APPLICATION OF INFORMATION TECHNOLOGIES FOR THE OPTIMIZATION OF ITINERARY WHEN DELIVERING CARGO BY AUTOMOBILE TRANSPORT

Current market conditions make the Ukrainian transport sector that provides services not only to domestic enterprises but also to international transportation, very important. According to data from the State Statistics Committee, automobile enterprises of Ukraine delivered 175.6 million tons of freights during 2017, which is 4.7 % larger than in the previous year. The number of enterprises engaged in freight transportation has also increased. There were 2.3 thousand such enterprises in Ukraine at the beginning of 2018, including 2.2 thousand automobile transportation companies [1]. An analysis of statistical data shows positive trends of growth, observed in the automobile transportation industry of Ukraine [2, 3], with, consequently, a further increase in volumes, structure, and geography of transportation of variAPPLICATION OF INFORMATION TECHNOLOGIES FOR THE OPTIMIZATION OF ITINERARY WHEN DELIVERING CARGO BY AUTOMOBILE TRANSPORT


Introduction
Current market conditions make the Ukrainian transport sector that provides services not only to domestic enterprises but also to international transportation, very important.According to data from the State Statistics Committee, automobile enterprises of Ukraine delivered 175.6 million tons of freights during 2017, which is 4.7 % larger than in the previous year.The number of enterprises engaged in freight transportation has also increased.There were 2.3 thousand such enterprises in Ukraine at the beginning of 2018, including 2.2 thousand automobile transportation companies [1].
An analysis of statistical data shows positive trends of growth, observed in the automobile transportation industry of Ukraine [2,3], with, consequently, a further increase in volumes, structure, and geography of transportation of vari-
The volumes of transported freight in the EU countries [4] also tend to grow and made up 1,921,434.5 million tkm in 2017, which is 3.99 % larger than in the previous year.
Development of optimal routes for delivery of goods in international traffic has a goal to increase the efficiency of international transportation.Delivery routes are constantly changing for various reasons, the main of which is the failure to match the volumes of supply against demand.Methods of calculation used in at present do not meet modern requirements of efficiency and require significant time for the relevant estimations.The application of improved methods will make it possible to bring down the cost of transportation of freight by automobile vehicles and improve the competitiveness of domestic carriers.
Given this, one of the urgent tasks of transportation industry at present is the task on optimization of schemes of freight delivery, especially under conditions when the volume of supply does not correspond to demand.Cargo transportation in most cases in practice are represented in the form of a transport problem (TP), which is a particular case of the general problem on linear programming (GPLP).To solve such problems, it is possible to apply the most known method for solving GPLPsimplex method, by first reducing the transport problem to the form of a problem on linear programming and by accounting for the specificity [5].
When managing cargo transportation, especially delivered by automobile vehicles, very often used in practice are the delivery and combined schemes of freight delivery whose optimization may employ the well-known traveling salesman problems.
Thus, the optimization of schemes of freight delivery by international traffic, applying modern information technologies, will make it possible to improve the efficiency of transportation and enhance the competitiveness of transport enterprises of Ukraine.

Literature review and problem statement
The application of information technologies in the field of transportation is gaining importance under modern conditions.Optimization of freight delivery schemes is one of the key tasks in the sector of transportation and logistical operations.In most market segments, delivery of goods adds the amount to its cost that is equivalent to the value of the goods.Along with this, it is worth noting that the use of modern information technologies for the optimization of such deliveries often helps save not less than 5-20 % of the total cost [6].
The issue of development of methods for the optimization of transportation is treated very seriously [7].Organization of cargo transportation, especially delivered by automobile transport, is associated with the use of multi-drop and multi-pick routes for freight delivery.Multi-pick (collected) routes are a kind of circular routes that imply gradual unloading (loading) of cargo in each subsequent route point and returning to the original point of the route.They in essence represent different kinds of the well-known traveling salesman problem.The specified problem gave rise to the development of a separate direction in graph theory, which is known as the search for Hamilton cycles in graphs [8].A Hamiltonian cycles problem in graph theory has received different generalization.One of these generalizations is the task of a salesman that in different modifications often occurs in transport logistics when planning transportations.The task of a salesman is a modified problem on assignment, however, in this case, the binding between the points should form a closed loop.There is also a generalization of the problem, the so-called generalized travelling salesman problem [9].But the use of this method to optimize international transport could not be implemented because it does not take into consideration the factors that are associated with the organization of customs control at the borders and the modes of operation of the drivers in accordance with the European agreement concerning the work of crews of vehicles engaged in international road transport (AETR) [3].
Cargo transportation in most cases in practice are represented in the form of TP, which is a special case of GPLP, it is also possible to tackle it by applying the most known method for solving GPLP -simplex method [10].However, the simplex method cannot be used to calculate the networks of international transport in the case when cargo transportation is not balanced for volumes [11].To solve a given problem, one must reduce the specified transport problem to a tabular form.
Thus, the relevant problem is the need to improve existing methods for the optimization of transportation by applying modern means of information technology.That is necessary to improve the effectiveness of international freight transportation, taking into consideration the features of the organization of customs controls at the borders between states, as well as operational modes of crews of vehicles engaged in international road transport [12].

The aim and objectives of the study
The aim of present study is to improve the efficiency of cargo transportation along international routes by applying modern tools of information technology.
To accomplish the aim, the following tasks have been set: 1.To improve a combinatorial method and develop software for solving a traveling salesman problem along a road transport network (RTN) taking into consideration the patterns in international automobile transportation and customs services and constraints for the throughput of transport infrastructure.
2. To apply the simplex method and develop software for solving a transport problem represented in a network form using the methods that imply reducing a transport problem, unbalanced in terms of transportation cargo volumes, to the transport problem of the balanced form, by introducing an additional fictitious transportation node, proportional and different, as well as by taking into consideration constraints for the throughput of routes along which freight is transported.

1. Representation of a salesman task in the road-transport network and approaches to solving it
Problem statement.There are n transport nodes in RTN.We assign a matrix of distances between them , .
Departing from the initial transport node A 0 , a truck must deliver or collect freight to/from all or the rest of the RTN transport nodes, calling them once, and return to the initial transport node A 0 .Therefore, the route of the truck is circular in its structure.It is required to determine the order to travel to the transport nodes of RTN so that the total distance travelled is minimal.Mathematical model of the problem.We introduce Boolean variables: z i,j =1 if a salesman moves from point A i to point A j ; z i,j =0 -otherwise.Where , 1,2, , ; , where u i , u j are the arbitrary positive integers.Condition (2) defines that a salesman enters each point once, except for the starting point.Condition (3) defines that he leaves each point once.Condition (4) ensures the closeness of the itinerary, which contains n points, and the absence of loops [10].Additionally, we shall introduce the following constraints: where d i,j is the throughput of a route section of cargo transportation from A i to А j , and x i,j is the volume of cargo transported between them.
It should be noted that this constraint should be considered only in the case of significant volumes of cargo transportation along the specified route.
The task of a traveling salesman refers to NP-complete problems, that is, in which at even at a relatively small number of places he visits (66 and more), it cannot be solved by a simple brute force method for all variants (combinatorial technique) by any theoretically possible computers in time less than several billion years.However, based on the practice of freight transportation, the number of transport nodes in multi-drop (combined) transportation routes is significantly less than the above value (20 and less).In addition, combinations of transport nodes, generated and entered into databases in advance, which make up the routes for transportation networks of the specified above dimensionality, represent the Hamiltonian cycles, thereby significantly reducing computational time.
When solving the task of a traveling salesman in real transportation networks, which lack transport links between each transportation node and all the rest, the number of combinations of transportation nodes is significantly reduced, which also has a positive effect on their solving time.

2. Representation of the transport problem in a tabular form and approaches to solving it
We shall represent the process of cargo transportation in a tabular form, that is, in the form of a transport table (TT) (Table 1), where: А i -points of cargo delivery (DP), each of which accumulates, respectively, а i of its volume ( 1, i m = ); B j -points of cargo utilization (UP), which placed orders for this cargo, respectively, of volume b j ( 1, j n = ); c i,,j -the unit cost of transporting a cargo from A i to B j , and x i,j is the volume of cargo transported between them.
The result of solving TP is the minimizing of the objective function C, which is the total cost of cargo transportation, that is, under condition of picking up the stocks of cargo from all its suppliers (а i ) and meeting all the orders (b j ) from all consumers, as well as fulfilling the following constraints (in contrast to constraints (5) with a different route): where d i,j is the throughput of the route of transporting a cargo from A i to B j .
Table 1 Initial transportation table Before filling the simplex table (ST) (Table 2) with values for the parameters from the original TT, we shall perform two preliminary steps, namely: -replace variables x 1,1 with x 1 , x 1,2 with x 2 , x 1,3 with x 3 , etc., x m,n with x mn , that is, we convert two indexes at variable x into one index by multiplying them (for instance, x 3,5 by x 15 ); -add, as required by the simplex method, to all the equations additional variables x mn+1 , x mn+2 and so on to x mn+m+n-1 , which will make up the initial basis solution to TP, by artificially assigning to their coefficients in the objective function C the values that are deliberately larger than all existing coefficients.
Each (m+n-1) ST line corresponds to one of the linear equations that represent either m equations m of TT lines (for example: x 1,1 +x 1,2 +…+x 1,n =a 1 ) or (n-1) equations of TT columns (for example: x 1,1 +x 1,2 +…+x 1,n =a 1 ).These equations are highlighted in Table 1 with a background color (x 1,1 ).Moreover, a value of 1 appears in the ST line in the case when the equation in TT has the corresponding variable x i .
The last ST line contains calculated values of the objective function C and indexes ∆ i , which are the indicators of the end of the optimization process (a condition for the completion of optimization process is non-negative value).
The above non-classic transformation of the transport problem, assigned in the form of a transportation table, into a simplex table, is the basis for its subsequent computer calculation.

Table 2
Initial simplex table An additional feature of the problem statement is its subsequent solving both for the transportation problems, non-balanced in terms of cargo volumes, and for the transportation networks with a large quantity of not only transportation nodes for delivery and utilization of goods but intermediate transport nodes as well.
We developed and compiled the database of infrastructure of Ukraine's transport system, which includes 300 major transportation nodes across the motor roads of international and European importance.The software for the optimization of international freight transportation in the transport system of Ukraine takes into consideration the throughputs of transport infrastructure, as well as capable of solving transport problems under condition of disbalanced volumes of cargo transportation.

Application of modern information technology tools
for the optimization of cargo delivery schemes along international routes 5. 1. Software-based improvement of the process of international cargo transportation in order to solve a traveling salesman problem We shall use an example of TP in the form of a solution to the problem on transporting a cargo from the Zhytomyr oblast to Poland.The cargo is wooden pellets, since wood and products made from it account for 23 % of the total export of Zhytomyr oblast.The cargo to be picked up in Poland is wooden furniture.The goods to be delivered were selected based on data from the State Statistics Service of Ukraine [1], laws and regulations, as well as the economic and social situation in this country.We chose 10 pick-up points in Zhytomyr Oblast with the largest volumes in the production of wood pellets, specifically: Dubrivka, Romaniv, Lubar, Malin, Ovruch, Novohrad-Volynskyi, Zhytomyr, Korosten, Berdichev, Radomyshl (Fig. 1).
The decisive factors when choosing the checkpoints (c/p) were the duration of customs operations, as well as the distance from the points of cargo dispatch.The following c/p are in the territory of Ukraine: Yagodin, Ustilug, Uhryniv, Rava-Ruska, Hrushiv, Krakovets, Shehyni, Smilnica.
Destinations points in Poland were selected based on the fact that there are consumers for wood pellets, as well as furniture factories that export their finished products to Ukraine.Thus, the destinations are the following: Słupsk, Verušuv, Racibórz, Elblag, Morąg; Brodnica, Warsaw, Kielce, Ostrowiec Świętokrzyski, Vengruv.
Employing the database management system Microsoft Access, we built a database with appropriate distances between c/p, points of departure and destination points (Fig. 2).
For convenience, the data on these distances are automatically converted into a file of the spreadsheet processor Microsoft Excel (Fig. 3).The software to solve a traveling salesman problem was developed using the algorithmic programming language Delphi [13].
To begin, we enter temporal characteristics of cargo handling operations along a given route.Loading and unloading the pellets and furniture will be mechanized.We assign service time at each c/p and the mean technical speed of a vehicle (V t =65 km/h).).It should be noted that when adding c/p to the route, the software operates in two modes: manual and automated.That is, c/p can be selected independently, or the software performs this automatically, by choosing the nearest one to the point of departure.The software generates a respective table, which clearly shows distances between the specified cities and c/p.We build possible closed routes and choose the shortest route among all options (Fig. 4).
By using a combinatorial technique, we build possible closed routes and choose the shortest route among all options (Fig. 4).
Next, based on the results of a combinatorial technique [14], the software generates the resulting shortest route consisting of sections that it includes (Fig. 5).
Thus, we determined that of all the software-identified 24 possible routes, the most efficient one would be:

Fig. 6. Graphical representation of a road transport network
At the edges of a given RTN, we assigned distances between the corresponding points, in kilometers, and the volumes of cargo, in tons (indicated at the vertices of the graph), which need to be picked up at DP (indicated with sign −) and must be delivered to UP (indicated with sign +).It is required to minimize transportation costs under condition that there should be 20 tons of cargo more than currently is, specifically:  Optimization of freight transportation in RTN will be conducted by using the computerized optimization system of cargo transportation in RTN (COSCT in RTN) [15].The result of work of a computerized system whose methodological basis for optimization is the simplex method is shown in Fig. 7. Preliminarily, we reduce the proposed transport problem to a tabular form -transport table (TT) (Table 1).
In the case of reducing open TP to the balanced form by introducing an additional fictitious DP A F , which in our case is the transit point С 1 (Yahodyn), this TT will contain the result of solving a given TP (Table 1), which is shown with bold numbers in it.3 gives optimal cargo transportation volumes, which are achieved along the most economically beneficial routes; the distances of freight delivery between the respective points are the numbers in the upper right corner of each table cell.These routes were obtained using the method of finding the shortest routes along a transportation network [16]; the actual volume of transportation work is 174,680 tkm.This is explained by the fact that cargo with a volume of 20 tons is not actually taken from fictitious point С 1 to UP B 1 .In Fig. 8 presented the results of the KSOVP on the balanced proportional method of RTN.The basis of this method lies is proportional to the volume of stocks (applications) reduction proposals (demand) cargoes of all without exception, PP or PS.In the case when demand for a cargo exceeds its supply, we calculate respective coefficient k for a reduction in the volume of orders from all DP ( j=1, n) according to formula: Next, these volumes of orders are decreased to magnitudes * (1 )* Table 4 gives TT with the results of optimization.Fig. 9 shows result of COSCT operation in RTN, balanced by a difference method.When using this method to reduce cargo transportations to the balanced form, difference module 1 1 which has the highest value for demand.This method is not applicable in the case when the greatest demand value, which decreases, is less than the module of difference In our case, this is В 3 and its new value is equal to В 3 =80−20=60 tons.Table 5 gives TT with the results of optimization.We shall try to obtain an optimal plan of cargo transportation in the unbalanced RTN (result of the work of the computerized system is shown in Fig 10), by preliminarily reducing the proposed TP to a tabular form -TT (Table 4).The result of solving this TP (Table 6) is given by bold numbers.Data from Tables 3-6 show the optimal plans to transport a cargo in the balanced and non-balanced RTN, obtained using the simplex method, do not converge.
In Fig. 11, the RTN dotted lines indicate the most economical (minimal) optimal plan of freight transportation (165,300 tkm), obtained using a difference method for reducing a TP, unbalanced in terms of cargo transportation, to the TP in the balanced form.

Discussion of results of the application of information
technologies for the optimization of cargo delivery schemes along international routes 6. 1. Solving a traveling salesman problem using software Our study addresses the use of modern information technologies in solving a salesman problem in order to optimize the process of compiling the itinerary for cargo transportation along international routes.The following factors were largely ignored when solving the set problem: -distances between the points of departure, destination points, and customs posts; -duration of service at checkpoints (customs clearance); -duration of cargo handling operations; -the mean speed of a vehicle; -rest time in line with AETR.
In our calculations, we used real data about the location of points of departure, destination points, and CP at the State border of Ukraine, distance between them, as well as the mean speed.
We report a procedure for the automation of process aimed at solving a traveling salesman problem by a combinatorial method, which takes into consideration existing requirements and constraints for the specificity and dimensionality of the problem.For convenience, data on the distance between the points of transportation are automatically converted into a file of the spreadsheet processor Microsoft Excel.The software for solving a traveling salesman problem was developed using the algorithmic programming language Delphi.
It should be noted that when adding intermediate transportation points to the route, the software operates in two modes: manual and automated.That is, intermediate points can be selected independently, or the software performs this automatically, by choosing the nearest one to the point of departure.
In addition, combinations of transportation nodes, generated and entered into databases in advance, which make up the routes for transportation networks of any dimensionality, represent the Hamiltonian cycles, thereby significantly reducing computational time.
Based on the experience of cargo transportation, the number of transport nodes in multi-drop (combined) routes of cargo transportation does not exceed 20.That is why all variants of cargo transportation routes, generated and entered into databases in advance, in transportation networks with a dimensionality below the magnitude specified above (20), which represent the Hamiltonian cycles, significantly reduce computational time.The software generates a corresponding table, which clearly shows distances between the specified cities and intermediate points.This allows us to build possible closed routes and choose the shortest variant.By using a combinatorial method, the software calculates the total distance of the route and selects the one that is the shortest.

2. Solving a transport problem using software
The proposed simplex method showed its effectiveness when solving both the balanced and non-balanced international cargo transportation.
By using the simplex method, we optimized the transport problem represented in a network form, by three methods for reducing a transport problem, unbalanced in terms of cargo transportation volumes, to the balanced form, specifically the introduction of an additional fictitious transportation node, proportional method, and a difference method.
Optimization of cargo transportation in RTN was carried out using a computerized optimization system of freight transport, by preliminarily reducing the proposed transport problem to the tabular form -a transportation table.
Underlying the method of computerized optimization system of freight transport in a road-transport network, balanced by a proportional method, is a decrease in proposal (demand), proportional to the volume of stocks (orders), for cargoes from all, without exception, delivery points (utilization).In the case demand for a cargo exceeds its proposal, we calculate a corresponding coefficient of reduction in the volume of orders from all checkpoints.
The application of a computerized optimization system of cargo transportation in a road transport network, supplemented by a decision support system, will make it possible practically always choose the most beneficial solutions among all the proposed.

Conclusions
1.The application of a combinatorial technique for solving a traveling salesman problem in a road transport network made it possible to take into consideration the existing requirements and constraints for the specificity and dimensionality of the problem, which allowed a reduction in the transportation costs, by 8 % on average, for delivering the goods to end users along international automobile routes.
2. The approach to employing the simplex method for solving a transport problem, represented in a network form, has proven its effectiveness when solving both balanced and non-balanced cargo transportation.The developed computer system implies a reduction, proportional to the volume of stock (orders), in the proposals (demand) for cargoes from all, without exception, supply points (utilization).That made to possible to reduce the volume of transportation operations, performed along international routes, by 6 % on average.

Fig. 1 .Fig. 2 .
Fig. 1.Automobile check-points at the Ukrainian-Polish border The length of the route is 2,390 km.Total time of transportation is 36.76hours, service time at the first c/p Ustiluh is 45 min, at the second (Zosin) -40 min, at the third (Dorohusk) -35 min, at the fourth (Yahodyn) -40 min.Duration of loading and unloading operations is 750 min.Total time to perform the route is 51.93 hours.

Fig. 3 .Fig. 4 .
Fig. 3. Distances between c/p, points of departure and destination points in the Microsoft Excel file format 180 that is, we have a TP, unbalanced (open) in terms of cargo transportation volume.First, we optimize the TP represented in a network form using the three methods to reduce a TP, unbalanced in terms of cargo transportation volume, to the balanced form, specifically: introduction of an additional fictitious transportation node, proportional and different.

Fig. 7 .
Fig. 7. Results of COSCT operation in RTN, balanced by the introduction of an additional fictitious transportation node Table3gives optimal cargo transportation volumes, which are achieved along the most economically beneficial routes; the distances of freight delivery between the respective points are the numbers in the upper right corner of each table cell.These routes were obtained using the method of finding the shortest routes along a transportation network[16]; the actual volume of transportation work is 174,680 tkm.This is explained by the fact that cargo with a volume of 20 tons is not actually taken from fictitious point С 1 to UP B 1 .

Fig. 5 .
Fig. 5. Itinerary of the specified closed route

Fig. 9 .
Fig. 9. Results of COSCT operation in RTN, balanced using a difference method

Fig. 10 .
Fig. 10.Result of COSCT operation in the unbalanced RTN

Table 3
Transport table, balanced by the introduction of an additional fictitious transportation node

Table 4
Transportation table, balanced using a proportional method

Table 5
Transportation table, balanced using a difference method

Table 6
Unbalanced transportation table