Logistics Scheduling Method, Electronic Device and Storage Medium

Abstract

The present invention discloses a logistics scheduling method, an electronic device and a storage medium. The logistics scheduling method includes obtaining logistics orders within a certain time period; determining the number of transport vehicles M and the number of paths M according to the total demand of goods in the logistics orders and the load limit of the transport vehicles; calculating a feasible solution for M transport vehicles to execute the logistics orders according to the positions (x i , y i ) of each customer point i and the position (x0, y0) of the distribution center; constructing a logistics scheduling optimization model according to the actual scenario and the optimization objective; constructing a hyper-heuristic optimization pool, and the hyper-heuristic optimization pool includes various development strategies; based on the logistics scheduling optimization model, adjusting the feasible solution by using the development strategies in the hyper-heuristic optimization pool, and outputting the optimal feasible solution. The present invention can obtain the corresponding optimal feasible solution according to the specific logistics distribution scenario, greatly improving the feasibility, adaptability and portability of the method.

Description

Technical Field

[0001] The present invention belongs to the technical field of data processing, and particularly relates to a hyper-heuristic based green logistics scheduling method, an electronic device and a storage medium. Background Art

[0002] Green logistics scheduling optimization has important practical significance for saving enterprise distribution costs, improving vehicle distribution efficiency, reducing the use of vehicles and environmental pollution.

[0003] In 2013, Bektas et al. confirmed that improving the distribution route can reduce vehicle environmental pollution by analyzing the correlation between vehicle load, driving mileage and carbon emissions (Demir E, Bektas T, Laporte G. The bi-objective pollution-routing problem[J]. European Journal of Operational Research, 2014, 232(3): 464-478.); He Dongdong et al. introduced an approximate calculation method of fuel consumption and carbon emissions to reduce the waste gas pollution generated in logistics distribution, and designed an improved tabu search algorithm for solving (He Dongdong, Li Yinzhen. Optimization model of multi-type green vehicle routing problem[J]. Computer Applications, 2018, 38(12): 3618-3624+3637.); Xiao Y studied the green vehicle routing optimization under the agricultural-supermarket docking mode, and proposed that this mode can effectively reduce resource waste and achieve green distribution (Montoya A, Mendoza J. A Multi-Space sampling heuristic for the green vehicle routing problem[J]. Transportation Research Part C Emerging Technologies, 2016, 70: 113-128.); Chen Yuguang et al. proposed a way to build an urban green joint distribution platform to improve the green level of distribution routes (Chen Yuguang, Chen Zhixiang. Research on distribution vehicle routing problem based on on-time delivery and minimum fuel consumption[J]. Chinese Journal of Management Science, 2016(1): 18-24.).

[0004] However, the actual green logistics distribution scenario is complex and changeable. At present, the research on green logistics scheduling problems at home and abroad can only be implemented under specific demand scenarios, and it is difficult to provide effective scheduling references for actual green logistics distribution scenarios. Summary of the Invention

[0005] The object of the present invention is to provide a logistics scheduling method, an electronic device and a storage medium, so as to solve the problem that the existing technology is only implemented in specific demand scenarios and is difficult to be implemented in actual green logistics distribution scenarios, and has poor portability.

[0006] The present invention solves the above technical problems through the following technical solutions: A logistics scheduling method includes the following steps:

[0007] Obtain logistics orders within a certain period of time, where the logistics orders include at least one order, and each order includes at least one customer point i and its location (x i , y i ) and the demand quantity q of goods i ;

[0008] Determine the number of transport vehicles M and the number of paths M according to the total demand quantity of goods in the logistics orders and the load limit of the transport vehicles; where M ≤ N, and N is the number of customer points in the logistics orders;

[0009] According to the location (x i , y i ) of each customer point i and the location (x 0 , y 0 ) of the distribution center, calculate a feasible solution for M transport vehicles to execute the logistics orders;

[0010] Construct a logistics scheduling optimization model according to the actual scenario and the optimization objective;

[0011] Construct a hyper-heuristic optimization pool, and the hyper-heuristic optimization pool contains various development strategies;

[0012] Based on the logistics scheduling optimization model, use the development strategies in the hyper-heuristic optimization pool to adjust the feasible solution, and output the optimal feasible solution.

[0013] Further, when the load limit of each transport vehicle is the same, the calculation formula for the number of transport vehicles M or the number of paths M is:

[0014]

[0015] where Q is the load limit of a single transport vehicle, is the rounding symbol;

[0016] When the load limits of each transport vehicle are different, the minimum M value that satisfies the following formula is the number of transport vehicles or the number of paths, and the specific formula is:

[0017]

[0018] where Q k is the load limit of the kth transport vehicle.

[0019] Furthermore, the specific implementation process of obtaining a feasible solution for M transport vehicles to execute the logistics order is as follows:

[0020] Step 3.1: Define the customer set V and the path set R, where V = {1, 2, …, i, …, N}, R = {1, 2, …, k, …, M}, and assume that initially, the M paths in the path set R are empty paths;

[0021] Step 3.2: Calculate the distance d from each customer point i in the set V to the distribution center i0 ;

[0022] Step 3.3: Calculate the insertion cost of inserting customer point i into the k-th path, and obtain the insertion costs of inserting customer point i into the M paths. The specific formula for the insertion cost is as follows:

[0023]

[0024]

[0025] where, is the insertion cost of inserting customer point i into the k-th path, is the transportation cost of the k-th path before inserting customer point i, is the transportation cost of the k-th path after inserting customer point i, |R k | is the number of customer points on the k-th path before inserting customer point i, |R k′ | is the number of customer points on the k-th path after inserting customer point i, is the distance between customer point j and customer point j + 1 on the k-th path, is the distance between the distribution center and the first customer point on the k-th path, is the distance between the |R k |th customer point and the distribution center on the k-th path, is the distance between the inserted customer point i and the distribution center on the k-th path;

[0026] Step 3.4: Sort the M insertion costs corresponding to customer point i in ascending order, and calculate the regret value of customer point i. The specific formula is as follows:

[0027]

[0028] where RV i is the regret value of customer point i, is the third smallest insertion cost among the M insertion costs corresponding to customer point i, is the smallest insertion cost among the M insertion costs corresponding to customer point i, The corresponding path is the path with the minimum insertion cost for customer point i;

[0029] Step 3.5: Repeat Steps 3.3 and 3.4 to obtain the regret value for each customer point i in set V;

[0030] Step 3.6: Extract the customer point i corresponding to the maximum regret value in set V max , and insert customer point i max at the end of its path with the minimum insertion cost;

[0031] Step 3.7: Delete customer point i from set V max ;

[0032] Step 3.8: Determine whether set V is empty; when set V is not empty, repeat Steps 3.3 to 3.8; when set V is empty, obtain a feasible solution for M transport vehicles to execute the logistics order.

[0033] Furthermore, the specific expression of the logistics scheduling optimization model is:

[0034]

[0035]

[0036] where V 0 = {0, 1, 2, …, N}, V = {1, 2, …, i, …, N}, R = {1, 2, …, k, …, M}, V is the customer set, V 0 is the set of customers and the distribution center, i, j = 0 represents the distribution center, j ≠ i, R is the path set; C is the transportation cost; d ij is the distance between point i and point j; is a binary variable, taking a value of 1 or 0. When , it means that in the k-th path, the transport vehicle travels from point i to point j; Q k is the load limit of the k-th transport vehicle.

[0037] Furthermore, the hyper-heuristic optimization pool includes a node reallocation strategy, a two-node exchange strategy, a vertex-arc exchange strategy, a two-arc exchange strategy, a three-arc exchange strategy, a chain relocation strategy, and a cross-exchange strategy.

[0038] Furthermore, based on the logistics scheduling optimization model, the specific implementation process of adjusting the feasible solution using the exploitation strategy in the hyper-heuristic optimization pool is:

[0039] Step 6.1: Perform full-strategy adjustment on the feasible solution for T rounds using all the development strategies in the hyper-heuristic optimization pool, to obtain the first feasible solution after T rounds and the normalized adjustment probability of each development strategy; where T≥1;

[0040] Step 6.2: Randomly select a development strategy from the hyper-heuristic optimization pool to perturb the first feasible solution, to obtain a second feasible solution;

[0041] Step 6.3: Select development strategies from the hyper-heuristic optimization pool according to the normalized adjustment probability of each development strategy to perform optimization adjustment on the second feasible solution, to obtain a third feasible solution, and calculate the selection adjustment probability of each development strategy after optimization adjustment;

[0042] Step 6.4: Based on the selection adjustment probability of each development strategy after optimization adjustment, repeatedly execute Steps 6.2 and 6.3 on the third feasible solution until the third feasible solution has no replacement for G consecutive times, then output the optimal feasible solution.

[0043] Further, in Step 6.1, the specific implementation process of the full-strategy adjustment for T rounds is as follows:

[0044] Step 6.11: Randomly sort all the development strategies in the hyper-heuristic optimization pool;

[0045] Step 6.12: Use the development strategy in the sorting to adjust the feasible solution, to obtain the adjusted feasible solution;

[0046] Step 6.13: When the adjusted feasible solution meets the constraint conditions in the logistics scheduling optimization model and the transportation cost of the adjusted feasible solution is less than the transportation cost of the feasible solution before adjustment, replace the feasible solution before adjustment with the adjusted feasible solution, and record the replacement times of the current development strategy;

[0047] Step 6.14: Proceed to the next development strategy, and repeatedly execute Steps 6.12 to 6.14 until all the development strategies in the hyper-heuristic optimization pool are completed, that is, the current loop is completed;

[0048] Step 6.15: Add 1 to the loop count, and determine whether the loop count is equal to the set loop count; if not, repeatedly execute Steps 6.12 to 6.15 on the feasible solution obtained in the current loop; if so, obtain the first feasible solution of the current round and the total replacement times of each development strategy; where the initial value of the loop count is 0;

[0049] Step 6.16: Increment the round number by 1, and check if the round number is equal to the set round number T. If not, repeat steps 6.12 to 6.16 for the first feasible solution obtained in the current round. If so, obtain the first feasible solution after T rounds and the total replacement times of each development strategy; the initial value of the round number is 0.

[0050] Preferably, after the full strategy adjustment, calculate the normalized adjustment probability according to the total replacement times of each development strategy. The specific calculation formula is:

[0051]

[0052] where, is the normalized adjustment probability of the i-th development strategy after the full strategy adjustment, is the corrected adjustment probability of the i-th development strategy after the full strategy adjustment, N T is the number of development strategies in the hyper-heuristic optimization pool, is the total replacement times of the i-th development strategy after the full strategy adjustment, σ is the correction parameter, and S is the number of loops in each round.

[0053] Furthermore, in step 6.2, the specific implementation process of the perturbation is as follows:

[0054] Step 6.21: Calculate the transportation cost C of the first feasible solution according to the logistics scheduling optimization model 1 ;

[0055] Step 6.22: Randomly select an integer between [M, 2M] and assign it to m;

[0056] Step 6.23: Randomly select a development strategy from the hyper-heuristic optimization pool to adjust the first feasible solution, obtain the adjusted feasible solution, and calculate the transportation cost of the adjusted feasible solution;

[0057] Step 6.24: Repeat step 6.23 (m - 1) times to obtain m adjusted feasible solutions and their transportation costs;

[0058] Step 6.25: Select the feasible solutions whose transportation costs are less than (1 + dev)C 1 from the m adjusted feasible solutions, where dev is the cost tolerance coefficient;

[0059] Step 6.26: Randomly select a solution from the feasible solutions whose transportation costs are less than (1 + dev)C 1 as the second feasible solution.

[0060] Preferably, in step 6.3, the roulette wheel selection algorithm is used to select a development strategy from the hyper-heuristic optimization pool according to the adjustment probability of each development strategy to optimize and adjust the second feasible solution.

[0061] Preferably, in step 6.3, the calculation formula for the selection and adjustment probability of each development strategy after optimization and adjustment is as follows:

[0062]

[0063] Wherein, is the selection and adjustment probability of the i-th development strategy in the (g + 1)-th optimization and adjustment process, is the normalized adjustment probability of the i-th development strategy in the g-th optimization and adjustment process, is the normalized adjustment probability of the i-th development strategy after full strategy adjustment, a is the historical experience coefficient, is the corrected adjustment probability of the i-th development strategy in the g-th optimization and adjustment process, σ is the correction parameter, is the replacement times of the i-th development strategy in the g-th optimization and adjustment process, is the selection times of the i-th development strategy in the g-th optimization and adjustment process, N T is the number of development strategies in the hyper-heuristic optimization pool.

[0064] Based on the same inventive concept, the present invention also provides an electronic device, including a memory and a processor, wherein a computer program capable of running on the processor is stored on the memory, and when the processor runs the computer program, the steps of the above-mentioned logistics scheduling method are executed.

[0065] Based on the same inventive concept, the present invention also provides a computer-readable storage medium, which is a non-volatile storage medium or a non-transient storage medium, on which a computer program is stored, and when the computer program is run by a processor, the steps of the above-mentioned logistics scheduling method are executed.

[0066] Advantageous Effects

[0067] Compared with the prior art, the advantages of the present invention are as follows:

[0068] A logistics scheduling method, an electronic device and a storage medium provided by the present invention. The method first constructs an initial feasible solution based on the insertion cost and the regret value; then optimizes the initial feasible solution by using all the development strategies in the hyper-heuristic optimization pool to obtain a first feasible solution; in order to avoid falling into a local optimum, a second feasible solution is generated based on the first feasible solution through perturbation, and a development strategy is selected according to the adjustment probability of the development strategy to optimize and adjust the second feasible solution, reducing the number of development strategies while ensuring the optimization effect and improving the optimization and adjustment efficiency; finally, the perturbation and optimization and adjustment steps are repeated to output the optimal feasible solution, which can obtain the optimal feasible solution and greatly reduce the logistics scheduling cost; the present invention can obtain the corresponding optimal feasible solution according to the specific logistics distribution scenario, greatly improving the feasibility, adaptability and portability of the method. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only one embodiment of the present invention, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0070] Figure 1 is a schematic diagram of the logistics scheduling method in the embodiment of the present invention;

[0071] Figure 2 is a schematic diagram of the adjustment principle of the node reallocation strategy in the embodiment of the present invention, where Figure a shows the internal reallocation of the path, and Figure b shows the reallocation between different paths;

[0072] Figure 3 is a schematic diagram of the adjustment principle of the two-node exchange strategy in the embodiment of the present invention;

[0073] Figure 4 is a schematic diagram of the adjustment principle of the vertex-arc exchange strategy in the embodiment of the present invention;

[0074] Figure 5 is a schematic diagram of the adjustment principle of the two-arc exchange strategy in the embodiment of the present invention, where Figure a shows the internal exchange of the path, and Figure b shows the exchange between different paths;

[0075] Figure 6 is a schematic diagram of the adjustment principle of the three-arc exchange strategy in the embodiment of the present invention;

[0076] Figure 7 is a schematic diagram of the adjustment principle of the chain repositioning strategy in the embodiment of the present invention, where Figure a shows the internal repositioning of the path, and Figure b shows the repositioning between different paths;

[0077] Figure 8 is a schematic diagram of the adjustment principle of the cross-exchange strategy in the embodiment of the present invention;

[0078] Figure 9 This is the flowchart for adjusting feasible solutions using the development strategy in the embodiments of the present invention. Detailed implementation manners

[0079] The following clearly and completely describes the technical solutions in the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part rather than all of the embodiments of the present invention. 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.

[0080] The following uses specific embodiments to elaborate on the technical solutions of the present application in detail. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be repeated in some embodiments.

[0081] Embodiment 1

[0082] As Figure 1 shown, a logistics scheduling method provided in this embodiment includes the following steps:

[0083] Step 1: Obtain logistics orders within a certain time period.

[0084] In this embodiment, a certain time period can be several hours, one day, several days, or one week. The logistics orders include at least one order, and each order includes at least one customer point i and its location (x i , y i ) and the demand quantity q i of the goods. Therefore, the logistics orders within a certain time period can be expressed as V = {1, 2,..., i,..., N}, that is, there are N customer points involved in the logistics orders, and V 0 = {0, 1, 2,..., N}, where 0 represents the distribution center.

[0085] Step 2: Determine the number of transport vehicles and the number of routes according to the logistics orders.

[0086] In this embodiment, when the load limit Q of each transport vehicle is the same, the calculation formula for the number of transport vehicles M or the number of routes M is:

[0087]

[0088] where Q is the load limit of a single transport vehicle, is the rounding symbol.

[0089] When the load limits of each transport vehicle are different, the minimum M value that satisfies formula (2) is the number of transport vehicles or the number of routes, specifically:

[0090]

[0091] Among them, Q k is the load limit of the k-th transport vehicle.

[0092] In the present invention, M ≤ N, that is, the demand q of goods at a single customer point i cannot be transported by more than one vehicle (for example, the demand for goods at a certain customer point is different and is transported by 2, 3 or 4 vehicles), and the demand for goods at multiple customer points can be transported by one vehicle (for example, the demand for goods at 3 customer points can be transported by one vehicle). Each transport vehicle corresponds to a path. Each transport vehicle starts from the distribution center 0 and returns to the distribution center 0, providing one-way distribution services to customer points 1 to N. The M transport vehicles respectively execute the corresponding paths to complete the distribution task in the logistics order.

[0093] Step 3: Construct a preliminary feasible solution.

[0094] According to the position (x i , y i ) of each customer point i and the position (x 0 , y 0 ) of the distribution center, a feasible solution for the M transport vehicles to execute the logistics order is calculated. The specific implementation process is as follows:

[0095] Step 3.1: Define the customer set V and the path set R, where V = {1, 2,..., i,..., N}, R = {1, 2,..., k,..., M}, and initially assume that the M paths in the path set R are empty paths.

[0096] Step 3.2: Calculate the distance d i0 from each customer point i in the set V to the distribution center 0.

[0097] Step 3.3: Calculate the insertion cost of inserting customer point i into the k-th path, and obtain the insertion costs of inserting customer point i into the M paths. The specific formula is:

[0098]

[0099] Among them, is the insertion cost of inserting customer point i into the k-th path, is the transportation cost of the k-th path before inserting customer point i, is the transportation cost of the k-th path after inserting customer point i, |R k | is the number of customer points on the k-th path before inserting customer point i, |R k′ | is the number of customer points on the k-th path after inserting customer point i, is the distance between customer point j and customer point j + 1 on the k-th path, is the distance between the distribution center and the first customer point on the k-th path, is the distance between the |R k |th customer point and the distribution center on the k-th path, is the distance between the inserted customer point i and the distribution center on the k-th path.

[0100] Exemplarily, when the customer points on the k-th path without inserting customer point i are 20, 11, 24, 7, 35 in sequence, then |R k | = 5; when the customer points on the k-th path with inserting customer point i are 20, 11, 24, 7, 35, i in sequence, then |R k | = 6.

[0101] Step 3.4: Sort the M insertion costs corresponding to customer point i in ascending order, and calculate the regret value of customer point i. The specific formula is:

[0102]

[0103] where RV i is the regret value of customer point i, is the third smallest insertion cost among the M insertion costs corresponding to customer point i, is the smallest insertion cost among the M insertion costs corresponding to customer point i, The corresponding path is the path with the minimum insertion cost for customer point i. The regret value method based on insertion cost can combine forward-looking information and improve the short-sighted behavior of the greedy heuristic.

[0104] Step 3.5: Repeat steps 3.3 and 3.4 to obtain the regret value of each customer point i in set V.

[0105] Step 3.6: Extract the customer point i corresponding to the maximum regret value in set V max , and insert customer point i max to the end of its path with the minimum insertion cost.

[0106] Step 3.7: Delete customer point i from set V max .

[0107] Step 3.8: Judge whether set V is empty; when set V is not empty, repeat steps 3.3 to 3.8; when set V is empty, obtain a feasible solution for M transport vehicles to execute the logistics order. In the feasible solution, all customer points are inserted into the path with the minimum insertion cost.

[0108] Step 4: Construct a logistics scheduling optimization model according to the actual scenario and optimization objective. The specific formula is:

[0109]

[0110] Among them, C is the transportation cost; d ij is the distance between point i and point j; is a binary variable, taking values of 1 or 0. When , it means that in the k-th path, the transport vehicle travels from point i to point j; otherwise Q is the load limit of the transport vehicle. Equation (7) represents the minimization of the driving mileage of all transport vehicles (i.e., the optimization objective is to minimize the transportation cost). Equations (9) and (10) ensure that the transport vehicle leaves point i after arriving at point i, and each customer point i is only arrived at or visited once. Equation (11) ensures that each transport vehicle starts from the distribution center, visits customer points according to the path in the feasible solution, and then returns to the distribution center. Equation (12) ensures that the total distribution volume of the k-th transport vehicle is less than its load limit requirement Q (the load limit Q of each transport vehicle is the same). When the load limits of each transport vehicle are different,

[0111] Step 5: Construct a hyper-heuristic optimization pool, which contains various development strategies.

[0112] In this embodiment, 7 development strategies are designed in the hyper-heuristic optimization pool, specifically the node reallocation strategy, the two-node exchange strategy, the vertex-arc exchange strategy, the two-arc exchange strategy, the three-arc exchange strategy, the chain relocation strategy, and the cross-exchange strategy. Among them, five are development strategies based on the exchange idea, and two are development strategies based on the reallocation idea. These 7 development strategies have the ability to develop and improve the existing feasible solutions, that is, the ability to optimize the initial feasible solutions constructed in Step 3.

[0113] Node reallocation strategy: As Figure 2 shown, first delete a customer point i on a non-empty path k, and then insert the deleted customer point i after the adjacent position of the deleted customer point i on path k; this strategy can also be applied between paths, deleting a customer point on a path and then inserting the deleted customer point into another path.

[0114] Two-node exchange strategy: As Figure 3 shown, it is applied between paths, and an arbitrary customer point is selected from each of the two paths for exchange.

[0115] Vertex-arc exchange strategy: As Figure 4 shown, it is applied between paths, and an arbitrary sub-path within one path is exchanged with an arbitrary customer point of another path.

[0116] Two-arc exchange strategy: As Figure 5As shown, arbitrarily select two non - overlapping sub - paths within the path for exchange; meanwhile, this strategy can also be applied between paths, where one sub - path is arbitrarily selected from each of the two paths for exchange.

[0117] Three - arc exchange strategy: As Figure 6 shown, arbitrarily select three non - overlapping arc segments

[0118] Arc(i,i + 1), Arc(j,j + 1), Arc(k,k + 1) within the path, and change them to Arc(i,j), Arc(i + 1,k), Arc(j + 1,k + 1); this strategy involves reversing the direction of the sub - path, that is, the sub - path segment {1,...,i} is concatenated with the reversed path segment {i + 1,...,j} which is concatenated with the reversed path segment {j + 1,...,k} and then concatenated with the path segment {k + 1,...,end}.

[0119] The idea of the chain re - location strategy is borrowed from Or - opt, but it does not reverse the path direction. As Figure 7 shown, select a sub - path of length L within the path and insert it after an arbitrarily selected customer point; this strategy is also applied between paths, that is, select a sub - path of length L within one path and insert it after an arbitrarily selected customer point of another path; execute this strategy in the order of h = 4, 3, 2.

[0120] Cross - exchange strategy: As Figure 8 shown, it only acts between paths and exchanges one sub - link selected from different paths.

[0121] Step 6: Adjustment of the feasible solution.

[0122] Based on the logistics scheduling optimization models (7) - (12), use 7 development strategies in the hyper - heuristic optimization pool to adjust the feasible solution obtained in step 3, and finally output the optimal feasible solution. As Figure 9 shown, the specific implementation process is as follows:

[0123] Step 6.1: Use all the development strategies in the hyper - heuristic optimization pool to perform a full - strategy adjustment on the feasible solution for T rounds, obtaining the first feasible solution after T rounds and the normalized adjustment probability of each development strategy; where T≥1. The specific implementation process of step 6.1 is as follows:

[0124] Step 6.11: Randomly sort the 7 development strategies in the hyper - heuristic optimization pool; for example, the sorting order of the 7 development strategies is: node re - allocation strategy, two - node exchange strategy, vertex - arc exchange strategy, two - arc exchange strategy, three - arc exchange strategy, chain re - location strategy, and cross - exchange strategy.

[0125] Step 6.12: Adjust the feasible solutions using the development strategy in sorting to obtain the adjusted feasible solutions. Exemplarily, in the first adjustment, the node reallocation strategy is used to adjust the feasible solutions.

[0126] Step 6.13: When the adjusted feasible solutions satisfy the constraint conditions in the logistics scheduling optimization model (i.e., equations (9) to (12)), and the transportation cost of the adjusted feasible solutions is less than the transportation cost of the feasible solutions before adjustment (the transportation cost can be calculated according to equations (7) and (8)), replace the feasible solutions before adjustment with the adjusted feasible solutions, and record the replacement times of the current development strategy.

[0127] Exemplarily, when the feasible solutions adjusted by the node reallocation strategy satisfy the constraint conditions of equations (9) to (12), and the transportation cost of the feasible solutions adjusted by the node reallocation strategy is less than the transportation cost of the feasible solutions in Step 3, replace the feasible solutions in Step 3 with the feasible solutions adjusted by the node reallocation strategy, and record the replacement times of the node reallocation strategy as 1; if not replaced, the replacement times are 0.

[0128] Step 6.14: Enter the next development strategy, and repeat Steps 6.12 to 6.14 until all the development strategies in the hyper-heuristic optimization pool are completed, that is, the current loop is completed.

[0129] In the second adjustment, use the two-node exchange strategy for adjustment and replacement with the feasible solutions adjusted and replaced by the node reallocation strategy as the input; in the third adjustment, use the vertex-arc exchange strategy for adjustment and replacement with the feasible solutions adjusted and replaced by the two-node exchange strategy as the input; in the fourth adjustment, use the two-arc exchange strategy for adjustment and replacement with the feasible solutions adjusted and replaced by the vertex-arc exchange strategy as the input; in the fifth adjustment, use the three-arc exchange strategy for adjustment and replacement with the feasible solutions adjusted and replaced by the two-arc exchange strategy as the input; in the sixth adjustment, use the chain relocation strategy for adjustment and replacement with the feasible solutions adjusted and replaced by the three-arc exchange strategy as the input; in the seventh adjustment, use the cross-exchange strategy for adjustment and replacement with the feasible solutions adjusted and replaced by the chain relocation strategy as the input. After completing the adjustment and replacement of the 7 development strategies, the adjustment of the current loop is completed, and the replacement times of each development strategy are obtained.

[0130] Step 6.15: Add 1 to the loop count, and determine whether the loop count is equal to the set loop count; if not, repeat Steps 6.12 to 6.15 for the feasible solutions obtained in the current loop (exemplarily, the feasible solutions adjusted and replaced by the cross-exchange strategy); if so, obtain the first feasible solution of the current round and the total replacement times of each development strategy; where the initial value of the loop count is 0.

[0131] In this embodiment, the number of cycles is set to 100.

[0132] Step 6.16: Increment the round number by 1, and determine whether the round number is equal to the set round number T. If not, repeat steps 6.12 to 6.16 for the first feasible solution obtained in the current round. If so, obtain the first feasible solution after T rounds and the total replacement times of each development strategy; where the initial value of the round number is 0.

[0133] In this embodiment, T = 1.

[0134] In this embodiment, after the full strategy adjustment, the calculation formula for the normalized adjustment probability of each development strategy is:

[0135]

[0136] Among them, is the normalized adjustment probability of the i-th development strategy after the full strategy adjustment, is the corrected adjustment probability of the i-th development strategy after the full strategy adjustment, N T is the number of development strategies in the hyper-heuristic optimization pool (in this embodiment, N T is 7), is the total replacement times of the i-th development strategy after the full strategy adjustment, σ is the correction parameter (according to experience, σ is 0.15), and S is the number of cycles in each round. During the full strategy adjustment, the number of runs or uses of the 7 development strategies is equal. The correction parameter can prevent the situation where the replacement times of a certain development strategy is 0, resulting in a selection of an adjustment probability of 0 in subsequent hyper-heuristic optimization. According to the Select a development strategy during the first optimization adjustment.

[0137] Step 6.1 is essentially a process of finding a local optimal solution. To jump out of the local optimal solution, perturbation is performed through step 6.2 in order to obtain the global optimal solution.

[0138] Step 6.2: Randomly select a development strategy in the hyper-heuristic optimization pool to perturb the first feasible solution output by step 6.1 to obtain a second feasible solution. The specific implementation process of the perturbation is as follows:

[0139] Step 6.21: Calculate the transportation cost C of the first feasible solution output by step 6.1 according to the logistics scheduling optimization model (Equations (7) and (8)) 1 .

[0140] Step 6.22: Randomly select an integer between [M, 2M] and assign it to m.

[0141] Step 6.23: Randomly select a development strategy from the hyper-heuristic optimization pool to adjust the first feasible solution, obtain the adjusted feasible solution, and calculate the transportation cost of the adjusted feasible solution (Equations (7) and (8)).

[0142] Step 6.24: Repeat Step 6.23 for m - 1 times to obtain m adjusted feasible solutions and their transportation costs (i.e., obtain the neighborhood solutions of the first feasible solution).

[0143] Step 6.25: Select the feasible solutions with transportation costs less than (1 + dev)C 1 from the m adjusted feasible solutions (the selected feasible solutions are put into the neighborhood elite pool), where dev is the cost tolerance coefficient.

[0144] In this embodiment, dev = 0.5.

[0145] Step 6.26: Randomly select a solution from the feasible solutions with transportation costs less than (1 + dev)C 1 from the neighborhood elite pool as the second feasible solution.

[0146] The perturbation optimization aims to open up new optimization directions, and the obtained neighborhood elite solutions still need to be further optimized through Step 6.3.

[0147] Step 6.3: According to the normalized adjustment probability of each development strategy, select a development strategy from the hyper-heuristic optimization pool to optimize and adjust the second feasible solution output in Step 6.2, obtain the third feasible solution, and calculate the selection adjustment probability of each development strategy after the optimization adjustment.

[0148] In this embodiment, according to the normalized adjustment probability of each development strategy, the roulette wheel selection algorithm is used to select a development strategy from the hyper-heuristic optimization pool to optimize and adjust the second feasible solution. In Step 6.3, during the first optimization adjustment, according to select the development strategy.

[0149] The calculation formula for the selection adjustment probability of each development strategy after the optimization adjustment is:

[0150]

[0151] where, is the selection adjustment probability of the i-th development strategy during the (g + 1)-th optimization adjustment process, is the normalized adjustment probability of the i-th development strategy during the g-th optimization adjustment process, is the normalized adjustment probability of the i-th development strategy after the full strategy adjustment, a is the historical experience coefficient, is the corrected adjustment probability of the i-th development strategy during the g-th optimization adjustment process, and σ is the correction parameter. is the replacement times of the \(i\)th development strategy in the \(g\)th optimization and adjustment process, is the selection times of the \(i\)th development strategy in the \(g\)th optimization and adjustment process, \(N\) T is the number of development strategies in the hyper-heuristic optimization pool. The correction parameter can prevent the situation that the replacement times of a certain development strategy is 0, resulting in a probability of 0 being selected in the next optimization and adjustment.

[0152] Step 6.4: Based on the selection adjustment probability of each development strategy after optimization and adjustment, repeat Steps 6.2 and 6.3 for the third feasible solution until the third feasible solution has no replacement for \(G\) consecutive times, then output the optimal feasible solution. In the first optimization and adjustment after full strategy adjustment and perturbation, according to select the development strategy in the hyper-heuristic optimization pool for optimization and adjustment; when repeating Step 6.3 for the third feasible solution, according to select the development strategy in the hyper-heuristic optimization pool for optimization and adjustment.

[0153] In this embodiment, \(G = 3\).

[0154] Embodiment 2

[0155] To verify the superiority of the method of the present invention, this embodiment uses the large instance GWKC20 in the dataset solomon for testing. The logistics scheduling method proposed by the present invention runs on Matlab R2020b with an AMD Ryzen5 4600H CPU@3.00GHz, a running memory of 16GB, and an operating system of Win10 Home Chinese Edition. The specific results are shown in Table 1.

[0156] Table 1 Operation results of GWKC large instance

[0157]

[0158] In Table 1, the items in bold black represent the relatively better solutions. As shown in Table 1, columns 1 to 4 represent the basic information of each instance, column 5 stores the currently known optimal solution BKS, column 6 is the optimal solution proposed obtained by running the method of the present invention 10 times, and column 7 GAP is the performance gap index.

[0159]

[0160] In the above formula, GAP i is the performance gap index of the \(i\)th test instance, refers to the cost of the currently known optimal solution BKS on the \(i\)th test instance, refers to the cost of the method of the present invention on the \(i\)th test instance.

[0161] To fully verify the effectiveness of the method proposed by the present invention, compared with the optimal solution BKS obtained by comprehensive comparison of the most cutting-edge algorithms at present, the comparison results show that the method proposed by the present invention can obtain a better solution than the optimal solutions obtained by the current most cutting-edge algorithms, and has good application prospects. As shown in Table 1, the method proposed by the present invention finds better solutions than the optimal solutions obtained by comprehensive comparison of the most cutting-edge algorithms at present in nine instances.

[0162] The above-disclosed are only the specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or variations, which should all be covered within the protection scope of the present invention.

Claims

1. A logistics scheduling method, characterized in that, it includes the following steps: Obtain logistics orders within a certain period of time, where the logistics orders include at least one order, and each order includes at least one customer point i and its location (x i , y i ), and the demand quantity q of goods i ; Determine the number of transport vehicles M and the number of paths M according to the total demand of goods in the logistics order and the load limit of the transport vehicles; where M ≤ N, and N is the number of customer points in the logistics order; According to the position (x i , y i ) of each customer point i and the position (x 0 , y 0 ) of the distribution center, a feasible solution for M transport vehicles to execute the logistics order is calculated; Construct a logistics scheduling optimization model according to the actual scenario and optimization objectives; Construct a hyper-heuristic optimization pool, and the hyper-heuristic optimization pool contains a variety of development strategies; Based on the logistics scheduling optimization model, use the development strategies in the hyper-heuristic optimization pool to adjust the feasible solution, and output the optimal feasible solution; Among them, the specific implementation process of using the development strategies in the hyper-heuristic optimization pool to adjust the feasible solution is: Step 6.1: Use all the development strategies in the hyper-heuristic optimization pool to perform T-round full-strategy adjustments on the feasible solution to obtain the first feasible solution after T rounds and the normalized adjustment probability of each development strategy; where T ≥ 1; Step 6.2: Randomly select a development strategy in the hyper-heuristic optimization pool to perturb the first feasible solution to obtain a second feasible solution; Step 6.3: Select the development strategies in the hyper-heuristic optimization pool according to the normalized adjustment probability of each development strategy to optimize and adjust the second feasible solution to obtain a third feasible solution, and calculate the selection and adjustment probability of each development strategy after the optimization and adjustment; Step 6.4: Based on the selection and adjustment probability of each development strategy after the optimization and adjustment, repeat steps 6.2 and 6.3 for the third feasible solution until the third feasible solution has no replacement for G consecutive times, and then output the optimal feasible solution.

2. The logistics scheduling method according to claim 1, characterized in that, When the load limit of each transport vehicle is the same, the calculation formula for the number of transport vehicles M or the number of paths M is: where Q is the load limit of a single transport vehicle, is the rounding symbol; When the load limit of each transport vehicle is different, the minimum M value that satisfies the following formula is the number of transport vehicles or the number of paths, and the specific formula is: Among them, Q k is the load limit of the k-th transport vehicle.

3. The logistics scheduling method according to claim 1, characterized in that, The specific implementation process of obtaining a feasible solution for M transport vehicles to execute the logistics order is: Step 3.1: Define a customer set V and a path set R, where V = {1, 2,..., i,..., N}, R = {1, 2,..., k,..., M}, and assume that the M paths in the path set R are empty paths initially; Step 3.2: Calculate the distance d from each customer point i in the set V to the distribution center i0 ; Step 3.3: Calculate the insertion cost of inserting customer point i into the k-th path to obtain the insertion costs of inserting customer point i into M paths. The specific formula for the insertion cost is: Among them, is the insertion cost of inserting customer point i into the k-th path, is the transportation cost of the k-th path without inserting customer point i, is the transportation cost of the k-th path when inserting customer point i, |R k | is the number of customer points on the k-th path without inserting customer point i, |R k ′| is the number of customer points on the k-th path when inserting customer point i, is the distance between customer points j and j + 1 on the k-th path, is the distance between the distribution center and the first customer point on the k-th path, is the |R on the k-th path k | is the distance between the last customer point and the distribution center on the k-th path, is the distance between the inserted customer point i and the distribution center on the k-th path; Step 3.4: Sort the M insertion costs corresponding to customer point i in ascending order and calculate the regret value of customer point i. The specific formula is: Among them, RV i is the regret value of customer point i, is the third-ranked insertion cost among the M insertion costs corresponding to customer point i, is the first-ranked insertion cost among the M insertion costs corresponding to customer point i, The corresponding path is the path with the minimum insertion cost for customer point i; Step 3.5: Repeat steps 3.3 and 3.4 to obtain the regret value of each customer point i in the set V; Step 3.6: Extract the customer point i corresponding to the maximum regret value in the set V max , and insert the customer point i max at the end of the path with the minimum insertion cost; Step 3.7: Delete customer point i from set V max ; Step 3.8: Determine whether the set V is empty; when the set V is not empty, repeat steps 3.3 to 3.8; when the set V is empty, obtain a feasible solution for M transport vehicles to execute the logistics order.

4. The logistics scheduling method according to claim 1, characterized in that, The specific expression of the logistics scheduling optimization model is: Among them, V 0 = {0, 1, 2, …, N}, V = {1, 2, …, i, …, N}, R = {1, 2, …, k, …, M}, V is the set of customers, V 0 is the set of customers and the distribution center, i, j = 0 represents the distribution center, j ≠ i, R is the set of paths; C is the transportation cost; d ij is the distance between point i and point j; is a binary variable, taking values of 1 or 0. When , it means that in the k-th path, the transport vehicle travels from point i to point j; Q k is the load limit of the k-th transport vehicle.

5. The logistics scheduling method according to claim 1, characterized in that, the hyper-heuristic optimization pool includes a node reallocation strategy, a two-node exchange strategy, a vertex-arc exchange strategy, a two-arc exchange strategy, a three-arc exchange strategy, a chain repositioning strategy, and a cross-exchange strategy.

6. The logistics scheduling method according to claim 1, characterized in that, in step 6.1, the specific implementation process of the full-strategy adjustment in the T-th round is as follows: Step 6.11: Randomly sort all the exploration strategies in the hyper-heuristic optimization pool; Step 6.12: Adjust the feasible solution using the exploration strategy in the sorting to obtain an adjusted feasible solution; Step 6.13: When the adjusted feasible solution satisfies the constraint conditions in the logistics scheduling optimization model and the transportation cost of the adjusted feasible solution is less than the transportation cost of the feasible solution before adjustment, replace the feasible solution before adjustment with the adjusted feasible solution, and record the replacement times of the current exploration strategy; Step 6.14: Proceed to the next exploration strategy, and repeat steps 6.12 to 6.14 until all the exploration strategies in the hyper-heuristic optimization pool are completed, that is, the current loop is completed; Step 6.15: Add 1 to the loop count, and determine whether the loop count is equal to the set loop count; if not, repeat steps 6.12 to 6.15 for the feasible solution obtained in the current loop; if so, obtain the first feasible solution in the current round and the total replacement times of each exploration strategy; where the initial value of the loop count is 0; Step 6.16: Add 1 to the round count, and determine whether the round count is equal to the set round count T; if not, repeat steps 6.12 to 6.16 for the first feasible solution obtained in the current round; if so, obtain the first feasible solution after the T-th round and the total replacement times of each exploration strategy; where the initial value of the round count is 0.

7. The logistics scheduling method according to claim 6, characterized in that, after the full-strategy adjustment, calculate its normalized adjustment probability according to the total replacement times of each exploration strategy, and the specific calculation formula is: Among them, is the normalized adjustment probability of the i-th development strategy after the full strategy adjustment, is the corrected adjustment probability of the i-th development strategy after the full strategy adjustment, N T is the number of development strategies in the hyper-heuristic optimization pool, is the total replacement times of the i-th development strategy after the full strategy adjustment, σ is the correction parameter, and S is the number of loops in each round.

8. The logistics scheduling method according to claim 1, characterized in that, in step 6.2, the specific implementation process of the perturbation is as follows: Step 6.21: Calculate the transportation cost C of the first feasible solution according to the logistics scheduling optimization model 1 ; Step 6.22: Randomly select an integer between [M, 2M] and assign it to m; Step 6.23: Randomly select an exploration strategy from the hyper-heuristic optimization pool to adjust the first feasible solution to obtain an adjusted feasible solution, and calculate the transportation cost of the adjusted feasible solution; Step 6.24: Repeat step 6.23 m - 1 times to obtain m adjusted feasible solutions and their transportation costs; Step 6.25: Select feasible solutions with transportation costs less than (1 + dev)C from the m adjusted feasible solutions, where dev is the cost tolerance factor; 1 ​ Step 6.26: Randomly select a solution from the feasible solutions with transportation cost less than (1 + dev)C 1 as the second feasible solution.

9. The logistics scheduling method according to claim 1, characterized in that, in step 6.3, select an exploration strategy in the hyper-heuristic optimization pool according to the adjustment probability of each exploration strategy and use the roulette wheel selection algorithm to optimize and adjust the second feasible solution.

10. The logistics scheduling method according to claim 1, characterized in that, in step 6.3, the calculation formula for the selection and adjustment probability of each exploration strategy after optimization and adjustment is: Among them, is the selection adjustment probability of the \(i\)-th development strategy in the \((g + 1)\)-th optimization adjustment process, is the normalized adjustment probability of the \(i\)-th development strategy in the \(g\)-th optimization adjustment process, is the normalized adjustment probability of the \(i\)-th development strategy after the full strategy adjustment, \(a\) is the historical experience coefficient, is the corrected adjustment probability of the \(i\)-th development strategy in the \(g\)-th optimization adjustment process, \(\sigma\) is the correction parameter, is the replacement times of the \(i\)-th development strategy in the \(g\)-th optimization adjustment process, is the selection times of the \(i\)-th development strategy in the \(g\)-th optimization adjustment process, \(N\) T is the number of development strategies in the hyper-heuristic optimization pool.

11. An electronic device, comprising a memory and a processor, wherein a computer program capable of running on the processor is stored on the memory, characterized in that, when the processor runs the computer program, it executes the steps of the logistics scheduling method according to any one of claims 1 to 10.

12. A computer-readable storage medium, which is a non-volatile storage medium or a non-transitory storage medium, on which a computer program is stored, characterized in that, when the computer program is run by a processor, it executes the steps of the logistics scheduling method according to any one of claims 1 to 10.

Images