A method for optimizing urban delivery vehicle routing based on realistic traffic conditions

Abstract

The present invention discloses a method for optimizing urban delivery vehicle routing based on a realistic traffic environment. The method comprises the following steps: temporarily ignoring two-stage vehicle routing, simplifying the urban delivery vehicle routing optimization problem based on the realistic traffic environment into a two-stage facility location selection problem; based on the solution of the two-stage facility location selection problem and in combination with the realistic urban traffic environment, transforming the urban delivery vehicle routing optimization problem based on the realistic traffic environment into several random multi-path vehicle routing problems, and further transforming them into several random multi-path traveling salesman problems; then, transforming each random multi-path traveling salesman problem into a corresponding approximate deterministic planning problem, and obtaining the optimal vehicle routing for the two stages by integrating the solution results of all random multi-path traveling salesman problems. The method has high computational efficiency, and the solution results are more consistent with the realistic urban traffic environment. It can provide more scientific and effective guidance for third-party logistics companies in carrying out urban distribution, thereby reducing distribution costs and improving distribution efficiency.

Description

Technical Field

[0001] The present invention relates to logistics optimization technology, and in particular to a method for optimizing urban delivery vehicle routes based on a realistic traffic environment. Background Art

[0002] Cities are the primary hubs for production and life, and also the most concentrated areas for logistics activities. With the continuous development of the economy and society, the rise of e-commerce, the improvement of residents' living standards, and changing consumption habits, the demand for urban distribution services characterized by small batches, multiple batches, and strong timeliness is rapidly growing. A two-tiered urban logistics network is currently one of the most commonly adopted logistics network architectures by third-party logistics companies. Heavy or medium-sized trucks are responsible for transporting goods from a suburban distribution center to several distribution centers in the city. After being reassembled at these distribution centers, goods are then delivered by light or mini trucks to Cainiao stations, express supermarkets, or customers. How to optimize urban distribution based on this two-tiered logistics network is a hot topic of concern in academia and industry.

[0003] There are two main existing urban distribution models: one is that operations managers determine which distribution centers to use based on experience, and then vehicle drivers decide which routes to use for urban distribution based on experience; the other assumes that there is only one path (the shortest path) between any two logistics nodes (distribution center, distribution center, customer) in a two-layer urban logistics network, and the generalized travel cost (travel time, distance or fee, etc.) of this path is fixed. The optimization of urban distribution vehicle routes in the two-layer logistics network is abstracted into a classic two-stage location-routing problem, which is solved and the solution is used to guide actual urban distribution.

[0004] However, when conducting urban distribution based on the experience of relevant personnel, it is difficult to guarantee that the decisions made are optimal. In addition, the real urban traffic environment is relatively complex. There are often multiple paths between two logistics nodes, and the generalized travel cost of each path is highly uncertain due to factors such as traffic congestion and weather conditions. Therefore, using the solutions of the classic two-stage location and routing problem to guide urban distribution has significant limitations. Summary of the Invention

[0005] Purpose of the invention: The purpose of the present invention is to provide a method for optimizing the path of urban distribution vehicles based on a realistic traffic environment. The method splits the path optimization problem of urban distribution vehicles based on a realistic traffic environment into three sub-problems: a two-stage facility location problem, a random multi-path vehicle path problem, and a random multi-path traveling salesman problem. By solving the three relatively simple sub-problems, the original complex combinatorial optimization problem is solved.

[0006] Technical solution: The present invention provides a method for optimizing the path of urban delivery vehicles based on a realistic traffic environment, comprising the following steps:

[0007] S1. Ignoring the two-stage vehicle routes from the distribution center to the delivery center and from the delivery center to the customer, the urban delivery vehicle routing optimization problem based on the real traffic environment is simplified into a two-stage facility location problem. A two-stage facility location model is established with the minimization of the total logistics service cost as the objective function and the satisfaction of the logistics needs of all customers and the operating capacity limit of the delivery center as the constraints. The location plan for the delivery center and the customer allocation plan are obtained by solving the problem.

[0008] S2. Based on the distribution center location selection and customer allocation scheme obtained in step S1, and in combination with the actual urban traffic environment, the urban distribution vehicle routing optimization problem based on the actual traffic environment is converted into several random multi-path vehicle routing problems. The selected distribution centers and the customers assigned to each selected distribution center are then grouped according to the rated load of the distribution vehicles. Each random multi-path vehicle routing problem is further converted into several random multi-path traveling salesman problems, and each random multi-path traveling salesman problem is modeled.

[0009] S3. Convert each random multi-path traveling salesman problem obtained in step S2 into a corresponding approximate deterministic planning problem and solve it. By integrating the solution results of all random multi-path traveling salesman problems, the optimal path for urban delivery vehicles based on the actual traffic environment is obtained.

[0010] Furthermore, the two-stage facility location model in step S1 is:

[0011] Objective function:

[0012] Constraints:

[0013]

[0014]

[0015]

[0016] Among them, S represents the candidate set of distribution centers; C represents the customer set; parameter f s 、w s They represent the activation cost and operating capacity of distribution center s, s∈S; d c represents the logistics demand of customer c, c∈C; H 0s H represents the unit transportation cost of the shortest route between the distribution center and the delivery center s; scThe unit transportation cost of the shortest route between distribution center s and customer c; the decision variable Z s Indicates whether the distribution center s is selected. If so, the value is 1, otherwise it is 0; u sc Indicates whether customer c is served by distribution center s. If so, the value is 1, otherwise the value is 0.

[0017] Furthermore, in step S2, a constrained K-means algorithm is used to group the selected distribution centers and the customers assigned to each selected distribution center according to the rated load of the delivery vehicles. The constrained K-means algorithm process is as follows:

[0018] (1) Input raw data: including the latitude and longitude coordinates of each demand point, the logistics demand of each demand point, and the rated load Q of the delivery vehicle; the demand point refers to the selected distribution center or the customer assigned to a selected distribution center;

[0019] (2) Initialization parameters: including the number of centroids The service capacity of each centroid is Q, and the longitude and latitude coordinates of K demand points are randomly selected as the initial coordinates of the K centroids; the centroid is the point represented by the mean of the longitude and latitude coordinates of a group of demand points;

[0020] (3) Arrange the demand points in descending order according to the logistics demand;

[0021] (4) Allocate the demand points one by one to the centroid closest to them with remaining service capacity;

[0022] (5) If all demand points have been allocated, proceed to step (6). If some demand points cannot be allocated, set K = K + 1 and repeat steps (2) to (4) until all demand points have been allocated.

[0023] (6) Update the coordinates of each centroid, that is, replace its current longitude and latitude coordinates with the mean of the longitude and latitude coordinates of the demand points served by each centroid;

[0024] (7) Repeat steps (2) to (6) for a set number of times and return the grouping results, in which the demand points served by each centroid are grouped.

[0025] Furthermore, each random multi-path traveling salesman problem in step S2 is modeled as follows:

[0026] Objective function:

[0027] Constraints:

[0028]

[0029]

[0030]

[0031] in:

[0032]

[0033] Constraints:

[0034]

[0035] Where V represents the point set consisting of the distribution center and the selected distribution center or a selected distribution center and its assigned customers; V′ represents a non-empty proper subset of V; E represents the edge set consisting of any two points in V; P ij represents the path set between points i and j in V; parameter represents the random travel cost of path p between points i and j, where represents the fixed travel cost of path p between points i and j, represents the random fluctuation of the fixed travel cost of path p; the decision variable X ij Indicates whether the delivery vehicle passes through point i and arrives at point j. If yes, the value is 1, otherwise the value is 0; Indicates whether the delivery vehicle passes through path p from point i to point j. If yes, the value is 1, otherwise the value is 0.

[0036] Furthermore, each random multi-path traveling salesman problem in step S3 is transformed into the following approximate deterministic planning problem:

[0037] Objective function:

[0038] Constraints:

[0039]

[0040]

[0041]

[0042] in:

[0043]

[0044] Among them, |P ij | represents the number of paths between points i and j; f Drepresents the objective function value of the traveling salesman problem consisting of the shortest path between any two points, |V| represents the number of points in V, b represents the cost of the shortest path between any two points in V, and β represents the path set P ij The degree of dispersion of the paths;

[0045] Solve the above approximate deterministic planning problem and obtain the random multi-path traveling salesman problem path solution X ij The optimal value of as well as The optimal value of when When the value is 1, if the travel cost of path p between points i and j is the minimum, then The value is 1, otherwise it is 0.

[0046] Furthermore, the objective function value of the random multi-path traveling salesman problem in step S3 is obtained through the following Monte Carlo simulation steps:

[0047] (1) Randomly generate several traffic environment scenarios The scenario set Φ and the scenario The random fluctuation value of the fixed toll cost of the lower path p

[0048] (2) According to the optimal solution Calculate the scenario planning problem equivalent to the random multi-path traveling salesman problem in the scenario The objective function value under

[0049]

[0050] in:

[0051] Indicates the path p between points i and j in the scenario The random travel cost under represents the fixed travel cost of path p between points i and j;

[0052] (3) According to step (2), Calculate the objective function value f of the random multi-path traveling salesman problem as follows: DAP :

[0053]

[0054] in, Indicates the situation Probability of occurrence;

[0055] (4) Repeat steps (1) to (3) for a set number of times and return to f DAP value.

[0056] The present invention provides an urban delivery vehicle route optimization system based on a real traffic environment, comprising:

[0057] A two-stage facility location problem development module is used to simplify the urban distribution vehicle routing optimization problem based on real-world traffic conditions into a two-stage facility location problem, without considering the two-stage vehicle routing from the distribution center to the delivery center and from the delivery center to the customer. The solution then yields a location solution for the delivery center and a solution for the customer allocation.

[0058] The random multi-path vehicle routing problem establishment module is used to transform the urban distribution vehicle routing optimization problem based on the actual traffic environment into several random multi-path vehicle routing problems based on the obtained distribution center location plan and customer allocation plan, combined with the actual urban traffic environment;

[0059] A random multi-path traveling salesman problem (TMSP) model is used to group selected distribution centers and customers assigned to each selected distribution center according to the rated load of the delivery vehicles, and further transform each random multi-path vehicle routing problem into several random multi-path traveling salesman problems.

[0060] An approximate deterministic programming problem solving module is used to transform each obtained random multi-path traveling salesman problem into a corresponding approximate deterministic programming problem and solve it;

[0061] The optimal path solving module is used to obtain the optimal path for urban delivery vehicles based on the real traffic environment by integrating the solution results of all random multi-path traveling salesman problems.

[0062] A device of the present invention includes a memory and a processor, wherein:

[0063] a memory for storing computer programs capable of running on the processor;

[0064] The processor is configured to execute the steps of the above-mentioned method for optimizing the path of urban delivery vehicles based on a real traffic environment when running the computer program.

[0065] A storage medium of the present invention stores a computer program, which, when executed by at least one processor, implements the steps of the above-mentioned method for optimizing the path of urban delivery vehicles based on a real traffic environment.

[0066] Beneficial effects: Compared with the existing technology, the method of the present invention divides the urban distribution vehicle path optimization problem based on the actual traffic environment into three sub-problems: a two-stage facility site selection problem, a random multi-path vehicle path problem, and a random multi-path traveling salesman problem, and solves them separately. It effectively solves the problem that the calculation time increases exponentially with the problem scale when solving complex combinatorial optimization problems, and the calculation efficiency is greatly improved; by considering the existence of multiple paths between any two logistics nodes and the uncertainty of the generalized travel cost of each path, the solution is more in line with the actual urban traffic environment, and can provide more scientific and powerful guidance for third-party logistics companies to carry out urban distribution, reduce distribution costs and improve distribution efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 This is a schematic diagram of the urban delivery vehicle route optimization based on the actual traffic environment of the present invention;

[0068] Figure 2 This is a schematic diagram of urban delivery vehicle routing optimization based on the classic two-stage location-routing problem;

[0069] Figure 3 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0070] In a real urban traffic environment, there are usually multiple paths between any two logistics nodes, and the travel cost of each path is highly uncertain due to factors such as traffic congestion and weather conditions. Therefore, the urban distribution vehicle path optimization based on the real traffic environment is an uncertain NP-hard combinatorial optimization problem, such as Figure 1 As shown in the figure, the solution is extremely difficult. It is often impossible to make the best decision when carrying out urban distribution based solely on the experience of relevant personnel. Ignoring the actual urban traffic environment, the optimization of urban distribution vehicle routes in the two-layer logistics network is abstracted into a classic two-stage location-routing problem, such as Figure 2 As shown, using its solution to guide urban distribution has great limitations.

[0071] The present invention proposes a method for optimizing urban delivery vehicle routing based on a realistic traffic environment. Without changing the essential characteristics of the original problem, the method splits the original problem into three sub-problems: a two-stage facility location problem, a random multi-path vehicle routing problem, and a random multi-path traveling salesman problem, and solves each of these sub-problems separately. The method includes the following steps: step 1, temporarily ignoring the two-stage vehicle routes from the distribution center to the delivery center and from the delivery center to the customer, simplifying the urban distribution vehicle route optimization problem based on the actual traffic environment into a two-stage facility location problem, and solving the location scheme of the distribution center and the customer allocation scheme; step 2, based on the distribution center location scheme and customer allocation scheme obtained in step 1, and in combination with the actual urban traffic environment, transforming the urban distribution vehicle route optimization problem based on the actual traffic environment into several random multi-path vehicle route problems, and then grouping the selected distribution centers and the customers assigned to each selected distribution center according to the rated load of the delivery vehicles, and further transforming each random multi-path vehicle route problem into several random multi-path traveling salesman problems; step 3, performing an approximate transformation on each random multi-path traveling salesman problem obtained in step 2, that is, transforming the random multi-path traveling salesman problem into a corresponding approximate deterministic planning problem, and then solving them one by one.

[0072] like Figure 3 The specific implementation steps are as follows:

[0073] S1. Without considering the two-stage vehicle routes for the time being, by constructing the following two-stage facility location model, the urban distribution vehicle route optimization problem based on the real traffic environment is simplified into a two-stage facility location problem.

[0074] Objective function:

[0075] Constraints:

[0076]

[0077]

[0078]

[0079] Among them, S represents the candidate set of distribution centers; C represents the customer set; parameter f s 、w s They represent the activation cost and operating capacity of distribution center s, s∈S; d c represents the logistics demand of customer c, c∈C; H 0s H represents the unit transportation cost of the shortest route between the distribution center and the delivery center s; sc The unit transportation cost of the shortest route between distribution center s and customer c; the decision variable Z sIndicates whether the distribution center s is selected. If so, the value is 1, otherwise it is 0; u sc Indicates whether customer c is served by distribution center s. If so, the value is 1, otherwise the value is 0.

[0080] Since the constructed two-stage facility location model belongs to a linear programming model, the decision variable Z can be obtained by embedding the Cplex solver in the specific implementation process of the algorithm. s The value of (location plan of distribution center) and u sc The value of (customer's allocation plan).

[0081] S2, based on the distribution center location plan and customer allocation plan obtained in step S1, and taking into account the actual urban traffic environment, the urban distribution vehicle path optimization problem based on the actual traffic environment is transformed into |∑ s Z s +1| random multi-path vehicle routing problem. Since the vehicle routing problem itself is an NP-hard combinatorial optimization problem, the difficulty of solving it is still very high when uncertainty is considered. Therefore, we use the constrained K-means algorithm, an unsupervised learning artificial intelligence algorithm, to group the selected distribution centers and the customers served by each selected distribution center. The constrained K-means algorithm process is as follows:

[0082] (1) Input raw data: including the latitude and longitude coordinates of each demand point (the selected distribution center or the customer assigned to a selected distribution center), the logistics demand of each demand point, and the rated load Q of the delivery vehicle;

[0083] (2) Initialization parameters: including the number of centroids (the points represented by the mean of the latitude and longitude coordinates of a set of demand points) The service capacity of each centroid is Q, and the longitude and latitude coordinates of K demand points are randomly selected as the initial coordinates of the K centroids;

[0084] (3) Arrange the demand points in descending order according to the logistics demand;

[0085] (4) Allocate the demand points one by one to the centroid closest to them with remaining service capacity;

[0086] (5) If all demand points have been allocated, proceed to step (6). If some demand points cannot be allocated, set K = K + 1 and repeat steps (2) to (4) until all demand points have been allocated.

[0087] (6) Update the coordinates of each centroid, that is, replace its current longitude and latitude coordinates with the mean of the longitude and latitude coordinates of the demand points served by each centroid;

[0088] (7) Repeat steps (2) to (6) 300 to 500 times and return the grouping results, in which the demand points served by each centroid are grouped.

[0089] Based on the results returned by the constrained K-means algorithm, each random multi-path vehicle routing problem can be split into K random multi-path traveling salesman problems. Each random multi-path traveling salesman problem is modeled as follows:

[0090] Objective function:

[0091] Constraints:

[0092]

[0093]

[0094]

[0095] in:

[0096]

[0097] Constraints:

[0098]

[0099] Where V represents the point set consisting of the distribution center and the selected distribution center or a selected distribution center and its assigned customers; V′ represents a non-empty proper subset of V; E represents the edge set consisting of any two points in V; P ij represents the path set between points i and j in V; parameter represents the random travel cost of path p between points i and j, where represents the fixed travel cost of path p between points i and j, represents the random fluctuation of the fixed travel cost of path p; the decision variable X ij Indicates whether the delivery vehicle passes through point i and arrives at point j. If yes, the value is 1, otherwise the value is 0; Indicates whether the delivery vehicle reaches point j from point i via path p. If yes, the value is 1, otherwise the value is 0.

[0100] S3. Since the random multi-path traveling salesman problem is still a random NP-hard combinatorial optimization problem, and given that the travel costs of different paths between any two logistics nodes in a real urban road network have left-tail asymptotic independence, each random multi-path traveling salesman problem can be transformed into the following approximate deterministic planning problem.

[0101] Objective function:

[0102] Constraints:

[0103]

[0104]

[0105]

[0106] in:

[0107]

[0108] Among them, |P ij | represents the number of paths between points i and j; f D represents the objective function value of the traveling salesman problem consisting of the shortest path between any two points, |V| represents the number of points in V, b represents the cost of the shortest path between any two points in V, and β represents the path set P ij The degree of dispersion of the paths.

[0109] Since the approximate deterministic programming model of the random multi-path traveling salesman problem belongs to the classic deterministic traveling salesman problem, the Concorde solver is directly embedded in the algorithm implementation process to obtain the random multi-path traveling salesman problem path solution X ij The optimal value of as well as The optimal value of (when When the value is 1, if the travel cost of path p between points i and j is the minimum, then The value is 1, otherwise 0).

[0110] The objective function value of the random multi-path traveling salesman problem can be obtained through the following Monte Carlo simulation steps:

[0111] (1) Randomly generate 50 to 100 scenarios (traffic accidents, road maintenance, traffic control, rainy days, snowy days and other different traffic environments) scenario set Φ and in the scenario The random fluctuation value of the fixed toll cost of the lower path p

[0112] (2) Based on the obtained Calculate the scenario planning problem equivalent to the random multi-path traveling salesman problem in the scenario The objective function value under

[0113]

[0114] in:

[0115] Indicates the path p between points i and j in the scenario The random travel cost under represents the fixed travel cost of path p between points i and j;

[0116] (3) According to step (2), Calculate the objective function value f of the random multi-path traveling salesman problem as follows: DAP :

[0117]

[0118] in, Indicates the situation Probability of occurrence;

[0119] (4) Repeat steps (1) to (3) 10 to 20 times and return to f DAP value.

[0120] At this point, by integrating the solution results of all random multi-path traveling salesman problems, the optimal path for urban delivery vehicles based on the real traffic environment is obtained.

[0121] The present invention provides an urban delivery vehicle route optimization system based on a real traffic environment, comprising:

[0122] A two-stage facility location problem development module is used to simplify the urban distribution vehicle routing optimization problem based on real-world traffic conditions into a two-stage facility location problem, without considering the two-stage vehicle routing from the distribution center to the delivery center and from the delivery center to the customer. The solution then yields a location solution for the delivery center and a solution for the customer allocation.

[0123] The random multi-path vehicle routing problem establishment module is used to transform the urban distribution vehicle routing optimization problem based on the actual traffic environment into several random multi-path vehicle routing problems based on the obtained distribution center location plan and customer allocation plan, combined with the actual urban traffic environment;

[0124] A random multi-path traveling salesman problem (TMSP) model is used to group selected distribution centers and customers assigned to each selected distribution center according to the rated load of the delivery vehicles, and further transform each random multi-path vehicle routing problem into several random multi-path traveling salesman problems.

[0125] An approximate deterministic programming problem solving module is used to transform each obtained random multi-path traveling salesman problem into a corresponding approximate deterministic programming problem and solve it;

[0126] The optimal path solving module is used to obtain the optimal path for urban delivery vehicles based on the real traffic environment by integrating the solution results of all random multi-path traveling salesman problems.

[0127] A device of the present invention includes a memory and a processor, wherein:

[0128] a memory for storing computer programs capable of running on the processor;

[0129] The processor is used to execute the steps of the above-mentioned method for optimizing the path of urban distribution vehicles based on a real traffic environment when running the computer program, and can achieve the technical effect consistent with the above-mentioned method.

[0130] A storage medium of the present invention stores a computer program, which, when executed by at least one processor, implements the steps of the above-mentioned method for optimizing the path of urban distribution vehicles based on a real traffic environment and can achieve the same technical effects as the above-mentioned method. Specific embodiment:

[0132] Taking the urban distribution business of a domestic third-party logistics company in Nanjing as an example, the regional scope is limited to 32.0302°-32.0587°N and 118.7633°-118.7961°E, and the actual road network is used; the latitude and longitude coordinates of the distribution center, the latitude and longitude coordinates, operating capacity, and activation cost of the alternative distribution centers, the latitude and longitude coordinates and logistics demand of customers, and the rated load of different types of distribution vehicles are shown in Tables 1, 2, 3, and 4 respectively.

[0133] Table 1 Attribute data of distribution center

[0134] Distribution center number Latitude coordinates Longitude coordinates 0 32.0309896 118.7911367

[0135] Table 2 Attribute data of alternative distribution centers

[0136] Alternative distribution center number Latitude coordinates Longitude coordinates Service Capabilities / Unit Activation cost / yuan 1 32.0364380 118.7676140 1000 1960 2 32.0309663 118.7826840 1200 1680 3 32.0365787 118.7875045 1200 1650 4 32.0400361 118.7647691 1500 1350 5 32.0384226 118.7924583 1200 1350 6 32.0523673 118.7783981 7500 1200 7 32.0532290 118.7820615 1500 1700 8 32.0519122 118.7874088 1000 1580 9 32.0397213 118.7806895 1200 1840 10 32.0568122 118.7664490 1000 1050

[0137] Table 3 Customer attribute data

[0138]

[0139]

[0140]

[0141]

[0142] Table 4 Delivery vehicle attribute data

[0143] Delivery vehicle type Rated load / Unit Heavy trucks 3500 light trucks 350

[0144] By implementing step 1 using the above data, we can obtain the following distribution center location plan and customer allocation plan:

[0145] Distribution center site selection plan:

[0146] [1,2,5,7,8], that is, in this embodiment, five distribution centers numbered 1, 2, 5, 7, and 8 will be selected for use.

[0147] Customer allocation plan:

[0148] Distribution center 1: [29,30,31,34,35,60,61,90,91,93,94,95,96,97,102,110], that is, distribution center numbered 1 will serve customers numbered 29, 30, 31, 34, 35, 60, 61, 90, 91, 93, 94, 95, 96, 97, 102, and 110;

[0149] Distribution center 2: [17, 23, 32, 33, 50, 51, 52, 53, 54, 62, 66, 85, 86, 98, 103, 104], that is, distribution center numbered 2 will serve customers numbered 17, 23, 32, 33, 50, 51, 52, 53, 54, 62, 66, 85, 86, 98, 103, 104;

[0150] Distribution center 5: [19, 20, 21, 22, 24, 40, 65, 70, 71, 72, 73, 74, 75, 76, 77, 79, 80, 83, 84, 87, 88, 89, 105], that is, distribution center numbered 5 will serve customers numbered 19, 20, 21, 22, 24, 40, 65, 70, 71, 72, 73, 74, 75, 76, 77, 79, 80, 83, 84, 87, 88, 89, 105;

[0151] Distribution center 7: [12,13,14,15,18,26,27,28,36,37,41,45,47,55,56,57,58,59,63,64,67,68,69,81,82,92], that is, distribution center numbered 7 will serve customers numbered 12, 13, 14, 15, 18, 26, 27, 28, 36, 37, 41, 45, 47, 55, 56, 57, 58, 59, 63, 64, 67, 68, 69, 81, 82, 92;

[0152] Distribution center 8: [11,16,25,38,39,42,43,44,46,48,49,78,99,100,101,106,107,108,109], that is, distribution center numbered 8 will serve customers numbered 11, 16, 25, 38, 39, 42, 43, 44, 46, 48, 49, 78, 99, 100, 101, 106, 107, 108, and 109.

[0153] Based on the obtained distribution center location and customer assignment plans, implementing step 2 can split the original problem into 5 + 1 = 6 random multi-path vehicle routing problems. Using the constrained K-means algorithm (400 iterations), we can obtain the grouping plan of the distribution centers served by the distribution center and the customers served by each selected distribution center. The original problem is transformed into 19 random multi-path traveling salesman problems as follows:

[0154] Distribution center 0: [2,5,8], [1,7], that is, the five selected distribution centers are divided into two groups, [2,5,8] and [1,7], and the distribution services are provided by the distribution centers;

[0155] Distribution center 1: [29, 31, 94, 95, 96], [90, 91, 93, 97, 102, 110], [30, 34, 35, 60, 61], that is, the 16 customers served by distribution center numbered 1 are divided into three groups: [29, 31, 94, 95, 96], [90, 91, 93, 97, 102, 110], and [30, 34, 35, 60, 61] for distribution services;

[0156] Distribution center 2: [23, 53, 54, 86, 98, 104], [32, 33, 50, 62, 66], [17, 51, 52, 85, 103], that is, the 16 customers served by distribution center numbered 2 are divided into three groups: [23, 53, 54, 86, 98, 104], [32, 33, 50, 62, 66], and [17, 51, 52, 85, 103] for distribution services;

[0157] Distribution center 5: [21, 22, 24, 65, 73, 76, 79], [19, 20, 40, 72, 74], [70, 71, 83, 87, 88], [75, 77, 80, 84, 89, 105], that is, the 23 customers served by distribution center numbered 5 are divided into four groups: [21, 22, 24, 65, 73, 76, 79], [19, 20, 40, 72, 74], [70, 71, 83, 87, 88], [75, 77, 80, 84, 89, 105] for distribution services;

[0158] Distribution center 7: [12,13,26,27,36,67], [15,37,55,56,57], [18,41,45,47,68,81,82,92], [14,28,58,59,63,64,69], that is, the 26 customers served by distribution center numbered 7 are divided into four groups: [12,13,26,27,36,67], [15,37,55,56,57], [18,41,45,47,68,81,82,92], [14,28,58,59,63,64,69] for distribution services;

[0159] Distribution center 8: [16,25,39,100,101,108,109], [11,44,48,49,78,99], [38,42,43,46,106,107], that is, the 19 customers served by distribution center numbered 8 are divided into three groups: [16,25,39,100,101,108,109], [11,44,48,49,78,99], [38,42,43,46,106,107] for distribution services.

[0160] Based on the obtained distribution center grouping and customer grouping schemes, implementing step 3 (randomly generating 100 scenarios and performing 10 Monte Carlo simulation iterations) yields the following solutions (optimal paths) for 19 random multipath traveling salesman problems:

[0161] Distribution center 0: [0,2,8,5,0], [0,1,7,0], that is, distribution center numbered 0 needs to arrange two heavy trucks. One heavy truck starts from the distribution center and serves distribution centers numbered 2, 8, and 5 in sequence before returning to the distribution center. The other heavy truck starts from the distribution center and serves distribution centers numbered 1 and 7 in sequence before returning to the distribution center.

[0162] Distribution center 1: [1,95,96,31,29,94,1], [1,97,93,110,91,90,102,1], [1,61,30,60,35,34,1], that is, distribution center numbered 1 needs to arrange 3 light trucks. The first truck departs from the distribution center and serves customers with numbers 95, 96, 31, 29, and 94 in sequence, and then returns to the distribution center. The second truck departs from the distribution center and serves customers with numbers 97, 93, 110, 91, 90, and 102 in sequence, and then returns to the distribution center. The third truck departs from the distribution center and serves customers with numbers 61, 30, 60, 35, and 34 in sequence, and then returns to the distribution center.

[0163] Distribution center 2: [2,53,23,54,104,98,86,2], [2,32,66,33,50,62,2], [2,17,52,85,51,103,2], that is, distribution center numbered 2 needs to arrange 3 light trucks. The first truck starts from the distribution center and serves customers with numbers 53, 23, 54, 104, 98, and 86 in sequence, and then returns to the distribution center. The second truck starts from the distribution center and serves customers with numbers 32, 66, 33, 50, and 62 in sequence, and then returns to the distribution center. The third truck starts from the distribution center and serves customers with numbers 17, 52, 85, 51, and 103 in sequence, and then returns to the distribution center.

[0164] Distribution center 5: [5,24,76,79,22,21,73,65,5], [5,19,20,74,40,72,5], [5,88,87,70,71,83,5], [5,89,75,80,84,105,77,5], that is, distribution center numbered 5 needs to arrange 4 light trucks. The first truck starts from the distribution center and serves customers with numbers 24, 76, 79, 22, 21, 73, and 65 in turn, and then returns to the distribution center. The second truck starts from the distribution center and serves customers with numbers 19, 20, 74, 40, and 72 in turn, and then returns to the distribution center. The third truck starts from the distribution center and serves customers with numbers 88, 87, 70, 71, and 83 in turn, and then returns to the distribution center. The fourth truck starts from the distribution center and serves customers with numbers 89, 75, 80, 84, 105, and 77 in turn, and then returns to the distribution center.

[0165] Distribution center 7: [7,13,12,36,26,27,67,7], [7,57,56,55,15,37,7], [7,41,45,47,82,18,92,81,68,7], [7,14,69,58,59,64,28,63,7], that is, the distribution center numbered 7 needs to arrange 4 light trucks. The first one starts from the distribution center and serves the trucks numbered 13, 12, 36, 26, The second bus departs from the distribution center and serves customers with numbers 57, 56, 55, 15, and 37 in sequence, then returns to the distribution center. The third bus departs from the distribution center and serves customers with numbers 41, 45, 47, 82, 18, 92, 81, and 68 in sequence, then returns to the distribution center. The fourth bus departs from the distribution center and serves customers with numbers 14, 69, 58, 59, 64, 28, and 63 in sequence, then returns to the distribution center.

[0166] Distribution center 8: [8,39,16,101,108,109,25,100,8], [8,44,49,78,11,48,99,8], [8,107,106,46,38,43,42,8], that is, distribution center numbered 8 needs to arrange 3 light trucks. The first truck departs from the distribution center and serves customers with numbers 39, 16, 101, 108, 109, 25, and 100 in turn, and then returns to the distribution center. The second truck departs from the distribution center and serves customers with numbers 44, 49, 78, 11, 48, and 99 in turn, and then returns to the distribution center. The third truck departs from the distribution center and serves customers with numbers 107, 106, 46, 38, 43, and 42 in turn, and then returns to the distribution center.

[0167] At this point, by integrating the solution results of 19 random multi-path traveling salesman problems, the optimal route arrangement for urban distribution vehicles based on the actual traffic environment was obtained. The operating cost of the distribution center was 5,900 yuan, the distribution cost of the first stage was 1,226 yuan, and the distribution cost of the second stage was 5,293 yuan. The total cost of urban distribution was 12,419 yuan.

Claims

1. A method for optimizing the path of urban delivery vehicles based on a realistic traffic environment, characterized in that: The following steps are involved: S1. Ignoring the two-stage vehicle routes from the distribution center to the delivery center and from the delivery center to the customer, the urban delivery vehicle routing optimization problem based on the real traffic environment is simplified into a two-stage facility location problem. A two-stage facility location model is established with the minimization of the total logistics service cost as the objective function and the satisfaction of the logistics needs of all customers and the operating capacity limit of the delivery center as the constraints. The location plan for the delivery center and the customer allocation plan are obtained by solving the problem. S2. Based on the distribution center location selection and customer allocation scheme obtained in step S1, and in combination with the actual urban traffic environment, the urban distribution vehicle routing optimization problem based on the actual traffic environment is converted into several random multi-path vehicle routing problems. The selected distribution centers and the customers assigned to each selected distribution center are then grouped according to the rated load of the distribution vehicles. Each random multi-path vehicle routing problem is further converted into several random multi-path traveling salesman problems, and each random multi-path traveling salesman problem is modeled. S3. Convert each random multi-path traveling salesman problem obtained in step S2 into a corresponding approximate deterministic planning problem and solve it. By integrating the solution results of all random multi-path traveling salesman problems, the optimal path for urban delivery vehicles based on the actual traffic environment is obtained. Specifically, each random multi-path traveling salesman problem is converted into the following approximate deterministic planning problem: Objective function: Constraints: Where β represents the path set P ij The degree of dispersion of the paths in is expressed as: |P ij | represents the number of paths between points i and j; f D represents the objective function value of the traveling salesman problem consisting of the shortest path between any two points, |V| represents the number of points in V, b represents the cost of the shortest path between any two points in V, V represents the point set consisting of the distribution center and the selected distribution center or a selected distribution center and its assigned customers; V′ represents a non-empty proper subset of V; E represents the edge set consisting of any two points in V; P ij represents the path set between points i and j in V; parameter represents the random travel cost of path p between points i and j, where represents the fixed travel cost of path p between points i and j, represents the random fluctuation of the fixed travel cost of path p; the decision variable X ij Indicates whether the delivery vehicle passes through point i and arrives at point j. If yes, the value is 1, otherwise the value is 0; Solve the above approximate deterministic planning problem and obtain the random multi-path traveling salesman problem path solution X ij The optimal value of as well as The optimal value of Indicates whether the delivery vehicle passes through path p from point i to point j. If yes, the value is 1, otherwise it is 0. When the value is 1, if the travel cost of path p between points i and j is the minimum, then The value is 1, otherwise it is 0.

2. The method for optimizing the path of urban delivery vehicles based on a realistic traffic environment according to claim 1, characterized in that: The two-stage facility location model in step S1 is: Objective function: Constraints: Among them, S represents the candidate set of distribution centers; C represents the customer set; parameter f s 、w s They represent the activation cost and operating capacity of distribution center s, s∈S; d c represents the logistics demand of customer c, c∈C; H 0s H represents the unit transportation cost of the shortest route between the distribution center and the delivery center s; sc The unit transportation cost of the shortest route between distribution center s and customer c; the decision variable Z s Indicates whether the distribution center s is selected. If so, the value is 1, otherwise it is 0; u sc Indicates whether customer c is served by distribution center s. If so, the value is 1, otherwise the value is 0.

3. The method for optimizing the path of urban delivery vehicles based on a realistic traffic environment according to claim 1, characterized in that: In step S2, the constrained K-means algorithm is used to group the selected distribution centers and the customers assigned to each selected distribution center according to the rated load of the delivery vehicles. The constrained K-means algorithm process is as follows: (1) Input raw data: including the latitude and longitude coordinates of each demand point, the logistics demand of each demand point, and the rated load Q of the delivery vehicle; the demand point refers to the selected distribution center or the customer assigned to a selected distribution center; (2) Initialization parameters: including the number of centroids d c represents the logistics demand of customer c. The service capacity of each centroid is Q. The longitude and latitude coordinates of K demand points are randomly selected as the initial coordinates of the K centroids. The centroid refers to the point represented by the mean of the longitude and latitude coordinates of a group of demand points. (3) Arrange the demand points in descending order according to the logistics demand; (4) Allocate the demand points one by one to the centroid closest to them with remaining service capacity; (5) If all demand points have been allocated, proceed to step (6). If some demand points cannot be allocated, set K = K + 1 and repeat steps (2) to (4) until all demand points have been allocated. (6) Update the coordinates of each centroid, that is, replace its current longitude and latitude coordinates with the mean of the longitude and latitude coordinates of the demand points served by each centroid; (7) Repeat steps (2) to (6) for a set number of times and return the grouping results, in which the demand points served by each centroid are grouped.

4. The method for optimizing the path of urban delivery vehicles based on a realistic traffic environment according to claim 1, characterized in that: Each random multi-path traveling salesman problem in step S2 is modeled as follows: Objective function: Constraints: in: Constraints:

5. The method for optimizing the path of urban delivery vehicles based on a realistic traffic environment according to claim 1, characterized in that: The objective function value of the random multi-path traveling salesman problem in step S3 is obtained through the following Monte Carlo simulation steps: (1) Randomly generate several traffic environment scenarios The scenario set Φ and the scenario The random fluctuation value of the fixed toll cost of the lower path p (2) According to the optimal solution Calculate the scenario planning problem equivalent to the random multi-path traveling salesman problem in the scenario The objective function value under in: Indicates the path p between points i and j in the scenario The random travel cost under represents the fixed travel cost of path p between points i and j; (3) According to step (2), Calculate the objective function value f of the random multi-path traveling salesman problem as follows: DAP : in, Indicates the situation Probability of occurrence; (4) Repeat steps (1) to (3) for a set number of times and return to f DAP The value of .

6. A city delivery vehicle route optimization system based on real traffic environment, characterized by: include: A two-stage facility location problem development module is used to simplify the urban distribution vehicle routing optimization problem based on real-world traffic conditions into a two-stage facility location problem, without considering the two-stage vehicle routing from the distribution center to the delivery center and from the delivery center to the customer. The solution then yields a location solution for the delivery center and a solution for the customer allocation. The random multi-path vehicle routing problem establishment module is used to transform the urban distribution vehicle routing optimization problem based on the actual traffic environment into several random multi-path vehicle routing problems based on the obtained distribution center location plan and customer allocation plan, combined with the actual urban traffic environment; A random multi-path traveling salesman problem (TMSP) model is used to group selected distribution centers and customers assigned to each selected distribution center according to the rated load of the delivery vehicles, and further transform each random multi-path vehicle routing problem into several random multi-path traveling salesman problems. The approximate deterministic programming problem solving module is used to transform each random multi-path traveling salesman problem obtained into a corresponding approximate deterministic programming problem and solve it. Specifically, each random multi-path traveling salesman problem is transformed into the following approximate deterministic programming problem: Objective function: Constraints: Where β represents the path set P ij The degree of dispersion of the paths in is expressed as: |P ij | represents the number of paths between points i and j; f D represents the objective function value of the traveling salesman problem consisting of the shortest path between any two points, |V| represents the number of points in V, b represents the cost of the shortest path between any two points in V, V represents the point set consisting of the distribution center and the selected distribution center or a selected distribution center and its assigned customers; V′ represents a non-empty proper subset of V; E represents the edge set consisting of any two points in V; P ij represents the path set between points i and j in V; parameter represents the random travel cost of path p between points i and j, where represents the fixed travel cost of path p between points i and j, represents the random fluctuation of the fixed travel cost of path p; the decision variable X ij Indicates whether the delivery vehicle passes through point i and arrives at point j. If yes, the value is 1, otherwise the value is 0; Solve the above approximate deterministic planning problem and obtain the random multi-path traveling salesman problem path solution X ij The optimal value of as well as The optimal value of Indicates whether the delivery vehicle passes through path p from point i to point j. If yes, the value is 1, otherwise it is 0. When the value is 1, if the travel cost of path p between points i and j is the minimum, then The value is 1, otherwise it is 0; The optimal path solving module is used to obtain the optimal path for urban delivery vehicles based on the real traffic environment by integrating the solution results of all random multi-path traveling salesman problems.

7. A device, characterized in that: comprising a memory and a processor, wherein: a memory for storing computer programs capable of running on the processor; A processor is configured to execute the steps of a method for optimizing the path of an urban delivery vehicle based on a real traffic environment as described in any one of claims 1 to 5 when running the computer program.

8. A storage medium, characterized in that: The storage medium stores a computer program, which, when executed by at least one processor, implements the steps of a method for optimizing the path of an urban delivery vehicle based on a real traffic environment as described in any one of claims 1 to 5.

Images