Just-in-time material distribution method using charged single-lift electric distribution vehicle

Abstract

The application discloses a kind of material distribution method for on-time of single load electric distribution vehicle with charging, comprising the following steps: obtaining the production plan and material demand information of automobile assembly line;Establish the information model of electric distribution trolley;Establish the objective function and constraint condition of on-time material distribution scheduling of single load electric distribution vehicle with charging;The material distribution scheduling model is solved based on branch and price algorithm to obtain the optimal material distribution scheduling strategy;Electric distribution vehicle executes material distribution task of station according to the scheduling strategy.Compared with prior art, the application considers the influence of battery capacity limit of electric distribution vehicle on scheduling result, can accurately solve the optimal electric distribution vehicle material distribution scheduling strategy, has the advantages of being able to effectively reduce the number of distribution vehicle, improve material distribution efficiency and stability etc..

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of material distribution, in particular to a just-in-time material distribution method using a single-load electric distribution vehicle charged. BACKGROUND

[0002] Mixed-model assembly line can meet the market individualization and diversification needs, enhance the enterprise response speed to the market, and thus is widely used in automobile, household appliance and other assembly-oriented manufacturing industries. The corresponding production logistics system is an important subsystem of modern manufacturing industry.

[0003] Workshop material distribution system is an important part of enterprise production system, and is also a subsystem of enterprise logistics system, responsible for the multi-species and small-batch distribution of materials required by the production workshop. Whether the production materials can be delivered to the production site in a timely and accurate manner is the key to the smooth operation of the enterprise production logistics system. Therefore, the just-in-time distribution of materials is the premise and guarantee of smooth production of manufacturing enterprises.

[0004] The ideal material distribution requirement on the assembly line is: "according to the correct conditions, the correct amount of correct materials is distributed to the correct place at the correct time", and the line-side inventory neither lacks materials nor accumulates. Therefore, how to model and optimize the workshop logistics distribution system to realize the just-in-time material supply of mixed-model production line is crucial to the smooth operation of the assembly line.

[0005] In the selection of transportation tools, electric trolley has the advantages of no emission of pollutants during driving and low running noise, and thus is more suitable for in-plant and out-of-plant "last mile" material distribution.

[0006] In solving the integer programming problem of electric trolley distribution, branch and price algorithm is widely used due to its fast solving speed. Branch and price algorithm searches all feasible solution spaces of constrained optimization problems, divides the entire feasible solution space into smaller and smaller branches, and calculates a lower bound or upper bound for the value of the solution in each branch. After each branch, the part of the branch whose limit exceeds the known feasible solution value is pruned, thereby narrowing the search range and improving the search efficiency, and thus is suitable for solving the material distribution problem of vehicles. The existing technologies using branch and price algorithm to solve the distribution problem are as follows:

[0007] (1) A collaborative distribution path optimization method based on branch-and-price cutting algorithm (CN114037180A) discloses a collaborative distribution path optimization method based on branch-and-price cutting algorithm, comprising the following steps: S1, establishing a set partitioning model, the model is established on the basis of feasible vehicle and unmanned aerial vehicle collaborative distribution path (feasible vehicle and unmanned aerial vehicle collaborative path refers to the path that meets the customer time window, demand, maximum service time and load of vehicle / unmanned aerial vehicle), on the basis of meeting the single service of each customer, the total distribution cost is minimized, wherein the total cost includes the fixed use cost of vehicle and the distribution cost of vehicle and unmanned aerial vehicle; S2, an accurate algorithm based on branch-and-price cutting is used to solve the set partitioning model, and the optimal vehicle and unmanned aerial vehicle collaborative distribution route is obtained.

[0008] (2) A distribution route optimization method based on vehicle and unmanned aerial vehicle collaboration (CN114462693A) discloses a distribution route optimization method based on vehicle and unmanned aerial vehicle collaboration, comprising the following steps: S1, establishing a mixed integer programming model for vehicle and unmanned aerial vehicle collaborative blood distribution; S2, a logarithmic-based method is used to change the above-mentioned mixed integer programming, and by increasing auxiliary linear constraints, the number of binary variables is significantly reduced; S3, the above-mentioned mixed integer programming model is divided into Benders master problem and Benders subproblem by using Benders representation, the branch-and-price cutting algorithm is used to solve the Benders subproblem, and then the selection of cluster center hospital, the distribution of non-cluster center hospital and the vehicle driving route are optimized, and the optimized cluster center hospital selection and distribution strategy are obtained.

[0009] However, unlike the constraints considered in the unmanned aerial vehicle distribution problem, the electric car distribution needs to consider the charging process of the electric distribution vehicle at the charging station in addition to the just-in-time distribution and path optimization, which makes the sub-problem essentially a primary shortest path problem with resource constraints considering the battery capacity constraint and time window constraint of the vehicle. If a one-way label algorithm based on dynamic programming is used, a large number of labels will be generated, reducing the solving speed. Moreover, if the decomposition method of the existing technology for the optimization model is directly applied to the electric car distribution problem, the number of constraints is still relatively large, and it is not the optimal solution. SUMMARY

[0010] In view of the above problems existing in the prior art, the present application provides a just-in-time material distribution method for electric distribution vehicles with single load capacity considering the influence of battery capacity limitation of electric distribution vehicles on scheduling results, which can accurately solve the optimal electric distribution vehicle material distribution scheduling strategy and has the advantages of being able to effectively reduce the number of distribution vehicles, improve the efficiency and stability of material distribution, etc.

[0011] The technical scheme of the present application is as follows:

[0012] The just-in-time material distribution method using single-load electric distribution vehicles with charging includes the following steps:

[0013] S1, obtaining information of an automobile assembly line, including but not limited to: production plan, material demand information of each station in the planning period, replenishment time window information, distance from the material supermarket to the assembly station, material loading and unloading time;

[0014] S2, establishing an information model of an electric distribution vehicle, including: maximum loading capacity, power level, power consumption rate;

[0015] S3, establishing a target function of just-in-time material distribution scheduling of single-load electric distribution vehicles with charging;

[0016] S4, determining the constraint conditions of the scheduling process, including: uniqueness constraint, task chain related constraint, time window constraint, vehicle constraint, power constraint;

[0017] S5, solving the material distribution scheduling model based on the branch and price algorithm to obtain the optimal material distribution scheduling strategy; the material distribution strategy includes the distribution task sequence executed by the electric distribution vehicle and the charging time of the electric distribution vehicle;

[0018] S6, each electric distribution vehicle executes the material distribution task of the station according to the scheduling strategy.

[0019] Further, the electric distribution vehicle uses point-to-point material distribution, that is, the electric distribution vehicle only carries the material of the corresponding station and distributes the material from the material supermarket to the corresponding station and then returns to the material supermarket.

[0020] Further, the electric distribution vehicle is a single-load vehicle with limited battery capacity, which needs to visit a charging station within the planning period.

[0021] Further, the material distribution scheduling in step S3 aims to minimize the number of electric distribution vehicles, and the target function is:

[0022]

[0023] wherein: y k is a binary variable, 1 if the distribution vehicle k executes the distribution task, otherwise 0; V is a set of distribution vehicles, k∈V.

[0024] Further, the constraint conditions of step S4 are as follows:

[0025] (1) Uniqueness constraint:

[0026]

[0027] Where: x ijk The variable is a binary variable. If the delivery vehicle k executes tasks i and j in sequence, it is 1; otherwise, it is 0. N′ is the set of delivery tasks and charging stations. n+1 is a virtual node, representing the end point of the task chain.

[0028] (2) Task chain related constraints:

[0029]

[0030]

[0031]

[0032] Where: N is the set of delivery tasks, N = {1, ..., n}; 0 is a virtual node, representing the starting point of the task chain;

[0033] (3) Time window constraint:

[0034] τ ik +(2D i +LT+UT)-M·(1-x ijk )≤τ jk ,

[0035]

[0036] τ ik +(2D i +g·(Qu ik ))-M·(1-x ijk )≤τ jk ,

[0037]

[0038]

[0039] Where: τ ik The start time for delivery vehicle k to perform task i; D i LT and UT represent the one-way travel time of the delivery vehicle to the workstation corresponding to task i; M is a large positive integer; Q is the battery capacity of the delivery vehicle; u ik The remaining battery power of delivery vehicle k when performing task i; [s i ,l i [ ] represents the time window of delivery task i; E represents the set of charging stations, e∈E; g represents the charging rate;

[0040] (4) Vehicle constraints:

[0041]

[0042] (5) Power constraint:

[0043] 0≤u jk ≤u ik -2D i h-δw i d i D i +Q(1-x ijk ),

[0044]

[0045]

[0046]

[0047] wherein: h is the power consumption rate when empty; δ is the power consumption increment per unit weight of material per unit time; w i is the unit weight of material of the distribution task i; d i is the material demand of the distribution task i.

[0048] Further, the steps of the branch-and-price algorithm in step S5 are as follows:

[0049] S5-1, using Dantzig-Wolfe decomposition technique to decompose the objective function and constraints into master problem and pricing sub-problem;

[0050] S5-2, relaxing the integer constraint condition of the master problem, solving to obtain the corresponding dual variable value;

[0051] S5-3, updating the corresponding coefficients of the pricing sub-problem according to the dual variable value, solving the pricing sub-problem; if a column that can improve the objective value of the master problem is found, it is added to the master problem;

[0052] S5-4, relaxing the master problem and sub-problem gradually, finding the optimal solution of the relaxed master problem;

[0053] S5-5, if the solution of the relaxed master problem is an integer solution, the algorithm is ended, otherwise, applying the branch strategy, returning to step S5-2.

[0054] Further, the objective function of the master problem is:

[0055]

[0056] The constraint conditions include:

[0057]

[0058]

[0059]

[0060] where: ω is a dispatching strategy of a single delivery vehicle, described as a feasible delivery sequence of a single delivery vehicle satisfying the constraints; c ω is the cost of dispatching ω; z ω is a binary variable, taking 1 if the dispatching strategy ω is selected in the optimal solution, otherwise taking 0; α iω represents that the dispatching strategy ω performs the delivery task i, otherwise 0;

[0061] The model is changed from to , that is, the linear relaxation master problem of the main problem is obtained; solving the linear relaxation master problem, denoted as π i , η is the corresponding dual variable value.

[0062] Further, the objective function of the pricing sub-problem is:

[0063]

[0064] Solving the minimum pricing sub-problem adopts a label algorithm.

[0065] Further, in step S5-5, the branching strategy branches the variable taking a fractional value of the relaxation master problem as follows:

[0066] S5-5-1, if the optimal objective value of the relaxation master problem is not less than the current optimal integer solution, then prune and end, otherwise execute step S5-5-2;

[0067] S5-5-2, if the optimal solution of the relaxation master problem is an integer solution, then replace the optimal solution of the relaxation master problem with the current optimal solution, and then prune and end, otherwise execute step S5-5-3;

[0068] S5-5-3, if the optimal objective value of the relaxation master problem is less than the current optimal integer solution, and part of the variables z ω is a decimal number, then branch.

[0069] Further, the branching step of step S5-5-3 is as follows:

[0070] S5-5-3-1, select the branching variable r * as the delivery path corresponding to the dispatching strategy ω * ;

[0071] S5-5-3-2, select a task node i and (j, i) ∈ r * ;

[0072] S5-5-3-3, the arc (j, i) on one branch is forbidden to access, and the arcs other than the arc (j, i) on the other branch are forbidden to access.

[0073] The beneficial technical effect of the present application is that:

[0074] (1) The present application combines the branch and bound algorithm and the column generation algorithm to solve the electric vehicle distribution problem considering battery capacity. The branch and bound algorithm used in the algorithm first decomposes the original problem into a main problem and a pricing sub-problem based on the Dantzig-Wolfe decomposition technique based on the mixed integer programming model, and then uses the column generation algorithm to find the linear relaxation optimal solution of the main problem by gradually iterating between the main problem and the sub-problem. Compared with the traditional branch and bound algorithm, the relaxation optimal solution can provide a tighter lower bound than the original problem relaxation, so that large-scale mixed integer programming models can be solved at a faster rate.

[0075] (2) Compared with the prior art, the present application uses the Dantzig-Wolfe decomposition algorithm in the branch and bound algorithm. This decomposition technique can be used for large-scale linear programming problems. The single weighted constraint replaces the polyhedral set constraint, thereby reducing the number of constraints, simplifying the model and improving the calculation efficiency.

[0076] (3) The prior art uses a one-way label algorithm based on dynamic programming, which generates a large number of labels. The present application uses a two-way label algorithm, which overcomes the problems of the prior art and improves the speed of solving the pricing sub-problem. BRIEF DESCRIPTION OF DRAWINGS

[0077] Figure 1 is a layout diagram of the just-in-time material distribution system of the embodiment;

[0078] Figure 2 is a whole flow chart of the branch and bound algorithm of the embodiment.

[0079] In the figure, the correspondence between the component names and the figure numbers is as follows: 1, product; 2, assembly station; 3, material; 4, electric distribution vehicle; 5, distribution path; 6, material supermarket; 7, charging station. DETAILED DESCRIPTION

[0080] The present application will be described in detail below in conjunction with the drawings and embodiments. Obviously, the described embodiments are only a part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0081] The just-in-time material distribution method of the single-load electric distribution vehicle of the embodiment is realized based on the point-to-point supply system of the material supermarket. As shown in Figure 1 The point-to-point material supermarket system includes work-in-process 1, assembly station 2, material 3, electric distribution vehicle 4, distribution path 5, material supermarket 6, and charging station 7.

[0082] The material supermarket 6 is arranged beside the assembly line to supply materials to the assembly station 2, and the material distribution of the material supermarket 6 is performed by the electric distribution vehicle 4 along the distribution path 5 to perform point-to-point material distribution, and the electric distribution vehicle 4 will go to the charging station 7 for charging when the electric quantity is consumed. In the embodiment, the just-in-time material distribution method of the single-load electric distribution vehicle with charging can be used to design the replenishment task sequence of the electric distribution vehicle 4 for each station 2.

[0083] Based on a just-in-time material distribution method of a single-load electric distribution vehicle with charging, a point-to-point material distribution scheduling from a material supermarket to an assembly line station is performed in the in-plant logistics stage, and a single-load electric distribution vehicle with charging is used for distribution. The electric distribution vehicle only carries materials corresponding to the station, distributes the materials from the material supermarket to the corresponding station, and returns to the material supermarket. The battery capacity of the electric distribution vehicle is limited, and it needs to visit the charging station within the planning period. The method includes the following steps:

[0084] S1, obtaining the production plan of the automobile assembly line, the material demand information of each station in the planning period, the replenishment time window information, the distance from the material supermarket to the assembly station, and the time required for material loading and unloading.

[0085] S2, establishing an information model of the electric distribution trolley, including: the maximum loading capacity of the electric distribution trolley, the electric quantity level of the electric distribution trolley, and the power consumption rate of the electric distribution trolley.

[0086] S3, establishing a target function of the just-in-time material distribution scheduling of the single-load electric distribution vehicle with charging:

[0087]

[0088] In the formula, y k is a binary variable, which is 1 if the distribution vehicle k performs the distribution task, and 0 otherwise; V is a set of distribution vehicles, k ∈ V.

[0089] S4, determining the constraint conditions of the scheduling process, including: uniqueness constraint, task chain related constraint, time window constraint, vehicle constraint, and electric quantity constraint.

[0090] (1) Uniqueness constraint:

[0091]

[0092] where x ijk is a binary variable, 1 if delivery vehicle k performs tasks i, j successively, otherwise 0; N is the set of delivery tasks and charging stations; n+1 is a virtual node, representing the end of the task chain.

[0093] (2) Task chain related constraints:

[0094]

[0095]

[0096]

[0097] where N is the set of delivery tasks, N = {1, …, n}; 0 is a virtual node, representing the start of the task chain.

[0098] (3) Time window constraints:

[0099] τ ik + (2D i + LT + UT) - M · (1 - x ijk ) ≤ τ jk ,

[0100]

[0101] τ ik + (2D i + g · (Q - u ik )) - M · (1 - x ijk ) ≤ τ jk ,

[0102]

[0103]

[0104] where τ ik is the start time of delivery vehicle k performing task i; D i is the one-way travel time of delivery vehicle to the workstation corresponding to task i; LT and UT are the loading and unloading times of the material box; M is a large positive integer; Q is the battery capacity of delivery vehicle; u ik is the remaining battery capacity of delivery vehicle k performing task i; [s i , l i ] is the time window of delivery task i; E is the set of charging stations, e ∈ E; g is the charging rate.

[0105] (4) Vehicle constraints:

[0106]

[0107] (5) Power constraint is:

[0108] 0≤u jk ≤u ik -2D i h-δw i d i D i +Q(1-x ijk ),

[0109]

[0110]

[0111]

[0112] where h is the power consumption rate when empty; δ is the power consumption increment per unit weight of material per unit time; w i is the unit weight of material for distribution task i; d i is the material demand for distribution task i.

[0113] S5, solving the material distribution scheduling model based on the branch and price algorithm to obtain the optimal material distribution scheduling strategy. Figure 2 The branch and price algorithm flowchart is shown, and the algorithm steps are as follows:

[0114] S5-1, using Dantzig-Wolfe decomposition technique to decompose the objective function and constraints into master problem and pricing sub-problem. After decomposition, the master problem is as follows:

[0115] The objective function of the master problem is:

[0116]

[0117] The constraint conditions include:

[0118]

[0119]

[0120]

[0121] where ω is a single distribution vehicle scheduling strategy, described as a single distribution vehicle in a feasible distribution sequence that meets the constraints; c ω is the cost of scheduling ω; z ω is a binary variable, which is 1 if the optimal solution selects the scheduling strategy ω, otherwise 0; α iω represents that the scheduling strategy ω executes the distribution task i is 1, otherwise 0.

[0122] S5-2, relax the integer constraints of the master problem, and solve it to get the corresponding dual variable values. Replace the integer constraints in the master problem with linear relaxation of the master problem. Solve the linear relaxation of the master problem using a linear programming solver, and denote the solution as i ,η as the corresponding dual variable values.

[0123] S5-3, update the corresponding coefficients of the pricing subproblem according to the dual variable values, and get the objective function of the pricing subproblem:

[0124]

[0125] Solve the pricing subproblem using the label algorithm. Any partial path ω from the virtual node 0 to the node i,i∈N' can be represented by the label . Wherein, represents the task sequence (0,…,e,…,i) visited by the path ω when reaching task node i; represents the objective value of the path ω when reaching task node i; represents the remaining power of the path ω when reaching task node i; represents the number of charging times of the path ω when reaching task node i; represents the earliest time of the path ω reaching task node i; represents the number of times of executing the delivery task k,k∈N of the path ω when reaching task node i.

[0126] The label at the virtual node 0 is the initial label F ω0 = ({0},0,Q,0,0,[0] N ). When expanding the label from task node i to task node j, the power non-negative constraint and the time window constraint need to be satisfied, only the label that satisfies the constraint can be further expanded, and the label recursive formula is as follows:

[0127]

[0128]

[0129]

[0130]

[0131]

[0132]

[0133] For labels of different paths reaching the same node, if and ​Then it is called dominate In the label algorithm, through the domination rule, the label that cannot produce the optimal solution can be determined in advance, and the dominated does not affect the solution of the subproblem, avoids the subsequent expansion of the dominated label, and can improve the solution speed of the label algorithm. It can be proved that the optimal solution expanded by the label is worse than the optimal solution obtained by the label .

[0134] When all labels are expanded to the termination node n+1, from all the feasible task sequences obtained, the task sequence with the minimum pricing subproblem objective value is found, which is the optimal solution of the pricing subproblem. If the objective value is negative, it indicates that a column is found that can improve the objective value of the main problem, which is added to the relaxed main problem; otherwise, the relaxed main problem has obtained the optimal solution.

[0135] S5-4, the relaxed main problem and the subproblem are iterated step by step to find the optimal solution of the relaxed main problem.

[0136] S5-5, if the solution of the relaxed main problem is an integer solution, the algorithm ends, otherwise, a branching strategy is applied, and step S5-2 is returned. The branching strategy branches the variable with fractional value of the relaxed main problem, and there can be the following three cases:

[0137] Case one: the optimal objective value of the relaxed main problem is not less than the current optimal integer solution, then the branch is pruned.

[0138] Case two: the optimal objective value of the relaxed main problem is less than the current optimal integer solution, and the optimal solution of the relaxed main problem is an integer solution, then the optimal solution of the relaxed main problem is replaced by the current optimal solution, and the branch is pruned.

[0139] Case three: the optimal objective value of the relaxed main problem is less than the current optimal integer solution, and part of the variables z ω is a fraction, then branching is performed. First, select the branching variable r * as the distribution path corresponding to the scheduling strategy ω * ; select a task node i and (j, i) ∈ r * ; on one branch, the arc (j, i) is prohibited from being accessed, and on the other branch, all arcs except the arc (j, i) are prohibited from being accessed.

[0140] S6, each electric distribution vehicle performs the material distribution task of the work station according to the scheduling strategy.

[0141] Although embodiments of the present application have been disclosed in connection with the illustrative embodiments set forth above, it should be understood that various substitutions, modifications and changes can be made by those skilled in the art to the illustrative embodiments without departing from the spirit and scope of the application. For example, although the present application has been described in connection with a single user, the present application can be used in connection with multiple users. Furthermore, the scope of the application should not be limited by the specific details provided in the description, and any sole method candidates should not be limited to the scope of the claims and their equivalents.

Claims

1. A timely material delivery method using a single-capacity electric delivery vehicle with charging capability, characterized in that, Includes the following steps: S1. Obtain information about the automotive assembly line, including but not limited to: production plan, material requirements for each workstation during the planning period, replenishment time window information, distance from the material supermarket to the assembly workstation, and time required for material loading and unloading. S2. Establish an information model for electric delivery vehicles, including: maximum loading capacity, power level, and power consumption rate; S3. Establish an objective function for just-in-time material delivery scheduling using single-capacity electric delivery vehicles equipped with charging systems; the electric delivery vehicles are single-capacity vehicles with limited battery capacity and need to access charging stations within the planning period; the material delivery scheduling aims to minimize the number of electric delivery vehicles, and the objective function is: Where: y k The variable is a binary variable, which is 1 if the delivery vehicle k is performing a delivery task, and 0 otherwise; V is the set of delivery vehicles, where k∈V; S4. Determine the constraints of the scheduling process, including: uniqueness constraints, task chain-related constraints, time window constraints, vehicle constraints, and power constraints; S5. Solve the material delivery scheduling model based on the branch pricing algorithm to obtain the optimal material delivery scheduling strategy; the material delivery scheduling strategy includes the delivery task sequence executed by the electric delivery vehicle and the charging time of the electric delivery vehicle. S6. Each electric delivery vehicle performs material delivery tasks at the workstation according to the aforementioned scheduling strategy.

2. The timely material delivery method using a single-capacity electric delivery vehicle with charging capability according to claim 1, characterized in that, The electric delivery vehicles adopt point-to-point material delivery, that is, the electric delivery vehicles only transport materials to the corresponding workstation, deliver the materials from the material supermarket to the corresponding workstation, and then return to the material supermarket.

3. The timely material delivery method using a single-capacity electric delivery vehicle with charging capability according to claim 1, characterized in that, The constraints for step S4 are as follows: (1) Uniqueness constraint: Where: x ijk This is a binary variable; if delivery vehicle k executes tasks i and j sequentially, it is 1; otherwise, it is 0. N' is the set of delivery tasks and charging stations; n+1 is a virtual node representing the end point of the task chain. (2) Task chain related constraints: Where: N is the set of delivery tasks, N = {1, ..., n}; 0 is a virtual node, representing the starting point of the task chain; (3) Time window constraint: Where: τ ik The start time for delivery vehicle k to perform task i; D i LT and UT represent the one-way travel time of the delivery vehicle to the workstation corresponding to task i; M is a large positive integer; Q is the battery capacity of the delivery vehicle; u ik The remaining battery power of delivery vehicle k when performing task i; [s i ,l i [ ] represents the time window of delivery task i; E represents the set of charging stations, e∈E; g represents the charging rate; (4) Vehicle constraints: (5) Power Constraint: Where: h is the power consumption rate under no-load conditions; δ is the power consumption increment per unit time for each additional unit weight of material; w i d represents the unit material weight for delivery task i; i The material requirements for delivery task i.

4. The timely material delivery method using a single-capacity electric delivery vehicle with charging capability according to claim 1, characterized in that, The steps of the branch pricing algorithm described in step S5 are as follows: S5-1. The objective function and constraints are decomposed into a main problem and a pricing subproblem using the Dantzig-Wolfe decomposition technique; the objective function of the main problem is: The constraints include: Where: ω is the scheduling strategy for a single delivery vehicle, described as a feasible delivery sequence for a single delivery vehicle that satisfies the constraints; c ω The cost of scheduling ω; z ω For binary variables, ω is set to 1 if the optimal solution selects the scheduling strategy ω, and 0 otherwise; α iω The scheduling strategy ω is 1 if delivery task i is executed, otherwise 0; Ω is the set of all feasible delivery sequences that satisfy the constraints. In this model Change to This yields the linearly relaxed principal problem; solve the linearly relaxed principal problem, denoted by π. i η is the value of the corresponding dual variable; The objective function of the pricing subproblem is: The labeling algorithm is used to solve the problem of minimizing the pricing sub-problem; S5-2. Relax the integer constraints of the master problem, solve it, and obtain the corresponding dual variable values; S5-3. Update the corresponding coefficients of the pricing subproblem based on the values ​​of the dual variables, and solve the pricing subproblem; if a column that improves the objective value of the main problem can be found, add it to the main problem; S5-4. Iterate through the relaxation of the main problem and subproblems to find the optimal solution to the relaxation of the main problem; S5-5. If the solution to the relaxation master problem is an integer solution, end the algorithm; otherwise, apply the branching strategy and return to step S5-2.

5. The timely material delivery method using a single-capacity electric delivery vehicle with charging capability according to claim 4, characterized in that: In step S5-5, the branching strategy for the variable that takes a fractional value in the relaxation master problem proceeds as follows: S5-5-1. If the optimal objective value obtained from relaxing the principal problem is not less than the current optimal integer solution, then prune and terminate; otherwise, proceed to step S5-5-2. S5-5-2. If the optimal solution to the relaxation master problem is an integer solution, then replace the current optimal solution with the optimal solution to the relaxation master problem, then prune and terminate; otherwise, proceed to step S5-5-3. S5-5-3. If the optimal objective value of the relaxation principal problem is less than the current optimal integer solution, and some variables z ω If the number is a decimal, then branching is performed.

6. The timely material delivery method using a single-capacity electric delivery vehicle with charging capability according to claim 5, characterized in that: The steps for the branch described in step S5-5-3 are as follows: S5-5-3-1 Selecting Branch Variables r * For the corresponding scheduling strategy ω * Delivery routes; S5-5-3-2, Choose any task node i such that (j,i)∈r * ; S5-5-3-3: On one branch, arc (j,i) is prohibited from accessing, and on another branch, all arcs except arc (j,i) are prohibited from accessing.

Images