Post-disaster multi-center material distribution method and system based on double repair mechanism

Abstract

The invention discloses a post-disaster multi-center material distribution method and system based on a double repair mechanism, and the method comprises the steps: inputting road network basic data, vehicle data, distribution point data and disaster-affected point data into a preset material distribution path planning model, and obtaining a material distribution strategy generated by the material distribution path planning model, wherein the material distribution path planning model takes the minimization of the total time of a vehicle from each distribution point to all disaster-affected points as an optimization problem, the optimization problem is solved through a memetic algorithm to obtain an optimal solution of path planning, and the memetic algorithm is used for repairing individuals with vehicle constraint violation in the initial population through a double repair mechanism. According to the post-disaster multi-center material distribution method and system based on the double-repair mechanism, the double-repair mechanism is provided, infeasible individuals in a population are repaired as much as possible, search of an algorithm can be carried out towards a feasible area, and an optimal material distribution strategy is obtained.

Description

Technical Field

[0001] The present invention relates to a post-disaster multi-center material distribution method and system based on a double repair mechanism, belonging to the technical field of material distribution. Background Art

[0002] The post-disaster material distribution problem is a specific application of the vehicle routing problem, in which the time requirements of each distribution point must be taken into account. Therefore, the post-disaster material distribution problem is usually described as a time-accumulated CVRP, whose goal is to minimize the sum of the time to reach all distribution points, which is different from the traditional CVRP, which aims to minimize the total distance traveled by the vehicle.

[0003] In the actual post-disaster material distribution problem, the management department will equip multiple material distribution centers according to the situation in the disaster area, so that the material distribution will be more efficient. This problem belongs to multi-center CCVRP, which is a typical combinatorial optimization problem. Compared with CVRP, the constraint on the number of vehicles in MDCCVRP is not only reflected in the total number of vehicles, but also in the maximum number of vehicles stationed at each distribution center.

[0004] It can be seen that in order to overcome the constraint problem on the number of vehicles, a post-disaster multi-center material distribution method and system based on a double repair mechanism is urgently needed. Summary of the invention

[0005] The purpose of the present invention is to overcome the shortcomings of the prior art and provide a post-disaster multi-center material distribution method and system based on a double-patch mechanism. A double-patch mechanism is proposed to patch up the infeasible individuals in the population as much as possible, so that the algorithm's search can be carried out towards the feasible area to obtain the optimal material distribution strategy.

[0006] To achieve the above object, the present invention is implemented by adopting the following technical solutions: In a first aspect, the present invention provides a post-disaster multi-center material distribution method based on a dual-repair mechanism, comprising: Obtain basic road network data, vehicle data, distribution point data and disaster-affected point data in the disaster-affected area; The basic road network data, vehicle data, distribution point data and disaster-affected point data are input into a preset material distribution path planning model to obtain a material distribution strategy generated by the material distribution path planning model, wherein: The material distribution path planning model takes minimizing the total time for vehicles to reach all disaster-stricken points from various distribution points as an optimization problem, and solves the optimization problem through a memetic algorithm to obtain the optimal solution for path planning. The memetic algorithm repairs the individuals that violate vehicle constraints in the initial population and the evolution process through a double repair mechanism.

[0007] Furthermore, the preset objective function and constraint conditions of the material distribution path planning model include: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) Among them, G = (V, E) is the disaster-stricken area, and the vertex set V includes V C and V D Two subsets, V C ={v1,v2,v3,…,vN} represents the set of disaster-affected points, V D ={vN+1,vN+2,vN+3,…,vN+M} represents the set of distribution centers, E is the edge set, representing the road, (vi,vj)∈E(vi,vj∈V) is the edge between any two vertices; the fleet K is stationed at V D At each parking lot, the vehicle capacity is Q, and the center d∈ V D The maximum number of vehicles stationed at is Kd; let the time for vehicle k to arrive at distribution point i be , the material demand of distribution point i is qi; if distribution point i is not served by vehicle k, then is 0; Formula (1) is the objective function; Formula (2) ensures that the vehicle leaves immediately after arriving at the disaster site; Formula (3) stipulates that each disaster site only delivers materials once; Formula (4) ensures that the vehicle eventually returns to the same distribution center, and Formula (5) constrains each vehicle to provide at most one service; Formulas (6) and (7) constrain the total number of vehicles and the maximum number of vehicles at each distribution center; Formula (8) constrains that the vehicle cannot exceed the maximum load Q; Formula (9) gives a method for calculating the time it takes for the vehicle to reach the disaster site, and avoids subpaths by using a sufficiently large number P; Formulas (10)-(11) define the vehicle arrival time; Formulas (12)-(13) define the binary variable Respectively indicate whether vehicle k starts from the distribution center d and whether it passes through the edge (i, j); The solution of the objective function is expressed as: (14) Among them, |RD| represents the number of paths starting from parking lot d, and the solution of the problem is the number of paths from all parking lots. composition.

[0008] Furthermore, the method of solving the optimization problem by using a memetic algorithm to obtain the optimal solution for path planning includes: Generate an initial population, and use a double patching mechanism to patch individuals in the population that violate vehicle constraints; Save the best feasible solution in the population to S; Determine whether the algorithm has reached the termination condition. If so, take S as the optimal solution for path planning. Otherwise, initialize the number of iterations and execute the iteration steps until the preset maximum number of iterations is reached. Then, use the double patching mechanism to patch the individuals in the population that violate vehicle constraints and update S. The iterative steps include: Use the crossover operator to generate sub-individuals, and perform local search on the generated sub-individuals; Determine whether the child individual is a non-clone individual of the population. If so, add the child individual to the population, sort the individuals in the population, and remove the last individual.

[0009] Furthermore, the initial population is generated by a multi-strategy initial population generation method, which includes a clustering-based initialization method, an optimal insertion method-based initialization method, and a random method, wherein: The number of individuals generated by the clustering-based initialization method is 1, the number of individuals generated by the optimal insertion method is 1, and the number of individuals generated by the random method is the population size minus 2.

[0010] Furthermore, for individuals in the initial population that violate vehicle constraints, a double patching mechanism is used to patch them, including: Check whether the number of vehicles in the current individual exceeds the preset maximum number of vehicles. If so, perform the repair step; if not, the algorithm ends; Create a copy of the current individual and perform RepTotV patching on the copy; Check whether the total number of vehicles in the patched copy is within the allowed range. If so, assign the copy to the original individual and accept the patching result. If not, perform further patching. If there is a vehicle crossing the distribution center in the patched current individual, implement RepDepV patching on the current individual.

[0011] Furthermore, the RepTotV patch includes: Step 1 OldInd←ind; Step 2 DRS←{}; Step 3: Keep the paths with the largest |K| loads in ind and move the remaining paths to DRS; Step 4 ICS[|K|]←{{},{},…,{}}; Step 5: For each path R in DRS Step 5.1 Sort the disaster-affected points in R from large to small according to the demand for supplies; Step 5.2 For each disaster point R[i] in R (i=1→|R|) If there is a path ind[j] in ind, so that after inserting the disaster point R[i], the capacity constraint Q can still be satisfied, then ICS[j]←ICS[j]∪R[i]. Otherwise, return OldInd and end the repair; Step 6 For each path ind[j] in ind (j=1→|K|) Use the best insertion method to insert all the disaster-affected points in ICS[j] into the path ind[j]; Step 7 Return to ind.

[0012] Furthermore, the RepDepV patching includes: Step 1: Define an array cnt[|VD|] with a length equal to the number of distribution centers to count the number of vehicles used at each distribution center, and initialize all elements to 0; Step 2 For each path R in ind Step 2.1 minCst←InfValue, where InfValue is the maximum value in the real number range; Step 2.2 For the distribution center d∈VD with insufficient number of vehicles (Kd>cnt[d]) Step 2.2.1 Calculate the additional vehicle travel time time(R,d) generated by changing the distribution center of route R to d. The calculation formula is as follows: (15) in, It represents the time from the distribution center to the first distribution point in R; Step 2.2.2 If minCst>time(R, d), then minCst←time(R, d), d*←d; Step 2.3 Replace the distribution center of R with d*; Step 2.4 cnt[d*]←cnt[d*]+ 1; Step 3: Use two traditional local search operators, single-point insertion and exchange, to perform neighborhood search on ind; Step 4 Return to ind.

[0013] In a second aspect, the present invention provides a post-disaster multi-center material distribution system based on a dual repair mechanism, characterized in that the system comprises: Acquisition module: used to obtain basic road network data, vehicle data, distribution point data and disaster-affected point data in the disaster-affected area; Path planning module: used to input the basic road network data, vehicle data, distribution point data and disaster-affected point data into a preset material distribution path planning model to obtain the material distribution strategy generated by the material distribution path planning model, wherein: The material distribution path planning model takes minimizing the total time for vehicles to reach all disaster-stricken points from various distribution points as an optimization problem, and solves the optimization problem through a memetic algorithm to obtain the optimal solution for path planning. The memetic algorithm repairs the individuals that violate vehicle constraints in the initial population and the evolution process through a double repair mechanism.

[0014] In a third aspect, the present invention provides a computer device, including a processor and a storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the method according to the first aspect.

[0015] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that when the program is executed by a processor, the steps of the method described in the first aspect are implemented.

[0016] Compared with the prior art, the present invention has the following beneficial effects: This post-disaster multi-center material distribution method and system based on the double-patch mechanism proposes a double-patch mechanism to repair the infeasible individuals in the population as much as possible, so that the algorithm's search can be carried out towards the feasible area to obtain the optimal material distribution strategy. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 It is a flow chart of the post-disaster multi-center material distribution method based on the dual repair mechanism; Figure 2 It is a flowchart of the meme algorithm; Figure 3 It is a convergence curve diagram of a typical example. DETAILED DESCRIPTION

[0018] The technical solution of the present invention is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations on the technical solution of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0019] The term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship. Embodiment 1

[0020] Figure 1 This is a flowchart of a post-disaster multi-center material distribution method based on a dual-patch mechanism in Embodiment 1 of the present invention. The post-disaster multi-center material distribution method based on a dual-patch mechanism provided in this embodiment can be applied to a terminal and can be executed by a post-disaster multi-center material distribution device based on a dual-patch mechanism. The device can be implemented by software and / or hardware, and the device can be integrated in a terminal, for example: any smart phone, tablet computer or computer device with communication function. See Figure 1 The method of this implementation specifically includes the following steps: Obtain basic road network data, vehicle data, distribution point data and disaster-affected point data in the disaster-affected area; The basic road network data, vehicle data, distribution point data and disaster-affected point data are input into a preset material distribution path planning model to obtain a material distribution strategy generated by the material distribution path planning model, wherein: The material distribution path planning model takes minimizing the total time for vehicles to arrive at all disaster-stricken points from each distribution point as an optimization problem, and solves the optimization problem through a memetic algorithm to obtain the optimal solution for path planning. The memetic algorithm repairs individuals in the initial population that violate vehicle constraints through a double repair mechanism.

[0021] MDCCVRP is a complex problem that belongs to NP-hard. The solution space grows explosively as the scale of the problem increases. For the convenience of algorithm description, the problem is first described symbolically. Assume that G=(V, E) is the disaster-stricken area, where the vertex set V includes two subsets VC and VD, VC={v1,v2,v3,…,vN} represents the set of disaster-stricken points, VD={vN+1,v N+2,v N+3,…,v N+M} represents the set of distribution centers, E is the edge set, representing the road, (vi,vj) ∈E(vi,vj∈V) is the edge between any two vertices. The fleet K is stationed at each parking lot of VD, with a vehicle capacity of Q, and the maximum number of stationed vehicles at the center d∈VD is Kd. Let the time for vehicle k to arrive at distribution point i be , the material demand of distribution point i is qi. If distribution point i is not served by vehicle k, then is 0. The vehicle starts from each point in VD, serves all points in VC once and only once, and then returns to the starting point in each corresponding VD, so that the total time for the material distribution vehicle to reach all garbage collection points is minimized under the premise of meeting the maximum load Q of the vehicle, that is: (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) Among them, formula (1) is the objective function; formula (2) ensures that the vehicle leaves immediately after arriving at the disaster site; formula (3) stipulates that each disaster site only delivers materials once; formula (4) ensures that the vehicle eventually returns to the same center point, and formula (5) constrains each vehicle to provide at most one service; formulas (6) and (7) constrain the total number of vehicles and the maximum number of vehicles at each center point; formula (8) constrains that the vehicle cannot exceed the maximum load Q; formula (9) gives a method for calculating the time it takes for the vehicle to reach the disaster site, and avoids sub-paths by using a sufficiently large number P; formulas (10)-(11) define the vehicle arrival time; formulas (12)-(13) define the binary variable They respectively indicate whether vehicle k departs from the distribution center d and whether it passes through the edge (i, j).

[0022] The solution of MDCCVRP can be expressed as: (14) Among them, |RD| represents the number of paths starting from parking lot d, and the solution to the problem consists of all parking lot paths.

[0023] Aiming at the characteristics of MDCCVRP, the present invention proposes an MA algorithm (MADRM) based on a double patching mechanism. The algorithm is punished by the initial population and undergoes several generations of evolutionary iterations to obtain a satisfactory solution. The evolutionary process mainly consists of selection and crossover. In order to further improve the quality of the solution, a local search is performed on the current solution. The specific algorithm process is described as follows.

[0024] Population initialization For algorithms based on iterative frameworks, the quality of the initial solution is of great significance. Moreover, MADRM is a population-based intelligent search algorithm, and the diversity of the population is particularly important for expanding the search space. In view of this, the present invention constructs a multi-strategy initial population generation method.

[0025] Clustering-based initialization method First, the disaster-stricken points in VC are distributed to the distribution centers in VD according to the principle of proximity. In this way, VC is divided around the center point. For each disaster-stricken point at the center point, the disaster-stricken points are sorted from near to far according to the distance to the center point. The vehicles deliver materials to the disaster-stricken points in the sorted order. If the vehicle capacity is full, the empty vehicle will deliver materials. The rules are repeated until all disaster-stricken points are delivered. This method generates 1 individual.

[0026] Initialization method based on optimal insertion method This method selects the best disaster point from the disaster points that have not been delivered materials and inserts it into the existing path under the condition of satisfying the capacity Q constraint. The so-called best point is the point at which the cumulative time added to the path is the smallest. This process continues until all disaster points have materials delivered. Due to the determinism of this method, it generates 1 individual.

[0027] Randomness Methods This method is similar to the deterministic method, except that the disaster-affected points are selected randomly each time. Due to the randomness, this method can generate multiple different individuals. In this way, the diversity of the population can be ensured. Here, the total number of individuals generated by this method is popsize – 2, where popsize is the subpopulation size.

[0028] After constructing the initial population Pop by the above three strategies, the target value and fitness value of the individuals are calculated to obtain the initial optimal feasible solution S.

[0029] Double patching mechanism (DRM) Since the path can satisfy the vehicle load constraint Q in the initial population and the subsequent evolution process, but the two constraints of the number of vehicles cannot be guaranteed, the present invention proposes a dual patching mechanism, which is mainly composed of two patching operators, as follows: Repair operator for total number of vehicles (RepTotV) If the number of vehicles used in the current individual ind exceeds the total number of vehicles |K|, then the following repairs are performed: Step 1 OldInd←ind; Step 2 DRS←{}; Step 3: Keep the paths with the largest |K| loads in ind and move the remaining paths to DRS; Step 4 ICS[|K|]←{{},{},…,{}}; Step 5: For each path R in DRS Step 5.1 Sort the disaster-affected points in R from large to small according to the demand for supplies; Step 5.2 For each disaster point R[i] in R (i=1→|R|) If there is a path ind[j] in ind, so that after inserting the disaster point R[i], the capacity constraint Q can still be satisfied, then ICS[j]←ICS[j]∪R[i]. Otherwise, return OldInd and end the repair; Step 6 For each path ind[j] in ind (j=1→|K|) The optimal insertion method is used to insert all disaster-affected points in ICS[j] into the path ind[j].

[0030] Step 7 Return to ind.

[0031] Repair operator for the number of vehicles at the distribution center (RepDepV) The RepDepV operator mainly relocates the distribution center points for all paths in individual ind to ensure that the number of vehicles at each center point does not exceed the limit. The specific steps are described as follows: Step 1: Define an array cnt[|VD|] with a length equal to the number of distribution centers to count the number of vehicles used at each distribution center, and initialize all elements to 0; Step 2 For each path R in ind Step 2.1 minCst←InfValue, where InfValue is the maximum value in the real number range; Step 2.2 For the distribution center d∈VD with insufficient number of vehicles (Kd>cnt[d]) Step 2.2.1 Calculate the additional vehicle travel time time(R,d) generated by changing the distribution center of route R to d. The calculation formula is as follows: (15) in, Represents the time from the distribution center to the first distribution point in R.

[0032] Step 2.2.2 If minCst>time(R, d), then minCst←time(R, d), d*←d; Step 2.3 Replace the distribution center of R with d*; Step 2.4 cnt[d*]←cnt[d*]+ 1; Step 3: Use two traditional local search operators, Single Insertion (SI) and Swap, to perform neighborhood search on ind.

[0033] Step 4 Return to ind.

[0034] Double patching mechanism The process of the double patching mechanism is as follows: Step 1 If the number of vehicles in the current individual ind is out of bounds (greater than |K|), go to step 2, otherwise go to step 6; Step 2 ind′←ind; Step 3: Perform RepTotV patching on ind′; Step 4: If the total number of vehicles in the patched ind′ is within the range |K|, execute step 5, otherwise go to step 7; Step 5: ind←ind′; Step 6 If there is a vehicle crossing the boundary at the distribution center in ind, perform RepDepV patching on ind; Step 7 Return to ind.

[0035] 3.3 Crossover and Local Search Operators (1) Crossover operator A path-based crossover operator is designed in MADRM. For parent individuals p1 and p2, two paths R1 and R2 are randomly selected from them respectively. R1 in p1 is replaced by R2 to generate a child individual child. In this way, if there are redundant material distribution points in child, they will be deleted. On the other hand, if there are missing distribution points in child, the best insertion method is used to fill them.

[0036] (2) Local search process For MA, its characteristic is that it has good local search capability because it includes a local search process. In the MADRM algorithm of the present invention, the local search process is mainly composed of five traditional local search operators, including single point insertion SI, double point insertion (Double Insertion, DI), Swap, 3-Opt and Arc-node Exchange (ANE). The five local search operators are called in a random order. When the current operator cannot improve the quality of ind, the next local search operator is started. Continue this process until the local search process is stopped when ind cannot be improved.

[0037] MADRM overall process Based on the design of the above parts, the steps of MADRM are as follows: Step 1: Use the three methods in 3.1 to generate an initial population Pop of size psize; Step 2: For individuals in Pop that violate vehicle constraints, a double patching mechanism is used to patch them; Step 3 Save the best feasible solution in Pop to S; Step 4 c←1; Step 5: Use the crossover operator in 3.3 to generate child individuals and perform local search on them; Step 6: If child is a non-clone of Pop, go to step 7; otherwise, go to step 9. Step 7 Pop←Pop∪{child}; Step 8: Sort the individuals in Pop and remove the last individual to maintain the size of psize; Step 9 c←c+1; Step 10: If c≤cycs, jump to step 5; Step 11: If the algorithm termination condition is not met, jump to step 4; Step 12: For individuals in Pop that have vehicle constraint violations, a double patching mechanism is used to patch them; Step 13: Save the best feasible solution in the current Pop to S′; Step 14 If S′ is better than S, then update S with it, that is, S←S′; Step 15 Return to S.

[0038] The algorithm flow of MADRM is as follows Figure 2 shown.

[0039] In order to verify the effectiveness of the MADRM method, the present invention selects the most representative methods POPMUSIC (E.Lalla-Ruiz and S. Voß, “A popmusic approach for the multi-depot cumulative capacitated vehicle routing problem,” Optimization Letters, vol. 14, no. 3, pp. 671–691, 2020.), VNS (500000) and PLS (X. Wang, T.-M. Choi, Z. Li, and S.Shao, “An effective local search algorithm for the multidepot cumulative capacitated vehicle routing problem,” IEEE Transactions on Systems, Man, andCybernetics: Systems, vol. 50, no. 12, pp. 4948–4958, 2020.). The test data uses the current popular test sets pr and p for the MDCCVRP problem, where the pr test set contains 10 test samples, and the number of disaster points ranges from 48 to 288. The p test set contains 23 test samples, and the number of disaster-affected points ranges from 50 to 360. The scale of disaster-affected points in the two test sets ranges from small to large, and the number of distribution centers is set according to the number and distribution of disaster-affected points.

[0040] MADRM is programmed in C++ and runs on Intel(R) Xeon(R) Gold 5218 2.30 GHz. After a thorough parameter adjustment process, the two important parameters, popsize and the number of inner loops, cycs, are set to 15 and popsize·min{5, |VC| / 20}, respectively, where min{5, |VC| / 20} is the smaller of 5 and |VC| / 20. In addition, the algorithm's termination condition is controlled by runtime. Since PLS is the most representative algorithm for considering the two constraints of the number of vehicles in MDCCVRP, in order to ensure fairness in comparison, the MADRM runtime of the present invention uses a runtime equivalent to it, which is mainly converted by the CPU main frequency of the machine. Since the CPU main frequency of the PLS running machine is 2.6GHz, the single running time of MADRM on the test sample is 2.6 / 2.3 times that of PLS. MADRM runs independently 30 times on each example, and a t-test with a significance level of 0.05 is performed at the same time.

[0041] Based on the above experimental settings, the experimental results of MADRM and other compared algorithms are shown in Tables 1 and 2, where Best, Avg and Time represent the best value, average value and running time (seconds). Mean represents the mean value on the entire test set, and BCW represents the number of cases where MADRM and the corresponding algorithm are significantly better (+), comparable and significantly worse (-) under the t-test. BKS is the optimal solution so far. If it is refreshed by MADRM, the corresponding result is marked with "†". If the result of each algorithm is the minimum value, it is displayed in bold.

[0042] Table 1 Results of each algorithm on the pr test set

[0043] From the results shown in Table 1, it can be seen that MADRM updated the optimal solution records of 6 examples in the entire pr test set, and reached the minimum value in the remaining 4 examples. For other algorithms, the number of test samples that reached the optimal solution record is: POPMUSIC: 0, VNS (500000): 1, PLS: 4. From the perspective of optimal performance, MADRM is better than POPMUSIC, VNS (500000) and PLS in the pr test set. From the perspective of average performance, the test set mean of MADRM is smaller than that of other algorithms, MADRM: 8073.17, PLS: 8168.16, VNS (500000): 8991.16, POPMUSIC: 22617.16. In all examples, the t-test shows that MADRM is significantly better than POPMUSIC and VNS (500000). For PLS, MADRM is significantly better than PLS in 8 cases, and there is no significant difference only in 2 cases.

[0044] Table 2 Results of each algorithm on the p test set

[0045] From the results in Table 2, we can see that in the 23 examples in the test set p, MADRM refreshed the optimal solution records of 18 test examples, reached the optimal solution record in 4 examples, and only the value on p04 was slightly larger than the optimal value of VNS (500000), which were 2627.92 and 2618.44 respectively. The number of examples in which POPMUSIC, VNS (500000) and PLS achieved the optimal solution record was 0, 4, and 4 respectively, and the total number of optimal solution records achieved by the three algorithms was 5. In terms of average performance, MADRM achieved the minimum value in 22 of the 23 examples, and only the minimum value on p04 was achieved by PLS. The t-test test shows that MADRM is significantly better than other algorithms in almost all examples.

[0046] From the analysis of the above results, it can be concluded that the MADRM proposed in the present invention has an obvious optimization effect in solving the problem of multi-center material distribution after disasters.

[0047] In order to examine the contribution of the double patching mechanism (DRM) to the algorithm, we remove it from MADRM and record the resulting algorithm as MA*. The patching operators RepTotV and RepDepV are only used when the population is initialized, and only feasible solutions are retained in the population. The experimental results are shown in Table 3.

[0048] Table 3 Comparison results between MADRM and MA*

[0049] From the results shown in Table 3, we know that MADRM achieved 9 and 20 minimum values ​​on the two test sets, accounting for the vast majority. The minimum values ​​achieved by MA* were 4 and 7, which is quite different from MADRM. From the t-test, we know that MADRM is significantly better than MA* in 8 and 14 test cases, and is only weaker than the latter in 4 cases. It can be seen that the dual repair mechanism (DRM) is of great significance to MADRM. For a more intuitive comparison, the convergence curves of the two algorithms on 8 representative cases are given here, as shown in the figure. Figure 3 As shown in the figure, the horizontal axis represents the algorithm running time, and the vertical axis represents the target value.

[0050] In the curve graph, in 6 cases (pr01, pr04, pr07, p10, p08 and p23), MADRM converges quickly soon after the algorithm starts and is always below the corresponding curve of MA*. Only in the case of p01, the MADRM convergence curve is above the corresponding curve of MA*, and in p15, the two curves overlap. These are consistent with the results in Table 3. Embodiment 2

[0051] A post-disaster multi-center material distribution system based on a dual repair mechanism, characterized in that the system comprises: Acquisition module: used to obtain basic road network data, vehicle data, distribution point data and disaster-affected point data in the disaster-affected area; Path planning module: used to input the basic road network data, vehicle data, distribution point data and disaster-affected point data into a preset material distribution path planning model to obtain the material distribution strategy generated by the material distribution path planning model, wherein: The material distribution path planning model takes minimizing the total time for vehicles to arrive at all disaster-stricken points from each distribution point as an optimization problem, and solves the optimization problem through a memetic algorithm to obtain the optimal solution for path planning. The memetic algorithm repairs individuals in the initial population that violate vehicle constraints through a double repair mechanism.

[0052] The post-disaster multi-center material distribution system based on the dual-patch mechanism provided in the embodiment of the present invention can execute the post-disaster multi-center material distribution method based on the dual-patch mechanism provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method. Embodiment 3

[0053] An embodiment of the present invention further provides a computer device, including a processor and a storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to perform the steps of the following method: Obtain basic road network data, vehicle data, distribution point data and disaster-affected point data in the disaster-affected area; The basic road network data, vehicle data, distribution point data and disaster-affected point data are input into a preset material distribution path planning model to obtain a material distribution strategy generated by the material distribution path planning model, wherein: The material distribution path planning model takes minimizing the total time for vehicles to arrive at all disaster-stricken points from each distribution point as an optimization problem, and solves the optimization problem through a memetic algorithm to obtain the optimal solution for path planning. The memetic algorithm repairs individuals in the initial population that violate vehicle constraints through a double repair mechanism.

[0054] For the details of each step in this embodiment, please refer to the first embodiment, which will not be described in detail here. Since this embodiment and the first embodiment adopt the same technical concept, it also has the technical effects described in the first embodiment. Embodiment 4

[0055] The embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the steps of the following method are implemented: Obtain basic road network data, vehicle data, distribution point data and disaster-affected point data in the disaster-affected area; The basic road network data, vehicle data, distribution point data and disaster-affected point data are input into a preset material distribution path planning model to obtain a material distribution strategy generated by the material distribution path planning model, wherein: The material distribution path planning model takes minimizing the total time for vehicles to arrive at all disaster-stricken points from each distribution point as an optimization problem, and solves the optimization problem through a memetic algorithm to obtain the optimal solution for path planning. The memetic algorithm repairs individuals in the initial population that violate vehicle constraints through a double repair mechanism.

[0056] For the details of each step in this embodiment, please refer to the first embodiment, which will not be described in detail here. Since this embodiment and the first embodiment adopt the same technical concept, it also has the technical effects described in the first embodiment.

[0057] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

[0058] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A method that specifies the functionality of a box or multiple boxes.

[0059] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce a manufactured product including an instruction method, which is implemented in the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0060] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0061] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the technical principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A post-disaster multi-center material distribution method based on a double repair mechanism, characterized in that: include: Obtain basic road network data, vehicle data, distribution point data and disaster-affected point data in the disaster-affected area; The basic road network data, vehicle data, distribution point data and disaster-affected point data are input into a preset material distribution path planning model to obtain a material distribution strategy generated by the material distribution path planning model, wherein: The material distribution path planning model takes minimizing the total time for vehicles to reach all disaster-stricken points from various distribution points as an optimization problem, and solves the optimization problem through a memetic algorithm to obtain the optimal solution for path planning. The memetic algorithm repairs the individuals that violate vehicle constraints in the initial population and the evolution process through a double repair mechanism.

2. The post-disaster multi-center material distribution method based on the double repair mechanism according to claim 1 is characterized in that: The preset objective function and constraints of the material distribution path planning model include: （1） （2） （3） （4） （5） （6） （7） （8） （9） （10） （11） （12） （13） Among them, G = (V, E) is the disaster-stricken area, and the vertex set V includes V C and V D Two subsets, V C ={v1,v2,v3,…,vN} represents the set of disaster-affected points, V D ={vN+1,vN+2,vN+3,…,vN+M} represents the set of distribution centers, E is the edge set, representing the road, (vi,vj)∈E(vi,vj∈V) is the edge between any two vertices; the fleet K is stationed at V D At each parking lot, the vehicle capacity is Q, and the center d∈ V D The maximum number of vehicles stationed at is Kd; let the time for vehicle k to arrive at distribution point i be , the material demand of distribution point i is qi; if distribution point i is not served by vehicle k, then is 0; Formula (1) is the objective function; Formula (2) ensures that the vehicle leaves immediately after arriving at the disaster site; Formula (3) stipulates that each disaster site only delivers materials once; Formula (4) ensures that the vehicle eventually returns to the same distribution center, and Formula (5) constrains each vehicle to provide at most one service; Formulas (6) and (7) constrain the total number of vehicles and the maximum number of vehicles at each distribution center; Formula (8) constrains that the vehicle cannot exceed the maximum load Q; Formula (9) gives a method for calculating the time it takes for the vehicle to reach the disaster site, and avoids subpaths by using a sufficiently large number P; Formulas (10)-(11) define the vehicle arrival time; Formulas (12)-(13) define the binary variable Respectively indicate whether vehicle k departs from the distribution center d and whether it passes through the edge (i, j); The solution of the objective function is expressed as: （14） Among them, |RD| represents the number of paths starting from parking lot d, and the solution of the problem is the number of paths from all parking lots. composition.

3. The post-disaster multi-center material distribution method based on the double repair mechanism according to claim 2 is characterized in that: The method of solving the optimization problem by using a memetic algorithm to obtain the optimal solution for path planning includes: Generate an initial population, and use a double patching mechanism to patch individuals in the population that violate vehicle constraints; Save the best feasible solution in the population to S; Determine whether the algorithm has reached the termination condition. If so, take S as the optimal solution for path planning. Otherwise, initialize the number of iterations and execute the iteration steps until the preset maximum number of iterations is reached. Then, use the double patching mechanism to patch the individuals in the population that violate vehicle constraints and update S. The iterative steps include: Use the crossover operator to generate sub-individuals, and perform local search on the generated sub-individuals; Determine whether the child individual is a non-clone individual of the population. If so, add the child individual to the population, sort the individuals in the population, and remove the last individual.

4. The post-disaster multi-center material distribution method based on the double repair mechanism according to claim 3 is characterized in that: The initial population is generated by a multi-strategy initial population generation method, which includes a population initialization method based on clustering, a population initialization method based on optimal insertion method, and a random population initialization method, wherein: The number of individuals generated by the clustering-based initialization method is 1, the number of individuals generated by the optimal insertion method is 1, and the number of individuals generated by the random method is the population size minus 2.

5. The post-disaster multi-center material distribution method based on the double repair mechanism according to claim 4 is characterized in that: For individuals in the initial population that violate vehicle constraints, a double patching mechanism is used to patch them, including: Check whether the number of vehicles in the current individual exceeds the preset maximum number of vehicles. If so, perform the patching step; if not, check whether the number of vehicles at each distribution center exceeds the maximum number of vehicles. If so, perform RepDepV patching, otherwise, the algorithm ends; The repairing steps include: Create a copy of the current individual and perform RepTotV patching on the copy; Check whether the total number of vehicles in the patched copy is within the allowed range. If so, assign the copy to the original individual and accept the patching result. If not, terminate the patching process. If there is a problem of vehicles crossing the distribution center in the patched current individual, the RepDepV patch is implemented on the current individual.

6. The post-disaster multi-center material distribution method based on the double repair mechanism according to claim 5 is characterized in that: The RepTotV patch includes: Copy the current path set ind to the original path set copy OldInd, and set the scattered path set DRS to empty; Keep the paths with the largest number of disaster-affected points |K| in the current path set ind, and move the remaining paths to the scattered path set DRS; Define an insertion disaster point set array ICS[|K|] with a length equal to the number of disaster points |K|, and set each element to an empty set; For each path R in the scattered path set DRS, the disaster-affected points are sorted from large to small according to the material demand. If there is a path ind[j] in the current path set ind, so that after inserting the disaster-affected point R[i], the capacity constraint Q can still be met, then the disaster-affected point R[i] is added to the disaster-affected point set ICS[j]. Otherwise, the original path set copy OldInd is returned and the repair is completed. For each path ind[j] in the current path set ind, j=1→|K|, use the best insertion method to insert all the disaster-affected points in the disaster-affected point set ICS[j] into the path ind[j]; Returns the current path set ind.

7. The post-disaster multi-center material distribution method based on the double repair mechanism according to claim 6 is characterized in that: The RepDepV patch includes: Define an array cnt[|VD|] with a length equal to the number of distribution centers to count the number of vehicles used at each distribution center, and initialize all elements to 0; For each path R in the current path set ind, set the minimum cost minCst to InfValue, where InfValue is the maximum value in the real number range; for the distribution center d∈VD that is not full, calculate the additional vehicle travel time time(R, d) caused by replacing the distribution center of path R with d. If the minimum cost minCst is greater than time(R, d), assign the minimum cost minCst to time(R, d), assign d* to d, replace the distribution center d of path R with d*, and increase cnt[d*] after replacement by 1; where Kd > cnt[d], Kd is the vehicle limit of distribution center d, and cnt[d] is the number of vehicles used by distribution center d; The current path set ind is searched in the neighborhood using two traditional local search operators: single-point insertion and exchange. Returns the current path set ind.

8. A post-disaster multi-center material distribution system based on a dual repair mechanism, characterized in that: The system comprises: Acquisition module: used to obtain basic road network data, vehicle data, distribution point data and disaster-affected point data in the disaster-affected area; Path planning module: used to input the basic road network data, vehicle data, distribution point data and disaster-affected point data into a preset material distribution path planning model to obtain the material distribution strategy generated by the material distribution path planning model, wherein: The material distribution path planning model takes minimizing the total time for vehicles to reach all disaster-stricken points from various distribution points as an optimization problem, and solves the optimization problem through a memetic algorithm to obtain the optimal solution for path planning. The memetic algorithm repairs the individuals that violate vehicle constraints in the initial population and the evolution process through a double repair mechanism.

9. A computer device, characterized in that: including processor and storage medium; The storage medium is used to store instructions; The processor is used to operate according to the instructions to execute the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method described in any one of claims 1 to 7 are implemented.

Images