thus obtaining the security situation $S_{c}$ of different devices and the security situation $S_{n}$ of distribution automation in both vertical and horizontal directions. The generation process of $S_{c}$ and $S_{n}$ has a high consistency, so only the generation process of $S_{n}$ is explained below.

According to the experience of professional personnel, the weights $q_{i} \left( {i = 1,2} \right)$ and $q_{1} + q_{2} = 100$ of different dimensions are determined. The security situation of distribution automation defined in Fig. 1 is generated by weighting the two dimensions of internal sources and external sources, and the values of different dimensions are determined by subtracting the probability $P_{j2}$ of taking 2 from the probability $P_{j0}$ of multiplying the weights of each dimension and taking 0. This is because even under the abnormal condition of distribution automation, the fluctuation of the threshold value $P_{j2}$ will not be significant, due to the fact that most devices are still in normal operation during the distribution automation process. For the purpose of distinction, $P_{j2}$ is multiplied by $\lambda$ to obtain:

$$S_{n} = \sum\limits_{j = 1}^{3} {w_{j} \left[ {P_{j0} - \lambda \cdot P_{j2} } \right]}$$

(5)

The higher the value of the security situation result $S_{n}$ obtained through Eq. (5), the greater the probability of it being attacked.

The calculation process of $S_{c}$ is basically the same as that of $S_{n}$. Under the condition that the device security situation is compared with the pre-set threshold and is low, it actively transmits abnormal signals to the distribution automation management center. When most devices have abnormal security situations, it will be reflected in the distribution automation security situation $S_{n}$ in real time, and the area where the abnormal distribution automation equipment is located can be determined based on the abnormal signal.

Experimental results and analysis

To verify the practical application performance of the multi-source information fusion distribution automation security situational awareness technology based on risk propagation paths studied in this paper, a small distribution network was selected as the experimental object. The security situational awareness factors of the experimental object were collected using the technology proposed in this paper, and 34 sensors were set up based on this. In order to fully and accurately obtain the security situational awareness factor of the distribution network, the 34 sensors are set up based on the following principle: The sensor setup needs to cover the key nodes and areas of the distribution network. These key nodes and areas are often important channels for power transmission, or critical paths for risk transmission. By setting sensors in these locations, the flow of power, the operating status of equipment and the change of environmental parameters can be monitored in real time, so as to timely capture and sense the security situation of the distribution network. The configuration of sensors should consider the topology and operation characteristics of the distribution network. Different distribution networks have different topologies and operation characteristics, so the setting of sensors also needs to be adapted to local conditions. Through in-depth analysis of the topology and operation characteristics of the distribution network, it is possible to determine which locations are the key points of risk transmission and which areas are the key areas of safety monitoring, so as to set up targeted sensors. The more sensors there are, the richer and more complete the data collected. This contributes to a more complete understanding of the security posture of the distribution network, improving the perceived accuracy and reliability. The sensor setup is shown in Fig. 4. The experimental object was subjected to security situational awareness using the technology proposed in this paper, and the results obtained are as follows.

Fig. 4

The sensor setting of the experimental object

Data fusion results

To analyze the performance of the multi-source information fusion in this paper, the conflict coefficient $K$ is introduced to indicate the degree of conflict between evidence. The $K$ value is positively correlated with the degree of conflict between evidence. However, when the $K$ value is increased to a certain value, the credibility of the evidence in the data fusion process of this paper technology is reduced, resulting in undesirable results. Figure 5 shows the calculation results of the $K$ value in the multi-source data fusion process of this paper technology.

Fig. 5

K value calculation results

From the analysis of Fig. 5, it can be concluded that when using this paper technology to fuse data from different sources, the K value is less than 0.3, indicating that the effect of using this paper technology for multi-source data fusion is good. This is because the method in this paper obtains the security situational awareness factor of distribution automation, and compares the data provided by different information sources to determine the consistency and difference of the data. According to the weight of security situational awareness factors, the abnormal values are filtered, the inconsistency is gradually reduced, and the effect of data fusion is improved.

Safety situational awareness results

The safety situational values of the experimental objects within one day were analyzed using the techniques in this paper, and the obtained results are shown in Fig. 6.

Fig. 6

Safety situational awareness results of the experimental subjects

Based on the analysis of Fig. 6, using the techniques in this paper to perceive the safety situational awareness of Node 4, the safety situational values at 5:00, 14:00, 23:00, and 24:00 were relatively high. This indicates that the probability of Node 4 being attacked is relatively high during these time points, and also demonstrates that using the techniques in this paper can achieve the analysis of the safety situational awareness of the experimental object. This is because this algorithm uses ant colony algorithm to optimize DS evidence fusion method, collects, transmits and fuses data provided by various information sources, and evaluates and determines the weight of each information source by adjusting pheromone concentration, path selection probability and other parameters, thus ensuring the accuracy of the analysis results.

Accuracy analysis of safety situational awareness results

Five randomly selected sensor nodes within the experimental object were used to perceive the safety situational awareness at different time points using the techniques in this paper. The perceived results were compared with the actual safety state to verify the accuracy of the safety situational awareness results obtained from the techniques in this paper. The results are shown in Table 1.

Table 1 Safety situational awareness results

From Table 1, it can be seen that the results of the security situation analysis of different nodes at different times, based on the technical analysis presented in this paper, are consistent with the actual security status of each node. Therefore, it can be concluded that the technology presented in this paper can accurately detect the security status of different nodes at different time points, and has strong practical value. This is because the ant colony algorithm is used to automatically adjust the pheromone concentration and path selection probability according to the distribution network state, obtain the optimal solution, and continuously optimize the weight distribution and information fusion process to ensure the consistency of the results.

Comparative analysis

To further verify the applicability of the technology presented in this paper, five different types of faults were simulated. Bayesian-based perception technology in literature (Ding et al. 2020) and RBF neural network-based perception technology in literature (Jiagen et al. 2019) were used as comparison technologies. The technology presented in this paper and the comparison technologies were respectively used to perceive different faults, and the time taken by different technologies during the perception process was compared. The results obtained are shown in Table 2.

Table 2 Results time time results of different perceptual techniques

The analysis of Table 2 shows that the response time of this method to different types of fault perception is 0.48 ms-1.34 m, while that of the method in reference (Ding et al. 2020) is 0.99 ms-2.23 ms and that of the method in reference (Ding et al. 2020) is 1.01 ms-1.87 ms. Compared with the contrast method, the perception response time of this method is shorter, because this method constructs the security situation awareness model of distribution automation, and quickly processes and analyzes the collected data, which effectively reduces the time and data volume of multi-source data processing.

In order to increase the rigor of the experiment, the perception accuracy of ant colony optimization algorithm was compared with the perception technology based on New Year greetings in literature (Ding et al. 2020) and the perception technology based on RBF neural network in literature (Jiagen et al. 2019). The comparison results are shown in Table 3.

Table 3 Comparison results of perceived accuracy of different technologies

Table 3 shows the comparative results of different technologies in terms of distribution security situation awareness accuracy. As can be seen from the data in the table, the perception accuracy of the ant colony optimization algorithm-based technology proposed in this paper is significantly higher than that described in literature (Ding et al. 2020) and literature (Jiagen et al. 2019) in all kinds of events.

In the perception of substation equipment removal event, the ant colony optimization algorithm reaches 91.34% accuracy, which is significantly higher than 82.43% and 81.57% in literature (Ding et al. 2020; Jiagen et al. 2019). This indicates that the method proposed in this paper has higher accuracy in identifying and predicting substation equipment removal events.

The ant colony optimization algorithm also shows high precision in the perception of disconnect switch, power failure, short circuit, disconnection and other events. Although the perception accuracy of different events is slightly different, on the whole, the method proposed in this paper has more stable perception performance than the techniques in literature (Ding et al. 2020; Jiagen et al. 2019).

Conclusions

In modern society, the security and stability of distribution automation system plays a vital role in energy supply and life support. In order to better monitor and manage the distribution automation system, build a reliable security situation awareness model and determine the area where abnormal equipment is located. The potential risks in the system can be effectively identified and evaluated by setting the security situation awareness factors of distribution automation and setting different types of sensors in the main areas. The ant colony algorithm is used to optimize the DS evidence fusion method, fuse the data of each perceptual factor, and deal with the weight distribution and information conflict to achieve more optimal data fusion. Constructing the security situation awareness model of distribution automation can quickly detect and locate abnormal equipment, improve the efficiency of fault diagnosis and maintenance, reduce power outage time, and provide intelligent decision support for subsequent operation and maintenance work. However, the security situation awareness of distribution automation system is a process of continuous evolution and continuous improvement. With the progress of technology and the change of demand, the security situation awareness model of distribution automation will be continuously improved in the follow-up research with advanced technologies and methods to ensure the stable operation and safe power supply of distribution system. With the permission of research time and conditions, the research scope will be broadened and applied to agriculture and chemical industry to provide more reliable services for the society and bring more convenience to people's lives.