Railway technology station flow distribution optimization method considering truck value concentration

Abstract

The invention aims to provide a railway technical station flow distribution optimization method considering wagon value concentration, relates to the technical field of railway transportation, and provides a wagon value concentration concept according to characteristics in order to improve the railway freight service quality and improve the transfer efficiency of high-value cargoes in a technical station. Therefore, the operation priority degree of the trucks loaded with different goods in the technical station is measured. According to the method, main factors influencing the value concentration of the trucks are analyzed, the value concentration of the trucks is comprehensively calculated through a weighted rank sum ratio method, the aims of minimizing the staying time of the trucks in a technical station and preferentially dispatching the trucks with the high value concentration are achieved, and flow distribution is conducted on the trucks with the different value concentrations in stages by changing the full-axis constraint of a freight train. A technical station traffic flow continuing mode considering the truck value concentration is formed, the total residence time of high-value-concentration trucks in a stage plan can be further shortened, the transfer operation efficiency of key cargos in the station is effectively improved, and the requirement that high-value-concentration trucks arrive and dispatch in the same group is met.

Description

Technical Field

[0001] The present invention relates to the technical field of railway transportation, and in particular, to a method for optimizing the distribution of railway technical stations considering the value concentration of freight cars. Background Art

[0002] With the development of social economy, the structure of social freight demand is also changing. The transportation demand for high-added-value and high-timeliness goods is gradually increasing. The traditional distribution method of technical stations is difficult to meet the transportation time limit requirements of "double-high" goods. At present, the high-speed railway network has been formed, and the freight capacity of the general-speed railway has been effectively released, and it has the conditions to efficiently transport "double-high" goods. Considering the "double-high" characteristics of goods, studying their reasonable organization process and distribution method at technical stations helps to improve the delivery speed of such goods and further enhance the competitiveness of railway transportation enterprises.

[0003] Currently, most of the research considering both the value of goods and the revenue of transportation enterprises focuses on single factors, such as studying the achievement level of transit time limit, the value of goods loaded on freight cars, etc. There is a lack of comprehensive evaluation of the overall transportation status of high-value goods, and it is difficult to achieve a better connection efficiency of the distribution plan at technical stations on the premise of ensuring that high-value freight cars depart as soon as possible; in addition, due to the complexity of the overall coordination of on-site work at railway technical stations, although existing research shows that the fixed-point departure mode under relaxed conditions has better economy, a flexible operation mode considering the rapid distribution of high-value freight cars has not been implemented in actual work. Summary of the Invention

[0004] In view of the deficiencies of the prior art, the present invention provides the following technical solutions:

[0005] A method for optimizing the distribution of railway technical stations considering the value concentration of freight cars, comprising the following steps:

[0006] S1. Analyze and evaluate the composition of the value concentration of freight cars, construct an evaluation index system for the value concentration of distribution resources at technical stations, analyze the weights of each evaluation index by the entropy weight method, then process the data by the deviation standardization method, and input it into the weighted rank sum ratio evaluation system to obtain the value concentration ranking of the freight cars arriving at the technical station;

[0007] S2. Establish a model, divide the distribution process into a high-value concentration distribution stage, an ordinary value concentration distribution stage, and a distribution plan adjustment stage, and form a distribution model that comprehensively considers the value concentration of freight cars and the total residence time of freight cars by adjusting the constraint conditions at different stages;

[0008] S3. Solve the model, which is solved in three stages:

[0009] The first stage is to solve the train formation and decomposition sequence at the station under the goal of maximizing the connection of vehicle flows and the initial distribution plan without considering the value concentration of freight cars;

[0010] In the second stage, under the maximum flow continuation decomposition and sorting sequence, the vehicle flow allocation is carried out for the trucks with high-value concentration and ordinary-value concentration successively.

[0011] In the third stage, for the trucks with high-value concentration that fail in vehicle flow allocation, the trains with the earliest departure are matched, and at the same time, the composition of the parked cars at the station is analyzed, and the formation content of the trains with lost lines is considered to be modified.

[0012] By designing a maximum neighborhood search algorithm, the first stage and the second stage are solved; combining the characteristics of determining the parked car resources at the station and determining the departure time of the trains with lost lines within the stage plan, referring to the network model solution algorithm in static vehicle flow allocation, an algorithm for adjusting the formation content of the trains with lost lines at the technical station considering the maximum departure of the trucks with high-value concentration is designed to solve the third stage.

[0013] Furthermore, in S1, five indicators including the number of trucks from the formation to the departure group, the time value of the goods, the delay time, the subsequent operations, and the shipper level are selected to construct an evaluation index system for the value concentration of vehicle flow allocation resources at the technical station. The specific steps for evaluating the value concentration of trucks are as follows:

[0014] S1.1, Data standardization. Except for the shipper level and the subsequent operation situation, the deviation standardization method is used for each evaluation index, and the following formula is applied:

[0015]

[0016] The data is transformed to the range of [0, 1], and then the interval of [0, 1] is equally divided into 10 parts. The data falling into each interval is divided into 10 grades, corresponding to the numbers 1 - 10 from low to high. For the sample data, it is transformed into a matrix with m samples, and each sample has n indicators.

[0017] S1.2, Ranking. Construct an evaluation matrix A of the evaluation object ij , where the values of the shipper level and the subsequent operation situation are based on the actual data, and the data of the time value of the goods and the delay time adopt the y after data standardization in S1.1 ij , rank each index in the evaluation matrix, and sort according to the type of the index. Among them, the benefit-type indexes are ranked from small to large, and the cost-type indexes are ranked from large to small. For the same index data, the average value is taken, and the obtained rank matrix is denoted as R ij ;

[0018] S1.3, Weight design. The entropy weight method is used to determine the weights of each index, denoted as [λ1, λ2, …, λ n ;

[0019] S1.4, Calculate the weighted rank sum ratio WRSR of the i-th evaluation object, denoted as δ WRSRi ;

[0020] S1.5, grading and sorting, sort the evaluation objects according to δ WRSRi Sort the evaluation objects.

[0021] Furthermore, the establishment of the allocation model in S2 includes the following steps:

[0022] S2.1, model assumptions: Assume that the technical station adopts the operation mode of single pushing and single humping, the trains arrive on time in the stage plan, and sudden situations and uncertainties during the operation of the technical stations within the region are not considered; Assume that the equipment and staff numbers of the station meet the operation requirements and can perform arrival and departure technical operations on arriving and departing trains in a timely manner; Do not consider abnormal and maintenance states such as locomotive servicing and faults, and default that the locomotive operates continuously during working hours without pauses in the middle; To simplify the modeling difficulty, assume that in the same arriving train at this technical station, the value concentration of freight cars with the same car flow group number is the same;

[0023] S2.2, establish the objective function: Split the allocation process of the technical station into two stages, and in both stages, the shortest stay time of freight cars in the station is used as the objective function:

[0024]

[0025] After preferentially allocating the high-value concentration freight cars in the arriving car flow, then allocate the remaining car flow according to the same objective function, and finally obtain the allocation plan with the shortest total stay time of freight cars within the current stage plan;

[0026] S2.3, model constraints: In the technical station allocation model, the main constraints considered include disintegration operation constraints, formation operation constraints, car flow connection constraints, and full-axle constraints for departing trains:

[0027] Train arrival and disintegration operation constraints, the constraints are expressed as follows:

[0028] T JTKS (Y h )≥T DD (Y h )+T DDJS h = 1, 2,..., m

[0029] T JTKS (Y h )≥T JTKS (Y h-1 )+T JT h = 2,..., m

[0030] Respectively represent that the actual disintegration time of the train with disintegration sequence Y h is not earlier than the earliest disintegration time of this train and the actual disintegration moment of the train with disintegration sequence Y h is not earlier than the arrival sequence Y h-1 That is, the disintegration end moment of the previous train;

[0031] The constraints for train formation operations are as follows:

[0032] T ZWJSBZ (Z l ) ≥ T CF (Z l ) - T CFJS l = 1, 2,..., n

[0033] T KSBZ (Z l ) ≥ T KSBZ (Z l-1 ) + T BZ l = 1, 2,..., n

[0034] They respectively represent the latest completion time T of the departure train with formation order Z l not earlier than its actual completion time of formation and the start time of formation of the departure train with formation order Z ZWJSBZ not earlier than the actual completion time of formation of the previous departure train; l The car flow connection constraint:

[0035] The car flow connection constraint:

[0036]

[0037] They respectively represent that only when the start time of formation of departure train j is later than the completion time of disintegration of arrival train i can arrival train i provide car flow for departure train j, and the number of wagons with car flow group number k provided by arrival train i for departure train j cannot exceed the number of wagons with the corresponding car flow group number in this arrival train; The constraint of associated formation content is:

[0038]

[0039] The full - axle constraint of departure trains:

[0040]

[0041] They respectively represent that there are requirements for the length of any departure train, which need to be restricted within [M min , M max , and there are requirements for the weight of any departure train, which need to be restricted within [W min , W max .

[0042] Furthermore, in S3, the maximum - neighborhood search algorithm includes the following steps:

[0043] a. Calculate the ideal arrival - departure connection car - flow matrix C without considering equipment occupancy conflicts and the current train - disintegration sequence LXThe actual arrival and departure connection vehicle flow matrix C considering equipment occupancy conflicts and train breakup sequences SJ , using the ideal arrival and departure connection vehicle flow matrix C LX Subtract the actual arrival and departure connection vehicle flow matrix C SJ , to obtain the vehicle flows that do not meet the vehicle flow connection constraint conditions under the current breakup sequence, that is, the vehicle flows corresponding to the positions greater than 0 in the matrix;

[0044] b. Find all positions greater than 0 in the arrival and departure vehicle flow connection difference matrix to generate a candidate set U of breakup adjustment points. Let u1 be the first adjustment point in the candidate set, and the total number of adjustment points be N TZ , each time select a point from the alternative set as the point to be adjusted to construct a neighborhood, and obtain the advance amount of the breakup time corresponding to the adjustment of the breakup sequence of the arrival train corresponding to this point, and the delay amount of the formation time corresponding to the adjustment of the formation sequence of the departure train corresponding to this point under the current breakup sequence;

[0045] At the same time, calculate the connection time difference between the arrival train and the departure train corresponding to the breakup adjustment point, and find all appropriate train breakup sequence adjustment amounts in the time change matrix that satisfy the change value being exactly greater than or equal to the absolute value of the train connection time difference at this point, so as to generate neighbors of the current solution; adjust the breakup sequence of the arrival and departure trains to which this part of the vehicle flow belongs. Using the arrival and departure train sequence of the stage plan as the initial breakup sequence, combined with time constraints such as technical operation time standards, calculate the actual arrival and departure vehicle flow connection matrix under the current breakup sequence of this technical station;

[0046] c. Solve the vehicle flow allocation results of all neighborhood solutions in turn until all neighborhood solutions of the adjustment points in N TZ are searched to obtain the optimal vehicle flow allocation plan T Best ; using the arrival and departure train sequence of the stage plan as the initial breakup sequence, calculate the arrival and departure train information of the stage plan under the current breakup sequence, and perform vehicle flow allocation with the current train arrival and departure sequence of the technical station as the breakup sequence of the trains in the vehicle flow allocation plan to obtain the initial solution T. Let T Best = T, T now = T, T now < T Best Take the train arrival and departure sequence as the current optimal breakup sequence.

[0047] Furthermore, in S3, the algorithm for adjusting the formation content of the trains losing lines at the technical station where high-value concentration freight cars depart to the maximum extent includes the following steps:

[0048] a. According to the car groups allocated to the trains losing lines, design an adjustment strategy for the car groups with failed vehicle flow allocation:

[0049] b. Regarding the problem of making the high-value concentration freight cars in station storage depart as soon as possible by adjusting the formation content of the train with lost lines as a static vehicle allocation problem, a vehicle allocation network model is established, and the maximum departure plan of the train with lost lines for adjusting the vehicle allocation content is obtained by solving the static vehicle allocation problem.

[0050] Compared with the prior art, the technical solution of the present application has the following beneficial effects:

[0051] When performing vehicle allocation at a technical station in the present invention, the arriving vehicle flow is classified according to the value concentration according to the fixed-point assembly mode. The decomposition vehicle allocation method of giving priority to the vehicle groups with high value concentration in vehicle allocation and the ordinary value concentration vehicle groups second can further reduce the total residence time of the high-value concentration freight cars within the stage plan, effectively improving the transfer operation efficiency of key goods at the station; in the multi-stage vehicle allocation plan considering the value concentration of freight cars finally formed, the high-value concentration vehicle groups are concentrated in the earliest departure trains that meet the connection requirements, and through the design of the algorithm, it is avoided that the goods with simultaneous arrival and departure requirements are split by the transfer station, meeting the requirements of the same group arrival and departure of the high-value concentration vehicle groups; compared with the existing heuristic algorithms, the neighborhood search algorithm based on the connection difference of arrival and departure vehicle flows has good applicability to the vehicle allocation optimization problem considering the value concentration of freight cars in this article, and can obtain the vehicle allocation optimization solution in a short time; the designed value matching search algorithm VMSA can obtain a specific plan in a short time when solving the combination of alternative vehicle flow directions of the train with lost lines, and because it is solved according to the optimal solution sequence under the maximum vehicle flow connection target, this plan can further improve the formation and departure efficiency of the high-value concentration vehicle groups while ensuring the superiority of the current vehicle allocation plan at the technical station. Description of the Drawings

[0052] Figure 1 It is a schematic diagram for analyzing the situation of freight car delay;

[0053] Figure 2 It is a schematic diagram of the dynamic vehicle allocation process at a railway technical station considering the value concentration of freight cars;

[0054] Figure 3 It is a schematic diagram of the network model for maximizing the departure adjustment of high-value concentration freight cars;

[0055] Figure 4 It is a schematic diagram of the location of the technical station;

[0056] Figure 5 It is a schematic diagram of the value concentration information of the arriving vehicle groups;

[0057] Figure 6 It is a schematic diagram of the vehicle allocation plan considering the value concentration of freight cars after adjustment;

[0058] Figure 7 It is a comparison diagram of two vehicle allocation plans;

[0059] Figure 8Schematic diagram of the allocation plan without considering the value concentration of freight cars;

[0060] Figure 9 Schematic diagram of the allocation plan considering the value concentration of freight cars. Specific implementation manner

[0061] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0062] Embodiment 1:

[0063] An optimization method for the allocation of railway technical stations considering the value concentration of freight cars includes the following steps:

[0064] 1. Analysis of the composition of the value concentration of freight cars at technical stations. As the smallest unit of railway transfer technical operations, a freight car can be regarded as a container for loading goods. During the transportation process, the value of the goods fluctuates to a certain extent with the progress of transportation. This process is formally similar to the change of liquid concentration over time. Therefore, it is described as "value concentration" and five types of indicators are selected, namely the number of freight cars in a group from arrival to departure, the time value of goods, the delay time, subsequent operations, and the shipper level, to construct an evaluation index system for the value concentration of allocation resources at technical stations:

[0065] Number of freight cars in a group from arrival to departure: Convert the group arrival situation of freight cars in the arriving train and the number of freight cars in the car group into a consideration factor for the priority level of car group operations in the allocation plan, that is, the composition factor of the value concentration defined in this application;

[0066] Time value of goods: Divide the composition of the goods value into three parts: the occupation of the shipper's funds by the goods, the loss caused by the extension of the transportation time, and the depreciation of the goods. Thus, the time value function of the goods is:

[0067]

[0068] Where P s is the value of a unit container of a certain category of goods, γ is the annual capital return rate of the shipper, ρ is the annual loss ratio of the goods, k is the annual depreciation rate of the goods; ω s is used to represent the total value of each standard freight car fully loaded with the same type of goods in actual operations:

[0069] Delay time: Based on the historical statistical data of the road network, combined with the data of trains departing from the originating station, the total running mileage of the section, the train running speed, and the technical operation time standard, the earliest arrival time T1 of the freight car at this station under ideal conditions, the latest arrival time T3, and the actual arrival time T2 of the train are determined. There are 4 cases of the numerical relationship among the three as shown in Figure 1 the following:

[0070] In the actual production operation process, due to the uncertainty of the operation time at each stage, deviations will inevitably occur. In the flow allocation problem studied in this paper, the key is the identification of delayed car groups. Therefore, cases 2 and 4 will be mainly discussed in the following (cases 1 and 3 are not discussed in this application). Since the current transportation implements the statistical standard of arrival time limit in days (d), although it brings certain convenience to the actual operation, it is difficult to meet the freight transportation demand under the current market competition environment. Therefore, referring to the calculation mode in minutes in the stage plan of the technical station, this application uses minutes as the unit for measuring the arrival time limit to calculate the delay time of the vehicle flow; in actual work, since the situation where the actual arrival time T2 of the goods at the current station is between the earliest ideal arrival time T1 and the latest ideal arrival time T3 is the most common, this type of situation will be analyzed first; since the goods delay time is a direct reflection of the current operation status of the freight car, for the technical station, it needs to bear the "burden" brought by the delay of the previous vehicle flow, and at the same time, the possible delay risk of the subsequent vehicle flow route should also be considered. Therefore, the actual arrival time T2 of the goods should be as close as possible to the earliest ideal arrival time T1. Thus, the freight car delay situation can be calculated by the following formula:

[0071] T YW = T2 - T1

[0072] Subsequent operation situation: During the operation of the freight train, the actual running time usually accounts for a relatively small proportion, about 1 / 3 of the total transportation time. Referring to the transportation industry indicators of the National Bureau of Statistics, the average transfer stay time of railway freight cars is 4.7h, and the travel speed of freight trains is 35.6km / h; calculated based on the average freight transport distance of 720.99km in 2022, when the number of transfers of the freight train is 1, 2, and 3 respectively, the proportion of the transfer time of the freight car is 23.2%, 46.4%, and 69.6% respectively. Obviously, the more transfers there are, the longer the transfer stay time; moreover, when the transportation plan of the goods is determined, its vehicle flow route and transfer station are relatively fixed. The freight car delay situation describes the transportation status of the freight car arriving at this station and before, while the subsequent operation requirements of the freight car are the advance evaluation of the operation situation after departing from this station;

[0073] Shipper level: By using three important customer behavior indicators, a "customer classification and value evaluation model based on RFM" is constructed, that is, the customer value is judged by the recent purchase time R (Recency), purchase frequency F (Frequency), and total purchase amount M (Monetary Value). Then, the K-means clustering method is used to divide railway freight customers into seven categories, namely important retention customers, general value customers, valueless customers, general retention customers, general retention customers, important retention customers, and important development customers. Among them, the shipper level is an efficiency-type indicator for the evaluation of the value concentration of freight cars. Therefore, the values are taken in descending order of importance, denoted as [ω7, ω6, ω5, ω4, ω3, ω2, ω1].

[0074] 2. Evaluation of the value concentration of freight cars based on the weighted rank sum ratio method. The entropy weight method is used to analyze the weights of each evaluation index, and then the data is processed by the deviation standardization method. The value concentration ranking of the arriving freight cars at this technical station is obtained by inputting into the weighted rank sum ratio evaluation system. The specific evaluation steps are as follows:

[0075] 2.1 Data standardization: The dimensionalities of the constituent factors of the value concentration index of the arriving freight cars at the technical station are different. Before evaluation, data standardization is required. This matrix has m samples, and each sample has n indicators. Except for the shipper level and subsequent operation conditions, the deviation standardization method can be used for each evaluation index. Applying the formula

[0076]

[0077] The data is transformed to between [0, 1], and then the [0, 1] interval is equally divided into 10 parts. The data falling into each interval is divided into 10 grades, corresponding to the numbers 1-10 from low to high. For the sample data, it is transformed into a matrix.

[0078] 2.2 Ranking: Construct the evaluation matrix A of the evaluation object ij , where the values of the shipper level and subsequent operation conditions are based on the actual data, and the data of the time value of goods and delay time adopt the y after data standardization in Step 1 ij ; Rank each index in the evaluation matrix. Sort according to the type of index. Among them, the efficiency-type indicators are ranked from small to large, and the cost-type indicators are ranked from large to small. For the same index data, the average value is taken. The obtained rank matrix is denoted as R ij ;

[0079] 2.3 Weight design: Considering the operation characteristics of railway technical stations, the entropy weight method is used to determine the weights of each index, denoted as [λ1, λ2,…, λ n ;

[0080] 2.4 Calculate the weighted rank sum ratio (WRSR) of the i-th evaluation object, denoted as δWRSRi , the calculation formula is:

[0081]

[0082] Sort in 2.5-point intervals, and sort the evaluation objects according to the estimated value WRSR;

[0083] The optimization goal of the flow distribution in this application is to preferentially group the individuals with higher value concentration among the freight cars with the same destination into the first-departing trains. During the process of calculating and comparing the value concentration, it is necessary to count the value concentration of each arriving train. Since the grouped arrival requirements of the freight cars themselves are difficult to quantify, to simplify the analysis process, it is assumed that the high-value-concentration freight cars with the same traffic flow group number in the same arriving train all have grouped arrival and departure requirements. Therefore, only the other four evaluation indicators are quantified during the rank sum ratio evaluation;

[0084] 3. Model establishment. The objective function of the current technical station flow distribution model can be divided into: a time type with the minimum total detention time of freight cars as the goal, a quantity type with the maximum number of total departing trains from the station as the goal, a revenue type with the maximum revenue of the departing trains from the station, and a comprehensive type that comprehensively considers multiple goals; This application mainly considers the optimization problem of the traffic flow connection of high-value-concentration freight cars. Its core is to shorten the detention time of high-value-concentration freight cars at the station while improving the in-transit operation efficiency of transfer freight cars. Therefore, the objective function adopts the minimum total detention time of freight cars; Thus, the flow distribution process is divided into a high-value-concentration flow distribution stage, a normal-value-concentration flow distribution stage, and a flow distribution plan adjustment stage. By adjusting the constraint conditions in different stages, a flow distribution model that comprehensively considers the value concentration of freight cars and the total detention time of freight cars is formed;

[0085] 3.1 Model assumptions. It is assumed that the technical station adopts a single-push and single-slide operation method, the trains arrive on time in the stage plan, and the sudden situations and uncertain factors during the operation process of the technical stations in the region are not considered; It is assumed that the equipment and staff numbers of the station meet the operation requirements and can perform arrival and departure technologies on the arriving trains and departing trains in a timely manner; The abnormal and maintenance states such as locomotive preparation and faults are not considered, and it is defaulted that the locomotive operates continuously during working hours without pauses in the middle; To simplify the modeling difficulty, it is assumed that the value concentration of the freight cars with the same traffic flow group number in the same arriving train at the technical station is the same;

[0086] 3.2 Objective function. Considering the value concentration of the arriving freight cars at the technical station, the flow distribution process of the technical station is split into two stages. Both stages use the shortest detention time of the freight cars at the station as the objective function. After preferentially distributing the high-value-concentration freight cars in the arriving traffic flow, the remaining traffic flow is distributed according to the same objective function, and finally a flow distribution plan with the shortest total detention time of the freight cars within the current stage plan is obtained. Therefore, the objective function is:

[0087]

[0088] 3.3 Constraints of the model. In the technical station flow allocation model, the main constraints considered include four types: disintegration operation constraints, formation operation constraints, car flow connection constraints, and full axle constraints for departure trains, as well as other logical constraints representing the operation process:

[0089] Train arrival and disintegration operation constraints:

[0090] After the arriving train arrives at the technical station, it needs to go through the arrival technical operation before it can become a train to be disintegrated. Its earliest disintegration time is the end time of the train inspection. The actual disintegration time before and after adjustment shall not be earlier than the train inspection technical time of this train. This constraint is expressed as follows:

[0091] a. The actual disintegration time of the train with disintegration sequence Y h is not earlier than the earliest disintegration time of this train

[0092] T JTKS (Y h )≥T DD (Y h )+T DDJS h = 1, 2,..., m

[0093] b. The actual disintegration time of the train with disintegration sequence Y h is not earlier than the disintegration end time of the previous train in arrival sequence Y h-1 i.e., the disintegration end time of the previous train

[0094] T JTKS (Y h )≥T JTKS (Y h-1 )+T JT h = 2,..., m;

[0095] Train formation operation constraints:

[0096] a. The latest formation end time T l of the departure train with formation sequence Z ZWJSBZ is not earlier than its actual formation end time

[0097] T ZWJSBZ (Z l )≥T CF (Z l )-T CFJS l = 1, 2,..., n

[0098] b. The start formation time of the departure train with formation sequence Z l is not earlier than the actual formation end time of the previous departure train

[0099] T KSBZ (Z l )≥T KSBZ(Z l-1 ) + T BZ where \(l = 1, 2, \ldots, n\);

[0100] Car flow connection constraint:

[0101] a. Only when the formation start time of departure train \(j\) is later than the disintegration end time of arrival train \(i\) can arrival train \(i\) provide car flow for departure train \(j\).

[0102]

[0103] b. The number of wagons with car flow group number \(k\) provided by arrival train \(i\) for departure train \(j\) cannot exceed the number of wagons with the corresponding car flow group number in this arrival train.

[0104]

[0105] c. Formation content association constraint

[0106]

[0107] Full - axle constraint for departure train:

[0108] Since wagons with high - value concentration are not considered to be incorporated into uncoupling and shunting trains, departure train \(j\) needs to be full - weight or full - length to depart.

[0109] a. It represents a requirement for the length of any departure train, which needs to be restricted within \([M min , M max

[0110]

[0111] b. It represents a requirement for the weight of any departure train, which needs to be restricted within \([W min , W max

[0112]

[0113] It should be noted that: In the car flow distribution stage of wagons with high - value concentration, since the number of wagons with high - value concentration is small, it is difficult to form a "fully high - value - concentration full - axle train" during the car flow distribution process. Therefore, in this stage, the car flow distribution model needs to be modified, and the full - axle constraint condition is not considered. At the same time, in the car flow distribution stage of wagons with ordinary - value concentration, since it is necessary to ensure that all wagons with high - value concentration depart, the current state of the departure trains incorporated with wagons with high - value concentration needs to be modified, and the interval of the full - axle constraint condition needs to be corrected according to the solution results of the first stage;

[0114] The model parameters and symbols involved in the above model establishment are shown in the following table:

[0115] Table 1 Explanation of parameters and symbols​​

[0116]

[0117] 4. Model solution: The railway technical station vehicle allocation model considering the value concentration of freight cars is a non - linear mixed - integer programming model, and its solution belongs to the NP - Hard problem. Existing research mostly uses heuristic algorithms for solution. However, heuristic algorithms themselves have problems such as unstable solution efficiency and being greatly affected by the initial solution. To reduce the difficulty of solution, this application is solved in three stages:

[0118] The first stage: Solve the station sorting sequence under the goal of maximizing the vehicle flow connection and the initial vehicle allocation plan without considering the value concentration of freight cars;

[0119] The second stage: Allocate vehicles for high - value - concentration and ordinary - value - concentration freight cars successively under the sorting sequence of maximizing the vehicle flow connection;

[0120] The third stage: Match the high - value - concentration freight cars with failed vehicle allocation to the earliest departure trains as much as possible, and at the same time analyze the composition of the in - station vehicles and consider modifying the formation content of the missed trains;

[0121] In the first and second stages, since the proportion of high - value - concentration car groups in the arriving vehicle flow at the station is usually not high, decision - makers can preferentially allocate these car groups. Limited by the urgency of formulating the stage plan and the uncertainty during the implementation of the stage plan, an efficient and easy - to - use vehicle allocation and key vehicle flow adjustment algorithm is required. Therefore, a maximization neighborhood search algorithm is designed for the solution of the first and second stages; in the third stage, when the stage plan includes high - value - concentration freight cars and there are missed trains at the technical station, it is necessary to consider adjusting the formation content of the missed trains to ensure the maximum departure of high - value - concentration freight cars. For this reason, combined with the determination of in - station vehicle resources and the determination of the departure time of missed trains in the stage plan, referring to the network model solution algorithm in static vehicle allocation, an algorithm for adjusting the formation content of missed trains at the technical station considering the maximum departure of high - value - concentration freight cars is designed;

[0122] Neighborhood search algorithm for maximizing the vehicle flow connection at the technical station: For the solution of the optimal sorting sequence, a local neighborhood search algorithm can be considered. By calculating the difference in the vehicle flow connection from departure, the car groups that do not meet the vehicle flow connection conditions are screened out. A neighborhood is established for the sorting sequence of the trains where these car groups are located, and the optimal solution that meets the vehicle flow connection is searched, so as to optimize the sorting sequence. Specifically as follows:

[0123] Step1: Calculate the ideal arrival - departure connection vehicle flow matrix C LX without considering equipment occupation conflicts and the current train sorting sequence and the actual arrival - departure connection vehicle flow matrix C SJ considering equipment occupation conflicts and the train sorting sequence respectively, and use the ideal arrival - departure connection vehicle flow matrix CLX Subtract the actual arrival and departure connection vehicle flow matrix C SJ , to obtain the vehicle flows that do not meet the vehicle flow connection constraint conditions under the current classification and marshalling sequence, that is, the vehicle flows corresponding to the positions greater than 0 in the matrix;

[0124] Step2 Find all positions greater than 0 in the arrival and departure vehicle flow connection difference matrix, generate the candidate set U of classification and marshalling adjustment points, let u1 be the first adjustment point in the candidate set, and the total number of adjustment points be N TZ ; Each time, select a point from the alternative set as the point to be adjusted to construct a neighborhood. Under the current classification and marshalling sequence, obtain the advance amount of the disintegration time corresponding to the adjustment of the disintegration sequence of the arriving train corresponding to this point (the result is a column vector with n rows and l columns), and the delay amount of the marshalling time corresponding to the adjustment of the marshalling sequence of the departing train corresponding to this point (the result is a row vector with l rows and m columns); At the same time, calculate the connection time difference between the arriving train and the departing train corresponding to the classification and marshalling adjustment point (this value is negative), and find all appropriate train classification and marshalling sequence adjustment amounts in the time change matrix that satisfy that the change value is exactly greater than or equal to the absolute value of the train connection time difference at this point, so as to generate the neighbors of the current solution. Adjust the classification and marshalling sequence of the arrival and departure trains to which this part of the vehicle flow belongs. Using the arrival and departure train sequence in the stage plan as the initial classification and marshalling sequence, combined with time constraints such as technical operation time standards, calculate the actual arrival and departure vehicle flow connection matrix under the current classification and marshalling sequence of this technical station;

[0125] Step3 Solve the vehicle flow distribution results of all neighborhood solutions in turn until all neighborhood solutions of the N TZ adjustment points are searched to obtain the optimal vehicle flow distribution plan T Best ; Using the arrival and departure train sequence in the stage plan as the initial classification and marshalling sequence, calculate the arrival and departure train information in the stage plan under the current classification and marshalling sequence, and perform vehicle flow distribution with the current train arrival and departure sequence of the technical station as the classification and marshalling sequence of the trains in the vehicle flow distribution plan to obtain the initial solution T, let T Best =T, T now =T, T now <T Best Take the train arrival and departure sequence as the current optimal classification and marshalling sequence;

[0126] Thus, the solution task of the first stage is completed, and the classification and marshalling sequence of the station under the maximum vehicle flow connection target and the initial vehicle flow distribution plan without considering the value concentration of freight cars are obtained; Among them, the train classification and marshalling sequence under the maximum vehicle flow connection target is used as the known condition to input the model for the vehicle flow distribution plan of high-value concentration freight cars in the second stage, and the initial vehicle flow distribution plan without considering the value concentration of freight cars is used as a reference to compare the optimization effects of subsequent vehicle flow distribution plans;

[0127] Technical station vehicle allocation algorithm considering the value concentration of freight cars: This application takes into account the flexible formation number of departure trains and conducts vehicle allocation based on the fixed-point assembly mode. On the basis of solving the maximum car flow connection and disassembly plan that meets the current technical station, the vehicle allocation process is decomposed, and freight cars with high value concentration are selected as vehicle allocation resources and input into the model for priority vehicle allocation. After obtaining the vehicle allocation plan with high value concentration, the model constraint conditions are modified in combination with the formation requirements of departure trains in the station stage plan, and the remaining arriving freight cars are input into the model again as vehicle allocation resources to achieve the secondary optimization of the vehicle allocation plan. The specific algorithm is as follows:

[0128] Step1 Read the value concentration information of freight cars in the arrival trains of the current stage plan, evaluate it using the weighted rank sum ratio evaluation method, and obtain the relative relationship of the value concentration of different car groups; technical station staff set the value concentration threshold according to the station car flow organization situation to determine high-value concentration car groups and ordinary-value concentration car groups;

[0129] Step2 Mark the high-value concentration car groups in the initial vehicle allocation plan without considering the value concentration of freight cars, and check whether each high-value concentration car group is assigned to the train with the earliest departure in the corresponding car flow group number. If this requirement has been met, the target plan is obtained; if not, go to Step3;

[0130] Step3 Take the train disassembly sequence under the maximum car flow connection target as the known condition, input the high-value concentration freight cars as vehicle allocation resources into the vehicle allocation model, and solve the result of the separate vehicle allocation of high-value concentration freight cars in the first stage;

[0131] Step4 Input the ordinary-value concentration freight cars as vehicle allocation resources into the vehicle allocation model, and solve the vehicle allocation result of ordinary-value concentration freight cars in the second stage by modifying the full-load constraint of departure trains;

[0132] Step5 Combine the vehicle allocation results of the two stages to obtain the dynamic vehicle allocation result considering the value concentration of freight cars; The dynamic vehicle allocation process of the technical station considering the value concentration of freight cars is as Figure 2 shown;

[0133] Maximum departure adjustment algorithm for high-value concentration freight cars: In the vehicle allocation stage of high-value concentration freight cars, since each departure train is in a state of waiting for vehicle allocation, for the combination of freight cars and departure trains that meet the car flow connection conditions, high-value concentration freight cars will be automatically assigned to the trains with the earliest departure during the vehicle allocation process at the technical station. For these trains, there may be a situation where due to insufficient car flow connection resources, the trains are short of axles and cannot depart, and finally form lost-line trains. Lost-line trains will cause waste of technical station resources and also result in certain benefit losses for railway transportation enterprises; For this part of car groups, there are usually 3 treatment methods:

[0134] a. If the train with disrupted line meets the connection time requirement of high-value freight car flow, and there is a combination of in-station cars that can form a full-axle train with the stranded high-value freight cars, then consider adjusting the formation direction of the train with disrupted line, and reallocate the flow to form a full-axle train, so as to achieve the purpose of reducing the total in-station residence time of freight cars while ensuring the departure efficiency of high-value concentration freight cars;

[0135] b. According to the train departure plan, find the departure train that is closest to the departure time of the train with disrupted line and whose formation content includes the high-value concentration car body of the train with disrupted line, judge whether the conversion length and traction weight of this train meet the adjustment requirements, and select appropriate car bodies for replacement;

[0136] c. Set the priority of this high-value concentration car body to the maximum, and give priority to the formation operation in the next stage plan.

[0137] Therefore, according to the car bodies allocated to the train with disrupted line, design the adjustment strategy for the car bodies with failed allocation:

[0138] Step1 Check the composition of the car flow group numbers of the freight cars with failed allocation and the formation direction of the departure train with failed formation. If there are ordinary value concentration freight cars with the same destination in the subsequent departure trains, give priority to replacing the ordinary value concentration car bodies with high-value concentration car bodies, and transfer to Step2; if there are no car bodies meeting the conditions in the subsequent departure trains, and the train with disrupted line meets the connection time requirement of high-value freight car flow, then transfer to Step3. If the train with disrupted line does not meet the connection time requirement of high-value freight car flow, then transfer to the last step;

[0139] Step2 For the car bodies with the same car flow group number in the subsequent departure trains, select the replacement car bodies of the high-value concentration car bodies according to the principle of giving priority to "longer in-station residence time and lower cargo value concentration", and update the allocation plan;

[0140] Step3 Classify the freight cars with failed allocation according to the car flow group numbers, and the car flow in the same direction as the high-value concentration freight cars with failed departure is the available formation resource, and enter it into the formation destination resource library of the new departure train;

[0141] Step4 According to all possible formation direction combinations of the train with disrupted line, convert the in-station car resources in the current plan into the allocation resources of the arriving trains and input them into the allocation model, and relax the full-axle limit of the train with disrupted line, and solve the optimal adjustment plan for the train with disrupted line;

[0142] Step5 Output the optimal allocation plan after adjustment.

[0143] Lost-line train adjustment network algorithm: The problem of making the high-value concentrated freight cars in the station storage depart as soon as possible by adjusting the formation content of the lost-line train can be analogously regarded as a static vehicle allocation problem and solved through a network model. For the station storage vehicle resources in the vehicle allocation plan obtained by solving the above algorithm, a special maximum flow algorithm is designed to solve the static vehicle allocation problem, and the maximum departure plan of the lost-line train for adjusting the vehicle allocation content is obtained.

[0144] Vehicle allocation network model. Here, the goal is to match the high-value concentrated freight cars in the station storage with car groups that meet the vehicle flow connection constraints, and at the same time change the formation content of the lost-line train so that these car groups form the to-be-departed vehicle flow and complete the vehicle flow connection according to the original scheduled formation and departure time of the lost-line train. Since the vehicle flow of the arriving trains often has more than one formation destination, and the formation content of the departing trains is often not single, when converting to a network model, a source and a sink are designed, denoted as S and T respectively. Among them, the source S represents the total number of vehicles arriving in this stage. In the actual plan, it is the station storage vehicle resources in the vehicle allocation plan considering the freight car value concentration. The sink T represents the total number of departing vehicles. In the actual plan, it is the freight car combination of the departing trains with the formation content changed.

[0145] Classify the car groups of the arriving trains and departing trains according to the vehicle flow group number. The left column represents the departure points d of each arriving train classified by the vehicle flow destination i , and the right column represents the receiving points f classified by the possible vehicle flow destinations of the lost-line train formation i . Connect the two columns of nodes with arcs. The capacity of the arc is the number of vehicles contained in the d i car group. The arriving and departing trains connected by arcs represent combinations that meet the connection constraints, and vice versa are vehicle flow combinations that do not meet the connection conditions. The simplified network model is obtained through the above method.

[0146] This network model is for targeted vehicle allocation of high-value concentrated car groups. To meet specific requirements, the following definitions are made:

[0147] a. Existing vehicle allocation resources: Refer to the station storage vehicle flow resources that meet the vehicle flow connection requirements of high-value concentrated car groups within the current stage plan.

[0148] b. Callable vehicle allocation resources: Refer to the vehicle flow resources that have been incorporated into the departing trains in the existing vehicle allocation plan and, according to the assembly requirements under relaxed conditions, do not affect the on-time departure of the original departing trains after being called.

[0149] c. Spare vehicle allocation resources: Refer to all the empty car resources that meet the coupling requirements in this technical station, which are used to ensure that the detained high-value concentrated freight cars depart as soon as possible.

[0150] d. Vehicle flow connection arc: Refer to the arc connecting nodes in the vehicle allocation network model, which has three attributes, namely node d i and node f iare connected, representing d i The car sets included in the train arriving at point meet f i the car flow connection requirements of the train departing from point; the initial flow on the arc is 0; the capacity on the arc is the number of freight cars of the corresponding distribution resource;

[0151] Due to the large number of technical arrival and departure car flows, when matching car sets for the to-be-marshaled car sets with high value concentration, car sets should be preferentially selected from the existing distribution resources, and then consider selecting from the "callable distribution resources"; when reconstructing the combination of the car flow directions of the train with lost lines, the possible car flow destinations of the train with lost lines should also be set in combination with the connection directions of the downstream technical stations; therefore, rules need to be designed to determine the node arrangement of the distribution network graph and reduce the number of unnecessary arcs. This application combines the value concentration and the car flow group number to design a distribution scheme network solution algorithm, and its network graph construction rules are as follows:

[0152] Rule 1: On the left side of the network graph is the source point T, and the first adjacent column is the input point d i , representing the distribution resources of this technical station, including: existing distribution resources, callable distribution resources; there is a one-way connection arc between the source point T and the node d of the input point column i , representing the car flow connection resources provided by this technical station;

[0153] Rule 2: The input point d i column is numbered {1, 2, 3,..., n} in the order from top to bottom, where the existing distribution resources are in the front and the callable distribution resources are in the back; within different types of distribution resources, their order is specified as: the ones with the same car flow group number as the high-value concentration freight car sets are placed at the top, and the car sets with the remaining car flow group numbers are sorted in layers, and the car sets with different car flow group numbers within the same car flow direction combination are arranged in descending order of value concentration; for example, for a certain technical station C, after considering the distribution of the freight car value concentration, there is a high-value concentration car set b1 in the station inventory, and its car flow group number is 1. Combining the regional road network distribution and the car flow group number setting, the possible car flow group number combinations are as follows: {[1], [1, 5], [1, 6], [1, 5, 6], [1, 8], [1, 9], [1, 8, 9]}. According to the arrangement method set in Rule 2, the order of the car sets represented by the input column d i is: {[1], [5], [6], [8], [9]}, corresponding to the nodes d i to d5 respectively. When the car sets with the same car flow group number come from n trains arriving at the station, the corresponding point d s is split into n nodes, and the nodes are arranged according to the disintegration order of the corresponding arriving trains;

[0154] Rule 3: Output column f jThe arrangement rule is the same as that of the input column. The difference is that since the output column is used to solve the possible departure combinations of the original lost-line columns after reorganization, the points of the output column do not need to be split;

[0155] All edges in the Rule 4 network diagram are single-direction edges;

[0156] The edge capacity between the Rule 5 output column and the sink T is set to the maximum number of cars allowed in this section;

[0157] The node types and arrangement orders of the input column and the output column are shown in the following table:

[0158] Table 2 Node Types and Arrangement Orders of the Input Column

[0159]

[0160] Table 3 Node Types and Arrangement Orders of the Output Column

[0161]

[0162] Combining the above rules and the car flow connection matrix under the optimal solution compilation order in the neighborhood search algorithm, screening out the points that meet the relevant rules and requirements and forming the connections of different point columns, a maximum departure adjustment network model for high-value concentration freight cars can be constructed, as Figure 3 shown;

[0163] Solution algorithm: For the maximum departure demand of high-value concentration freight cars, all car groups that may be grouped with the detained high-value concentration freight cars are screened out by the construction rules adopted in the construction of the network model. It is also necessary to solve the problem of preferentially matching the most suitable car group to the high-value concentration car group. Combining the flow allocation requirements of the high-value concentration car group, a special maximum flow solution algorithm - the Value Matching Search Algorithm (VMSA) is designed. In the link of solving the flow augmentation chain, car groups with a higher degree of fit with this car group are preferentially selected for flow allocation.

[0164] When constructing the flow augmentation chain, the node with the smallest label in the previous node group is preferentially selected as the alternative node. If there is an available flow augmentation space in the connection arc between the alternative node and the starting point, that is, the flow is less than the capacity, then this node is used as a node in the flow augmentation chain. According to this rule until reaching the sink, it is considered that a flow augmentation chain has been found; traverse all possible edges in the network diagram using Pycharm in the above manner, and finally solve the maximum flow network distribution diagram;

[0165] After obtaining the maximum - flow network, among the connecting arcs between the output flow and the sink, search for possible combination patterns according to the possible combinations of the car - flow directions of the dropped - line trains. For example, the maximum formation number within a section is set to 55, and the minimum formation number is set to 45. The x / y on the arc represents the flow on the edge and the remaining capacity. Assume the output train is f i For the arcs between the output train f and the sink T, only the first three arcs have flow [l1, l2, l3]. Depending on the case of the numerical example, the following situations may exist:

[0166] a. If the combination of the car - flow directions corresponding to f1, f2, and f3 is a possible combination, and the sum of the flows corresponding to the three arcs (l1 + l2 + l3) ∈ [45, 55], then it is considered that an adjustment plan for the dropped - line train has been found;

[0167] b. If the combination of the car - flow directions corresponding to f1, f2, and f3 is a possible combination, but l1 + l2 + l3 < 45, then consider running sectional - pickup trains or local trains. If the running requirements are not met, it is considered that the flow - distribution fails;

[0168] c. If the combination of the car - flow directions corresponding to f1, f2, and f3 is a possible combination, but l1 + l2 + l3 > 55, then deduct the excess freight cars in the order from bottom to top until the formation - quantity requirements are met;

[0169] d. If the combination of the car - flow directions corresponding to f1, f2, and f3 is not a possible combination, then it is considered that the adjustment of the dropped - line train fails;

[0170] In summary, if (l1 + l2 + l3) ∈ [45, 55], then the flow - distribution plan of the dropped - line train is adjusted to the car - flow group numbers corresponding to the three nodes f1, f2, and f3, and the departure time is the departure time of the original dropped - line train.

[0171] Example 2:

[0172] According to Example 1, this example is a specific numerical - example analysis of a railway technical - station flow - distribution optimization method considering the value concentration of freight cars. A certain regional railway network is as Figure 4 shown. The technical station C is a regional marshalling station. The technical operation time for trains to arrive and depart at this technical station is 30 minutes each, and the marshalling and un - marshalling time is 20 minutes each. The train weights corresponding to different marshalling destinations are shown in Table 4. The requirements for the formation number of departure trains are [45, 55], and the full - load axle constraint is [27000t, 33000t]. Select the arrival trains at Station C from 8:00 to 11:52 and the departure trains from 10:45 to 14:25 as the basis for flow - distribution at the technical station C. The information of the arrival trains at Station C within the stage is shown in Table 5, and the information of the departure trains is shown in Table 6;

[0173] Table 4 Train weights corresponding to different marshalling car - flow group numbers at the technical station

[0174]

[0175] Table 5 Information on the arriving car flow at technical stations

[0176]

[0177] Table 6 Information on the direction of the departing car flow at technical stations

[0178]

[0179] 1. Multi-stage car flow allocation plan considering the value concentration of freight cars:

[0180] Obtain the information on the constituent factors of the value concentration of the arriving car flow at technical stations, input it into the weighted rank sum ratio evaluation system to obtain the ranking of the value concentration of the arriving car flow. Taking the serial number at the end of the arriving train as the X-axis, the car flow group number as the Y-axis, and the value concentration of the car group as the Z-axis, draw the distribution diagram of the value concentration of the arriving car groups (see Figure 5 );

[0181] Select those with a WRSR evaluation value greater than or equal to 0.8 as high-value concentration freight cars, that is, WRSR ≥ 0.8;

[0182] Table 7 Distribution table of high-value concentration freight cars

[0183]

[0184] Table 8 First-stage - Results table of preferential car flow allocation for high-value concentration freight cars

[0185]

[0186] Table 9 Summary table of two-stage car flow allocation plan considering the value concentration of freight cars

[0187]

[0188]

[0189] The high-value concentration car group 10007 - 9 is marshaled into train 10008. Since this train does not meet the departure requirements, the car flow allocation fails, and the car flow allocation plan for this car group needs to be adjusted. Screen the departing trains that meet the requirements of the car flow group number from the earliest to the latest according to the departure order. It is found that according to the screening of the car flow group number, the subsequent trains that meet the requirements are train 10018 and train 10022; According to the calculation of the car flow connection conditions, it can be known that train 10018 meets the requirement of marshaling the high-value concentration car group 10007 - 9. Therefore, this car group is sequentially adjusted into train 10018, and a car flow allocation plan considering the value concentration of freight cars is formed.

[0190] Comparing the two results, it is found that compared with the traditional flow allocation scheme, the flow allocation scheme considering the value concentration of freight cars has the following optimizations: In the flow allocation process without considering the value concentration of freight cars, the high-value concentration car group 3 / 10007 / 6 is divided into two parts and respectively marshaled into the departure trains No. 10012 and No. 10016. However, in the flow allocation scheme considering the value concentration of freight cars, the entire car group is marshaled into train No. 10012, meeting the requirement of the high-value concentration car group for grouped arrival and departure; for the car groups with the arrival traffic flow directions of 8 and 9, the flow allocation scheme considering the value concentration of freight cars arranges the car groups that meet the traffic flow connection in the earliest departure train within the time window. Compared with the flow allocation scheme without considering the value concentration of freight cars, the car groups with the traffic flow group numbers of 8 and 9 are concentrated and marshaled in the departure train No. 10018, improving the transfer efficiency of the high-value concentration car groups at this station and providing guarantee for the flow allocation of the high-value concentration car groups at the subsequent technical stations.

[0191] In summary, compared with the existing flow allocation scheme, the multi-stage flow allocation scheme of the technical station considering the value concentration of freight cars realizes the requirement of the fastest departure of the high-value concentration car groups under the condition that the total detention time of freight cars in the stage plan remains unchanged. The total detention time of the high-value concentration car groups in the station is reduced by 620 minutes, and the average detention time of the high-value concentration freight cars in the station is reduced from 190.86 minutes to 178.76 minutes, with a decrease of 6.3%.

[0192] 2. Adjustment scheme for the formation content of trains with lost lines:

[0193] Construct a network model for adjusting the formation content of the traffic flow of trains with lost lines. In the input column [d1, d2, d3, d4], the existing flow allocation resources are arranged in priority, and then the points of the input column are arranged in the order of the callable flow allocation resources; in the output column [f1, f2, f3], the points with the same traffic flow direction as the high-value concentration freight cars are arranged in priority, and then the arrangement is carried out in a way that can form a marshalling combination with the high-value concentration traffic flow group number; for example, [1, 6, 7] and [1, 8, 9] are respectively the marshalling combinations that can be formed by the departure trains. If there are high-value concentration freight cars with the traffic flow group number of 1 in the station's stored cars, the arrangement order of the output column is designed as [1, 6, 7, 8, 9]; in the network diagram, the data (a, b) on the directed arc, where a refers to the capacity of the directed arc, specifically referring to the upper limit of the number of freight cars that can be allocated by this flow allocation scheme, and b refers to the number of freight cars allocated in the current iteration stage.

[0194] Using the value matching search algorithm VMSA, it is obtained that the marshalling direction included in the adjusted departure train No. 10008 is [1, 6, 7]. Since the number of freight cars meeting the conditions in the maximum flow network exceeds the full-axle constraint, adjustments are made according to the value concentration of freight cars and the marshalling destination. The number of formed cars is 54, and the formation content is shown in Table 10.

[0195] Table 10 Formation content of the newly issued train after adjustment

[0196]

[0197] In summary, the adjusted vehicle allocation plan considering the value concentration of freight cars is obtained. The analysis is carried out in the form of a horizontal bar chart. Taking the number of assembled cars as the horizontal axis and the number of departure train trips as the vertical axis, the vehicle allocation plan considering the value concentration of freight cars is as follows Figure 6 shown. In the figure, the classification of the value concentration of freight cars and the level of numerical values are distinguished according to colors. Among them, blue represents the vehicle group with ordinary value concentration, and orange represents the vehicle group with high value concentration. The darker the color, the higher the value concentration of the freight cars in the corresponding vehicle group;

[0198] The adjusted vehicle allocation plan of the technical station considering the value concentration of freight cars changes as shown in the following table compared with the original plan:

[0199] Table 11 Comparison of the results of each vehicle allocation plan

[0200]

[0201] By comparing the vehicle allocation plans before and after adjustment, it is found that the number of departure trains in the vehicle allocation plan adjusted by using the value matching search algorithm VMSA increases by 1 column; the vehicle stay time at the station is reduced by 159.3 h; the average stay time of freight cars at the station and the average stay time of high-value-concentration freight cars at the station are reduced by 0.27 h and 0.29 h respectively. Among them, in the adjusted and optimized plan, the average stay time of high-value-concentration freight cars at the station is reduced by 0.49 h compared with the original plan, and the optimization ratio reaches 15.41%.

[0202] 3. Analysis of vehicle allocation results:

[0203] For the vehicle allocation plan solved by the VMSA algorithm, a comparison chart of the vehicle allocation plan considering the value concentration of freight cars adjusted by the VMSA algorithm and the vehicle allocation plan not considering the value concentration of freight cars is drawn, as Figure 7 shown. Among them, each departure train corresponds to two vehicle allocation plans. The upper one is the vehicle allocation plan considering the value concentration of freight cars, and the lower one is the vehicle allocation plan not considering the value concentration of freight cars; for the original plan and the new vehicle allocation plan obtained by the VMSA algorithm, the distribution of each vehicle group is as Figure 8 、 Figure 9 shown;

[0204] By comparing the vehicle allocation plan solved by the traditional vehicle allocation model and the VMSA algorithm, it can be found that the solution result of the VMSA algorithm has the following advantages: Figure 7 As shown in, this algorithm ensures that the high-value concentration departs as soon as possible, and moves the high-value-concentration vehicle groups originally distributed in the subsequent trains to the previous trains that meet the connection requirements; after adjustment, the under-axle trains in the original plan depart successfully, and as many ordinary-value-concentration freight cars that meet the conditions as possible are incorporated into the departure trains, effectively reducing the total stay time of freight cars at the station.

[0205] The preferred embodiments of the present invention disclosed above are only used to help illustrate the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the art can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A railway technical station flow optimization method considering freight car value concentration, characterized in that: The following steps are involved: S1, truck value concentration composition analysis and evaluation, constructing a technical station distribution resource value concentration evaluation index system, analyzing the weight of each evaluation index by the entropy weight method, and then processing the data by the deviation standardization method, inputting the weighted rank sum ratio evaluation system to obtain the value concentration ranking of trucks arriving at the technical station; S2, model establishment, the distribution process is divided into high value concentration distribution stage, ordinary value concentration distribution stage and distribution scheme adjustment stage, by adjusting the constraints at different stages, to form a distribution model that comprehensively considers the truck value concentration and the total stay time of the truck; S3, model solution, is divided into three stages: In the first stage, the station disassembly sequence under the maximum vehicle flow connection goal and the initial flow allocation plan without considering the value concentration of freight cars are solved; In the second stage, the high-value concentration and ordinary-value concentration trucks are allocated in turn under the maximum vehicle flow sequence. In the third stage, high-value freight cars that failed in the allocation are matched with the earliest possible departure trains. At the same time, the composition of the stored cars at the station is analyzed, and the contents of the lost train formation are considered to be modified. The first and second stages are solved by designing a maximized neighborhood search algorithm. The third stage is solved by designing a technical station lost line train formation adjustment algorithm that takes into account the maximum departure of high-value concentration freight cars, combining the characteristics of station inventory resources and the departure time of lost line trains within the stage plan, referring to the network model solution algorithm in static distribution.

2. According to claim 1, a railway technical station flow optimization method considering freight car value concentration is characterized in that: In S1, five indicators, namely the number of trucks from the group to the departure group, the time value of the cargo, the delay time, the subsequent operation and the level of the cargo owner, are selected to construct an indicator system for evaluating the value concentration of the technical station distribution resources. The specific steps for evaluating the value concentration of trucks are as follows: S1.1, data standardization, except for the cargo owner grade and subsequent operation status, each evaluation index uses the deviation standardization method and applies the following formula: Transform the data to [0, 1], and then divide the interval [0, 1] into 10 equal parts. Divide the data in each interval into 10 levels, corresponding to numbers 1-10 from low to high. For the sample data, transform it into a matrix. The matrix has m samples, and each sample has n indicators. S1.2, rank, construct the evaluation matrix A of the evaluation object ij , where the values ​​of the cargo owner level and subsequent operation status are based on actual data, and the cargo time value and delay time data are based on the y after data standardization in S1.1 ij , rank each indicator in the evaluation matrix, and sort according to the type of indicator. The benefit indicators are ranked from small to large, and the cost indicators are ranked from large to small. The average value is taken for the same indicator data. The obtained rank matrix is ​​recorded as R ij ; S1.3, weight design, the entropy weight method is used to determine the weight of each indicator, denoted as [λ1,λ2,…,λ n ]; S1.4, calculate the weighted rank sum ratio WRSR of the i-th evaluation object, denoted as δ WRSRi ; S1.5, sort by bin, according to δ WRSRi Sort the evaluation objects.

3. According to claim 1, a railway technical station flow optimization method considering freight car value concentration is characterized in that: The establishment of the flow distribution model in S2 includes the following steps: S2.1, model assumptions, assume that the technical station adopts a single-push and single-slide operation mode, and the trains arrive on time in the stage plan, without considering the emergencies and uncertainties in the operation process of the technical station in the region; assume that the equipment and staff of the station meet the operation requirements and can carry out the technical operations of the arriving and departing trains in time; do not consider abnormalities and maintenance status such as shunting preparation and failure, and assume that the shunting operation is continuous during working hours without suspension; to simplify the modeling difficulty, it is assumed that the value concentration of freight cars with the same vehicle group number in the same arriving train of the technical station is the same; S2.2, establish the objective function, split the technical station distribution process into two stages, and the shortest truck stay time at the station is used as the objective function in both stages: After giving priority to high-value trucks in the arriving traffic, the remaining traffic is allocated according to the same objective function, and finally the allocation plan with the shortest total stay time of trucks in the current stage is obtained; S2.3, model constraints. In the technical station flow distribution model, the main constraints considered include dismantling operation constraints, marshaling operation constraints, train flow connection constraints and departure train full axle constraints: The train arrival disassembly operation constraints are expressed as follows: T JTKS (Y h )≥T DD (Y h )+T DDJS h=1,2,...,m T JTKS (Y h )≥T JTKS (Y h-1 )+T JT h=2,...,m Respectively represent the disintegration order is Y h The actual dismantling time of the train is not earlier than the earliest dismantling time of the train and the dismantling order is Y h The actual dismantling time of the train is not earlier than the arrival order of Y h-1 That is, the disintegration of the previous sequence car is completed; The train marshaling operation constraints are: T ZWJSBZ (Z l )≥T CF (Z l )-T CFJS l=1,2,...,n T KSBZ (Z l )≥T KSBZ (Z l-1 )+T BZ l=1,2,...,n Respectively indicate that the grouping order is Z l The latest marshaling time of the departure train is T ZWJSBZ No earlier than the actual end of the marshaling time and the marshaling order is Z l The marshaling start time of the departing train is not earlier than the actual marshaling end time of the preceding departing train; Traffic flow continuation constraints: They respectively indicate that only when the marshaling start time of the departure train j is later than the disassembly end time of the arrival train i, can the arrival train i provide traffic for the departure train j, and the number of freight cars with traffic group number k provided by the arrival train i to the departure train j cannot exceed the number of freight cars with the corresponding traffic group number contained in the arrival train; the marshaling content association constraints are: Departure train full axle constraints: They respectively indicate that the length of any departure train has a requirement and needs to be limited to [M min ,M max ], and the weight of any departing train is required to be limited to [W min ,W max ].

4. According to claim 1, a railway technical station flow optimization method considering freight car value concentration is characterized in that: In S3, the maximum neighborhood search algorithm includes the following steps: a. Calculate the ideal arrival and departure train flow matrix C without considering equipment occupancy conflicts and the current train disassembly sequence. LX The actual arrival and departure train flow matrix C considering equipment occupancy conflicts and train disassembly sequence SJ , using the ideal arrival and departure traffic matrix C LX Subtract the actual arrival and departure traffic matrix C SJ , get the traffic that does not meet the traffic continuation constraint conditions under the current disassembly order, that is, the traffic corresponding to the position greater than 0 in the matrix; b. Find all positions greater than 0 in the departure flow connection difference matrix, generate a candidate set U of de-marshalling adjustment points, let u1 be the first adjustment point in the candidate set, and the total number of adjustment points is N TZ , each time a point in the candidate set is selected as the point to be adjusted to construct a neighborhood, and the corresponding disassembly time advance amount corresponding to the disassembly order adjustment of the arriving train at the point under the current disassembly order is obtained, and the corresponding marshaling time delay amount corresponding to the marshaling order adjustment of the departing train at the point is obtained; At the same time, the connection time difference between the arriving train and the departing train corresponding to the disassembly adjustment point is calculated, and all suitable train disassembly order adjustment quantities that satisfy the change value just greater than or equal to the absolute value of the train connection time difference at this point are found from the time change matrix, thereby generating the neighbors of the current solution; the disassembly order of the arriving and departing trains to which this part of the traffic belongs is adjusted, and the stage-planned arrival and departure train sequence is used as the initial disassembly order. Combined with time constraints such as technical operation time standards, the actual arrival and departure traffic flow connection matrix under the current disassembly order of the technical station is calculated; c. Solve the flow distribution results of all neighborhood solutions in turn until all N TZ The neighborhood solutions of all adjustment points in get the optimal flow allocation solution T Best ; Take the stage planned train arrival and departure sequence as the initial disassembly sequence, calculate the stage planned train arrival and departure information under the current disassembly sequence, take the current train arrival and departure sequence of the technical station as the disassembly sequence of the train in the flow allocation plan, and get the initial solution T, let T Best =T,T now =T,T now <T Best The train arrival and departure sequence is taken as the current optimal train disassembly sequence.

5. According to claim 1, a railway technical station flow optimization method considering freight car value concentration is characterized in that: In S3, the algorithm for adjusting the train formation content of the technical station lost line to maximize the departure of high-value trucks includes the following steps: a. According to the train set assigned to the lost line train, design the adjustment strategy of the train set that failed to be assigned: b. The problem of adjusting the content of lost-line trains to enable high-value freight cars stored at stations to depart as soon as possible is regarded as a static flow distribution problem. By establishing a flow distribution network model, the static flow distribution problem is solved to obtain the maximum departure plan for lost-line trains with adjusted flow distribution content.

Images