Available online at www.sciencedirect.

com

ScienceDirect
Transportation Research Procedia 20 (2017) 47 – 52

12th International Conference "Organization and Traffic Safety Management in large cities",
SPbOTSIC-2016, 28-30 September 2016, St. Petersburg, Russia

Methods of Multi-Criteria Optimization in Problems of Simulation

of Trucking Industry
Belokurov Sergey, Belokurov Vladimir *, Zolnikov Vladimir, Cherkasov Oleg
Voronezh State G.F. Morozov Forestry Engineering University, 8 Timirjazeva str., Voronezh, 394087, Russia

Abstract

Efficiency of a transport enterprise can be substantially enhanced by optimization methods at the stages of synthesis, analysis and
decision-taking, which makes it possible to obtain, in specific situation, the best of possible solutions. The simulation of motor
transport enterprise’s operation is closely associated with using vector numeric schemes that practically present structural and
parametric models of motor transport enterprise’s functions. This is connected with the fact that motor transport enterprises are
defined by a relatively large quantity of their specific and individual properties that become sophisticatedly interdependent in the
course of their functioning.
The article describes the basic features and methods of multi-criterion optimization in simulating the motor transport enterprises’
operations.
©©2017
2016TheTheAuthors.
Authors. Published
Published by Elsevier
by Elsevier B.V.B.V.
This is an open access article under the CC BY-NC-ND license
Peer-review under responsibility of the organizing committee of the 12th International Conference "Organization and Traffic
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility
Safety Management of the organizing committee of the 12th International Conference “Organization and Traffic Safety
in large cities".
Management in large cities”
Keywords: Multi-criterion optimization, simulation, choice of solutions, motor transport enterprise, solution quality.

1. Main text

In organizing and managing of motor transport enterprises (MTE) (for example, the passenger ones) the tasks of
efficient use of the passenger transport were always essential, which amounted to building a rational routing system
of the city, optimum traffic schedule, passenger servicing quality, passenger safety, etc. All this raises certain

* Corresponding author. Tel.: +0-000-000-0000 ; fax: +0-000-000-0000 .

E-mail address: [email protected] *

2352-1465 © 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the 12th International Conference “Organization and Traffic Safety
Management in large cities”
doi:10.1016/j.trpro.2017.01.010
48 Belokurov Sergey et al. / Transportation Research Procedia 20 (2017) 47 – 52

requirements for the efficiency of motor transport enterprise operation management. A systematic approach,
methods and models of a systematic analysis and the theory of mathematical simulation are the most effective tools
for addressing this problem [Belokurov and Skryl (2011/2002/2008), Hibbs (2000), Kaufmann (1988)]. This is
connected with the fact that the motor transport enterprises’ systems under study are described by a fairly large
quantity of their specific features, which become sophisticatedly interdependent during functioning and are
described by multi-criterion optimization models (MCOM) [Belokurov (2008/2004)].
Modern systems of object application, such as models of a motor transport system and, generally, of passenger
transport of a large city are described, as a rule, by a fairly large quantity of qualitative and quantitative factors and
the existence of complex dependencies between them.
The simulation and optimization of parameters and functioning modes of a MTE is a difficult task of a large
scale. One of the means of solving this task is employment of an efficient apparatus of multi-criterion optimization
[Belokurov et al. (2004), Belokurov and Belokurov (2000)].
The use of MCOM procedures implies a number of organizational and computational restrictions. So, for
instance, the exhaustive scanning of a large number of versions of non-dominated solutions from iteration to
iteration results in over-filling of the computing environment memory, in which case the search time increases and
the accuracy of obtained solutions decreases.
Conversion of the initial vector form of presenting criteria to the scalar mode calls for theoretical substantiation of
the causes of choosing one or other "main" criterion or means of building a criterion-based convolution, which, in
turn, reduces the efficiency of the solution obtained and calls for substantiation of the adequacy of the resultant
scalar model [Belokurov et al. (2004), Belokurov (2009)].
The main drawback of the existing methods of screening and choosing the solutions is that the selection of part of
the "good", from the viewpoint of the set task of versions is based, as a rule, on a random choice. Under such
circumstances there are no objective reasons for choosing one or other part of the multitude of non-dominated
solutions.
A step-by-step search of solutions and the selection conducted by using it is described by different means of
formalization linked to a specific method. Acceptance of a final solution on the formed multitude of non-dominated
alternatives is also hampered [Belokurov (2009)]. These specific features of MCOM schemes present, in their turn,
higher requirements for mathematical provision and software.
Obviously, the most convenient method of addressing the task is the use of efficient procedures leading to the
improved quality of the solution obtained. Thus it is necessary to choose the solution on a sound basis both in search
iteration and in choosing the final solution.
Let’s formalize the mathematical model of the task of multi-criterion choice. Let there be m of initial versions of
solution, each of which has a certain multitude X i  D(i 1, m,D D1 u D2 u ...u Dm ) of its realization (x is a
symbol of Descartes product). Each realization of the X version is determined by a certain multitude X – quantitative
indicator of task parameters.
The choice of a concrete solution is linked to a certain value of the optimization criterion vector value
Q (Q1 ( X ),...,Qn ( X )) .
Then MCOM task is generally reduced to the model [Belokurov and Skryl (2011), Belokurov et al. (2009)]:

Q (Q1 ( X ), Q2 ( X ),...,Qn ( X )) oOpt,

X D

D : D1 u D2 u ...u Dm ,
(1)
X {Ai d X i d Bi , i 1, m},
fP ( X i ) d 0, i 1, m,P 1,2,3,...,
where: Opt is an operator realizing some optimization principle; X is a version of solution determined by the
multitude X = ( X1 ,...,X m ) – technical-economic, or quantitative parameters and setting parametric restrictions
for the area of search; Q – is a vector of optimization criteria; A, B – are parametric restriction for the area of search
D, representing a multitude of possible versions of solution; fμ – are functional restrictions the area of search D.
 Belokurov Sergey et al. / Transportation Research Procedia 20 (2017) 47 – 52 49

In optimization as per model (1) in conditions of its realization on a computer, a number of problems emerge,
such as:
1. Flexible tuning for an object area, analytical setting parametric and functional restrictions in D.
2. Choice of a numeric optimization scheme. Possible use of arbitrary scheme combinations for best results.
3. Efficient restriction of power growth of a multitude of not worst solutions on search iterations with a
minimum loss of significant versions.
4. Resort to efficient methods of extrapolating expert evaluations not only in accepting the final solution, but also
directly on search iterations.
5. Choice of the search direction with the help of the search control subsystem. A possible change on any
parameter iteration and mode.
6. Choice of an objective condition of a final search.
7. Isolation and ranging of non-dominated solutions by degree of usefulness with employment of expert
procedures.
The solution of the above mentioned problems involves the study of existing MCOM models, the building of
invariant, from the viewpoint of the assigned task, models and the search algorithms.
From the position of the systematic approach [Belokurov (2009), Belokurov et al. (2009/2005)] its is possible to
single out three main stages of model realization (1):
1. Singling out a permissible multitude of solutions meeting the restrictions in the region D.
2. Singling out, from the permissible multitude of solutions, a sub-multitude of non-dominated alternatives.
3. Obtaining, in a sense optimal, solution to a non-dominated multitude of alternatives.
As far as the number of possible alternatives is very high and a conflict may occur between optimization criteria,
the MCOM task has to be addressed at the second stage with subsequent determination of the final solution relative
to the multitude obtained [Belokurov (2008), Belokurov et al. (2004)].
At any stage of numeric MCOM a need arises to choose optimization parameters or criteria and versions of using
one or other method as well as part of the already formed multitude of solution versions. In this connection it would
be reasonable to describe this choice with the aid of the existing tool of selection theory.
Formalization of solutions selection on search of iterations in the language of selection theory is one of the
principally new approaches in the study of numeric MCOM schemes enabling the construction of efficient man-to-
machine algorithms and their readjustment at any stage which provides a powerful tool of flexible control used in
solution search and acceptance.
Effective control of power growth in Pareto multitude and substantiated choice of solutions on search iterations
improves the quality of the obtained result, reduces the total time of resolving the MCOM problem while
formalization of typical choice situations using the built functions and choice mechanisms makes it possible to raise
the results of theoretical studies to the level of accessible and understandable practical recommendations and
software packages for supporting the accepted solutions with adaptive screening procedures [Belokurov et al.
(2010), Belokurov et al. (2010), Gasnikov (2010)].
The building of the optimization model is linked to formalization of the informative task setting, i.e. its record
with the use of mathematical relations in form (1). In this case, it is necessary to determine the interdependence
between the input and output parameters and establish criteria of optimality and functional restrictions in analytical
or algorithmic form. This stage is also associated with rating (or normalization) of the control parameters.
There is a linear normalization xiH xi / [i ,
where [ i is a constant equal to a unit of measuring the parameter xi and a logarithmic normalization
x i
H
ln(xi / [i ) . The advantage of the logarithmic normalization is in operating with relative accruements of
parameters.
The model typification is linked to its transformation and reduction to a form suited to using standard
optimization procedures. There is no all-purpose algorithm effectively resolving all of the optimization problems.
One and the same algorithm, in resolving different problems, can operate with different efficiency. Therefore a
multi-method mode of resolving optimization tasks is used in practice.
50 Belokurov Sergey et al. / Transportation Research Procedia 20 (2017) 47 – 52

We propose a block-diagram, invariant to the object region, reflecting the main stages and tasks arising in vector
optimization models of motor transport enterprise control in Fig.1 [Belokurov (2008/2004)].
The choice of optimization method is a very important task significantly impacting the optimization result. The
optimization methods are divided in classes, for each of which its own solution algorithms are developed. However,
even in a framework of a single class the resolution of optimization problems may need various optimization
procedures.
After that, parameters (constant values) are established, depending on the chosen method. The choice of the initial
approach is an iterative sequence of synthesis, analysis, and decision-taking. At the stage of synthesis, in accordance
with a chosen algorithm, the follow-on values of input parameters are determined, i.e. the next version of solution is
synthesized.
The stage of analysis is linked to evaluation of the obtained version, i.e. calculation of quality values and
restrictions corresponding to the obtained entry parameter values.
At the stage of decision taking it is determined whether the current version is acceptable in terms of the assigned
task. If a satisfactory version is obtained, the optimization process is completed. If not, a transfer is made to the next
iteration. Normally, the quality control of the version obtained is carried out in the computing procedures
automatically. In this case, at the initial stage of optimization, a criterion is determined for breakdown algorithm
whose performance completes the job. On each iteration the breakdown criterion is checked automatically and, in
the event of its non-performance, the iterative optimization process goes on.
If the breakdown criterion is fulfilled (the optimization algorithm completed the function), but the present version
does not suit the designer, the following actions can be performed:
x correct the initial approach;
x change the method parameters;
x change the method;
x correct the optimization method.
Thus, the decision-taking stage in simulating the effective operation of the MTE calls for substantiation of the
models proposed at the stage of their synthesis and analysis. Therefore, before passing over to analysis of the
multitude of non-dominated solutions based on the methods of expert evaluations extrapolation, a theoretical
investigation is supplied of the models and algorithms for MCOM tasks.
It should be noted that the studied methods of multi-criterion optimization in simulating the MTE operation may
be widely used also in other applied tasks. It is possible to include in them the tasks of designing, planning, resource
distribution and many others related to MCOM simulation.
Belokurov Sergey et al. / Transportation Research Procedia 20 (2017) 47 – 52 51

Fig. 1. General optimization scheme for MTE management tasks.

52 Belokurov Sergey et al. / Transportation Research Procedia 20 (2017) 47 – 52

References

Belokurov, S.V. (2002). The Technology of Building the Selection Scheme for Vector Optimization Tasks. Bulletin of the Voronezh Institute
of the Russian Interior Ministry, 1 (10): 100–104.
Belokurov S.V., Skryl S.V., Belokurov, V.P. (2010). Decision-Taking for Effective Management of Transport Systems Based on Selection
Situations. Proceedings of the Russian Academy of Sciences. (2): 6–12.
Belokurov, S.V. (2008). Effective Algorithm of Choosing Non-Dominated Solutions in Numeric Vector Schemes. Bulletin of the Voronezh
Institute of the Russian Interior Ministry, (2): 141–145.
Belokurov, S.V. (2008). Transport Systems Simulation in Choosing a Solution on Search Iteration in Numeric Vector Schemes. Transport.
Science, technology, management: scientific information. Proceedings of the Russian Academy of Sciences, (11): 12–14.
Belokurov, S.V. (2009). Optimization of Multipurpose Transport Tasks in Using the Algorithm of Analysis and Screening on Iterations of
Solution Search. Bulletin of the Russian Academy of Sciences, (6):2–4.
Belokurov, S.V., Belokurov, V.P. (2000). Models of Multi-Criterion Phased Choice in Information Systems Used in Controlling Transport
Processes. Bulletin of the Russian Academy of Sciences, (7): 11–14.
Belokurov, S.V., Bugayev, Yu.V., Serbulov, Yu.S. (2005). Models of Choosing Non-Dominated Versions in Numeric Multi-Criterion
Optimization Schemes. Voronezh, 199 p.
Belokurov, S.V., Pitolin, M.V., Karpov, G.M. (2009). Structure of the Procedure of Dynamic Systematic Analysis of Transport Systems.
Bulletin of the Voronezh Institute of Russian Federal Corrective Service, (4): 67–69.
Belokurov, S.V., Skryl, S.V. (2011). Traffic Management Models (Based on Examples of Russian Internal Affairs Ministry Department of
Traffic Control). Voronezh: Publishing House of the Voronezh Institute of the Interior Ministry, 265 p.
Belokurov, S.V., Skryl, S.V., Belokurov, V.P. (2010). Special Features of the Model of Optimum Management of Solutions Screening Based
on Synthesis of Choice Theory in Transport Systems. Bulletin of the Russian Academy of Sciences, (1): 5–9.
Belokurov, S.V., Velichko, S.V., Solovei, D.Ye. (2004). Models and Algorithms of Decision-Taking on Search Integrations in Numeric
Vector Schemes. Bulletin of the Saratov State Aerospace University n.a. S.P. Korolyov, 1 (14): 196–200.
Belokurov, S.V., Velichko, S.V., Solovei, D.Ye. (2004). Synthesis of Choice Functions on Search Integrations in Numeric Models of Multi-
Criterion Optimization. Voronezh: Publishing House of the Voronezh State University, 125 p.
Gasnikov A.V. (2010). Introduction to Mathematical Simulation of Transport Traffic. Moscow: Moscow Institute of Physics and Technology,
452 p.
Hibbs, J. (2000). Transport policy: The myth of integrated planning. London: The institute of economic affairs, 111 р.
Kaufmann A., Gupta, M.M. (1988). Fuzzy Mathematical Models in Engineering and Management Science. Amsterdam: North-Holland,
195 p.