Retail cabinet inventory path decision optimization method and equipment

Abstract

The invention discloses a retail cabinet inventory path decision optimization method and equipment, and the method comprises the following steps: (1), building an inventory path optimization model of a retail cabinet, and enabling the inventory path optimization model to take the minimum sum of the inventory cost, the stockout cost and the vehicle path cost at the current planning period as an objective function, taking a replenishment constraint, a maximum inventory level constraint, a vehicle capacity constraint, a retail cabinet flow balance constraint, a vehicle closed-loop constraint and an elimination sub-loop constraint as constraint conditions; (2) obtaining inventory path related information; and (3) solving the inventory path optimization model by adopting a tabu search method according to the inventory path related information to obtain an optimized inventory decision and path decision. According to the invention, the replenishment decision and the path planning decision are considered at the same time during decision making, and the efficiency is higher.

Description

Technical Field

[0001] The present invention relates to retail cabinet optimization technology, and in particular to a retail cabinet inventory path decision optimization method and equipment. Background Art

[0002] Retail cabinets are a new retail model. Different from the traditional sales model through offline stores, retail cabinet sales are sold in small batches of snacks, drinks, snacks and other commodities by laying retail cabinets in office buildings, hospitals, universities, subways, industrial parks and other areas, to meet the last 100 meters of shopping needs of consumers in these closed or semi-closed places. Compared with the traditional retail industry, the retail cabinet industry does not need to rent offline stores or invest in human resources. It is completely purchased by consumers. It has the advantages of low initial investment, low operating costs, small footprint, and flexible site selection, and is favored by many retail giants and capital.

[0003] Retail cabinet operation is a relatively complex decision-making problem. Generally, operators will make replenishment plans on a daily basis. The daily decision content is the replenishment plan for the next day, including the list of retail cabinets to be replenished, the details of the goods replenished by each smart retail cabinet, and the replenishment route sequence of the operating driver. The operator's goal is to keep the inventory of goods in the cabinet at a high level at all times to maximize the coverage of customer needs, while at the same time reducing the burden on the replenishment driver as much as possible, that is, optimizing the driver's replenishment route.

[0004] At present, most retail cabinet operators adopt the solution of first making replenishment decisions and then route planning decisions. That is, first, based on the daily sales volume of goods in the cabinet and the remaining inventory in the cabinet, with the goal of minimizing the loss of out-of-stock, select the list of retail cabinets that need to be replenished the next day; secondly, based on these selected retail cabinets, with the goal of minimizing the delivery path cost, help the staff plan the replenishment route; the last step is manual screening to eliminate those retail cabinets that are geographically isolated and drivers do not have time to replenish. The advantage of this decision is that the process is clear and easy to execute, but it is relatively extensive and inefficient. Summary of the invention

[0005] In view of the problems existing in the prior art, the object of the present invention is to provide a highly efficient retail cabinet inventory path decision optimization method and device.

[0006] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:

[0007] A retail cabinet inventory path decision optimization method comprises the following steps:

[0008] (1) Establishing a retail counter inventory path optimization model, wherein the objective function of the inventory path optimization model is to minimize the sum of inventory cost, out-of-stock cost and vehicle path cost in the current planning period, and the constraints are replenishment constraint, maximum inventory level constraint, vehicle capacity constraint, retail counter flow balance constraint, vehicle closed loop constraint and sub-loop elimination constraint; wherein the objective function of the inventory path optimization model is:

[0009]

[0010]

[0011] In the formula, h i represents the unit inventory cost, l i Indicates the unit out-of-stock loss, Respectively represent inventory levels The positive and negative parts of represents the remaining inventory level of retail counter i at the end of the t-th planning period, and serves as the initial inventory level of the t+1-th planning period. express The positive part, represents the remaining inventory level of retail counter i at the end of the t-1 planning period, c ij represents the path cost from retail counter i to retail counter j, represents a 0-1 variable indicating whether edge (i, j) is the route of vehicle k in period t. , it means that edge (i, j) is the route of vehicle k in period t, otherwise it is not; represents a 0-1 variable indicating whether retail cabinet i is replenished in the tth planning period. When , it means that retail cabinet i is replenished in the t planning period, otherwise it means that it is not; represents a 0-1 variable indicating whether vehicle k replenishes retail cabinet i in the tth planning period. When , it means that vehicle k replenishes retail counter i in the tth planning period, otherwise it means no; Represents a set of points, where vertex 0 represents the supply warehouse, and the other vertices Represents a collection of retail counter locations. represents a collection of vehicles, represents the set of planning periods, represents the demand for retail cabinet i in the tth planning period, represents the replenishment quantity received by retail cabinet i in the tth planning period, represents the replenishment quantity of vehicle k to retail cabinet i in the tth planning period;

[0012] (2) Obtaining information related to the inventory path, including the warehouse location, the location of each retail counter, the initial inventory level of each retail counter in the current planning period, the maximum inventory level of each retail counter, the unit out-of-stock loss of each retail counter, and the demand and vehicle capacity of each retail counter in the current planning period;

[0013] (3) Based on the inventory path related information, the taboo search method is used to solve the inventory path optimization model to obtain the optimized inventory decision and path decision.

[0014] Furthermore, the replenishment constraints are specifically:

[0015]

[0016]

[0017] In the formula, C i represents the maximum inventory level of retail cabinet i.

[0018] Furthermore, the maximum inventory level constraint is specifically:

[0019]

[0020] In the formula, They represent the remaining inventory levels of the warehouse at the end of the t and t-1 planning periods, respectively. t represents the warehouse's incoming inventory in each period, C i represents the maximum inventory level of retail cabinet i;

[0021] The vehicle capacity constraint is specifically:

[0022]

[0023] In the formula, Q represents the capacity of the vehicle, represents a 0-1 variable indicating whether the k-th vehicle visits the warehouse in period t. If the vehicle visits the warehouse, then otherwise

[0024] Furthermore, the retail counter flow balance constraint is specifically:

[0025]

[0026] In the formula, represents a 0-1 variable indicating whether the edge (j,i) is the route of vehicle k in period t. When , it means that edge (j,i) is the route of vehicle k in period t, otherwise it is not;

[0027] The vehicle closed-loop constraints are specifically:

[0028]

[0029] In the formula, They are 0-1 variables used to indicate whether the edge (0,j) and (j,0) are the routes of vehicle k in period t. When the variable is 1, it means that the edge is its route, otherwise it is not. The 0 in (0,j) and (j,0) indicates a warehouse. represents a 0-1 variable indicating whether the k-th vehicle visits the warehouse in period t. If the vehicle visits the warehouse, then otherwise

[0030] Furthermore, the elimination of the sub-ring constraint is specifically as follows:

[0031]

[0032] In the formula, represents a subset of the retail cabinet set, represents a 0-1 variable indicating whether vehicle k replenishes retail cabinet m in the tth planning period. , it means that vehicle k replenishes retail cabinet m in the tth planning period, otherwise it means no.

[0033] Furthermore, step (3) specifically includes:

[0034] (3.1) Construct two empty taboo tables and They are used to store the retail cabinet and replenishment period tuples (i, t) respectively, and initialize the perturbation length and the maximum number of iterations, where i represents the retail cabinet number and t represents the number of planning periods;

[0035] (3.2) Construct an initial solution and use it as the current solution; the structure of the solution is represents the retail cabinet replenishment decision in solution s, represents the replenishment quantity decision in solution s, represents the inventory level decision in solution s, represents the path decision in solution s, Respectively represent the solution s The value of

[0036] (3.3) Determine whether the termination condition is met, that is, whether the number of iterations is greater than the maximum number of iterations. If so, output the historical optimal solution; if not, execute (3.4);

[0037] (3.4) Calculate the neighborhood solution of the current solution using four different neighborhood operators, and select the best solution from the domain solutions as the candidate solution according to the contempt criterion; wherein the four different domain operators are specifically Remove operator, Insert operator, Relocate operator and Replace operator;

[0038] (3.5) Determine whether the candidate solution is better than the historical optimal solution. If not, execute (3.6). If yes, execute (3.7).

[0039] (3.6) Randomly perturb the candidate solution according to the perturbation length and execute (3.7);

[0040] (3.7) Improve the candidate solution or the solution after random perturbation of the candidate solution, including the improvement of the replenishment period of each retail cabinet and the improvement of the path plan, and take the improved solution as the optimal solution and execute (3.8);

[0041] (3.8) Update the taboo table according to the current solution and the optimal solution;

[0042] (3.9) Increase the number of iterations by 1 and return to execute step (3.3).

[0043] Furthermore, step (3.2) specifically includes:

[0044] (3.2.1) Without considering the impact of path cost, solve the following model to obtain the minimum inventory cost

[0045]

[0046] (3.2.2) Through the heuristic method, according to the obtained Find the time when the delivery cost is the lowest At this time As a feasible solution;

[0047] (3.2.3) The feasible solution is improved, and the improved solution is used as the initial solution.

[0048] Furthermore, step (3.8) specifically includes:

[0049] The current solution s is the best solution and is adopted as solution s , Indicates that if retail cabinet i is replenished in the tth planning period in solution s, but in solution s , If retail counter i is not replenished in the tth planning period, the tuple (i, t) is added to the taboo table , which means that retail cabinet i is prohibited from replenishing in period t in the following iterations;

[0050] If in solution s, retail cabinet i is not replenished in the tth planning period, but in solution s, In the tth planning period, retail cabinet i is replenished, then the tuple (i, t) is added to the taboo table , which means that retail cabinet i will maintain replenishment in period t in the next several iterations;

[0051] Among them, the taboo lengths of the two taboo tables are Where L and λ are fixed parameters, n is the number of retail cabinets, |R(s)| is the number of feasible paths in solution s, is 0 to Randomly pick a number between.

[0052] Furthermore, step (3.7) specifically includes:

[0053] (3.7.1) Improve the replenishment period of each retail counter: Add a two-dimensional array DeltaCost to record the change in the total cost caused by the change in the replenishment period. Then the element DeltaCost[i][t] of DeltaCost records if The estimated change in total cost from 0 to 1 is Then DeltaCost[i][t] records if Estimate the change in total cost from 1 to 0; then get the element with the smallest negative number from each row of DeltaCost, extract the value of the index number (i, t) of the element, and record it as (i0, t0). If the solution before the improvement is but The value after promotion If the previous solution is not improved but The value after promotion Finally, according to the OU policy Synchronously promote and get the promoted value

[0054] (3.7.2) According to the improved Re-plan the path and get the upgrade Value After the improvement As the solution after the current solution s is improved.

[0055] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the above method.

[0056] Compared with the prior art, the present invention has the following beneficial effects: the present invention provides a retail cabinet inventory path decision optimization method and device, and the present invention can reduce operating costs and improve efficiency by establishing a mathematical model of the inventory path problem and making joint optimization of replenishment decision and path planning. The taboo search algorithm proposed in the present invention can greatly improve the solution efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 It is a flowchart of a retail cabinet inventory path decision optimization method provided by an embodiment of the present invention;

[0058] Figure 2 Schematic diagram of four operators provided by the present invention;

[0059] Figure 3 It is a flow chart of the taboo algorithm provided by the present invention. DETAILED DESCRIPTION

[0060] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0061] Embodiment 1

[0062] First, the problem of the present invention is analyzed. It is necessary to abstract the retail cabinet operation replenishment problem into a mathematical model of the Inventory Routing Problem (IRP). The decision-making elements of the retail cabinet inventory routing problem include warehouses, retail cabinets, and vehicles. The operator needs to decide which retail cabinets to replenish in the future planning period, as well as the vehicle replenishment route. Warehouses and retail cabinets can both generate inventory costs and out-of-stock losses, and vehicles will generate path costs during the replenishment process. Therefore, the goal of the problem can be defined as minimizing the sum of these costs.

[0063] Through the above analysis, this embodiment provides a retail cabinet inventory path decision optimization method, such as Figure 1 As shown, the following steps are included:

[0064] (1) Establish an inventory path optimization model for retail counters. The inventory path optimization model takes the minimum sum of inventory cost, out-of-stock cost and vehicle path cost in the current planning period as the objective function, and takes replenishment constraint, maximum inventory level constraint, vehicle capacity constraint, retail counter flow balance constraint, vehicle closed loop constraint and sub-loop elimination constraint as constraint conditions.

[0065] Among them, the objective function of the inventory path optimization model is:

[0066]

[0067] In the formula, h irepresents the unit inventory cost, l i Indicates the unit out-of-stock loss, Respectively represent inventory levels The positive and negative parts of represents the remaining inventory level of retail counter i at the end of the t-th planning period, and serves as the initial inventory level of the t+1-th planning period. express The positive part, represents the remaining inventory level of retail counter i at the end of the t-1 planning period, c ij represents the path cost from retail counter i to retail counter j, represents a 0-1 variable indicating whether edge (i, j) is the route of vehicle k in period t. , it means that edge (i, j) is the route of vehicle k in period t, otherwise it is not; represents a 0-1 variable indicating whether retail cabinet i is replenished in the tth planning period. When , it means that retail cabinet i is replenished in the t planning period, otherwise it means that it is not; represents a 0-1 variable indicating whether vehicle k replenishes retail cabinet i in the tth planning period. When , it means that vehicle k replenishes retail counter i in the tth planning period, otherwise it means no; Represents a set of points, where vertex 0 represents the supply warehouse, n represents the number of retail cabinets, and other vertices Represents a collection of retail counter locations. represents a collection of vehicles, represents the set of planning periods, represents the demand for retail cabinet i in the tth planning period, represents the replenishment quantity received by retail cabinet i in the tth planning period, It represents the replenishment quantity of retail cabinet i by vehicle k in the tth planning period.

[0068] The replenishment constraints are specifically:

[0069]

[0070]

[0071] In the formula, C i represents the maximum inventory level of retail cabinet i.

[0072] The maximum inventory level constraint is specifically:

[0073]

[0074] In the formula, They represent the remaining inventory levels of the warehouse at the end of the t and t-1 planning periods, respectively. t represents the warehouse's incoming inventory in each period, C i represents the maximum inventory level of retail cabinet i;

[0075] The vehicle capacity constraint is specifically:

[0076]

[0077] In the formula, Q represents the capacity of the vehicle, represents a 0-1 variable indicating whether the k-th vehicle visits the warehouse in period t. If the vehicle visits the warehouse, then otherwise This constraint is used to limit the sum of the replenishment quantities of all retail counters on the route to be less than the vehicle capacity.

[0078] The retail counter flow balance constraint is specifically:

[0079]

[0080] In the formula, represents a 0-1 variable indicating whether the edge (j,i) is the route of vehicle k in period t. When , it means that the edge (j,i) is the route of vehicle k in period t, otherwise it is not. The flow balance constraint is used to constrain the out-degree and in-degree of retail cabinet i to be equal, that is, flow balance.

[0081] The vehicle closed-loop constraints are specifically:

[0082]

[0083]

[0084] In the formula, Respectively represent 0-1 variables used to indicate whether the edge (0,j) and (j,0) are the routes of vehicle k in period t. When the variable is 1, it means that the edge is its route, otherwise it is not. The 0 in (0,j) and (j,0) represents the warehouse. This constraint is used to limit each vehicle to start from warehouse 0 and return to warehouse 0 in each period.

[0085] The elimination of sub-ring constraints is specifically:

[0086]

[0087] In the formula, represents a subset of the retail cabinet set, represents a 0-1 variable indicating whether vehicle k replenishes retail cabinet m in the tth planning period. , it means that vehicle k replenishes retail cabinet m in the tth planning period, otherwise it means no. This constraint is used to eliminate Sub-ring

[0088] (2) Obtaining information related to the inventory path, including the warehouse location, the location of each retail counter, the initial inventory level of each retail counter in the current planning period, the maximum inventory level of each retail counter, the unit out-of-stock loss of each retail counter, and the demand and vehicle capacity of each retail counter in the current planning period.

[0089] (3) Based on the inventory path related information, the taboo search method is used to solve the inventory path optimization model to obtain the optimized inventory decision and path decision.

[0090] like Figure 3 As shown, this step specifically includes:

[0091] (3.1) Construct two empty taboo tables and They are used to store the retail cabinet and replenishment period tuples (i, t) respectively, and to initialize the disturbance length and the maximum number of iterations, where i represents the retail cabinet number and t represents the number of planning periods.

[0092] (3.2) Construct an initial solution and use the initial solution as the current solution.

[0093] The retail cabinet inventory routing problem requires obtaining: the inventory level of the retail cabinet and the warehouse in each period, the set of retail cabinet replenishment points in each period, the retail cabinet replenishment quantity in each period, and the replenishment route in each period. Let S represent the solution space of the retail cabinet inventory routing problem, and s represents a feasible solution in the solution space S. Then s consists of four parts, namely s = {y(s), q(s), i(s), R(s)}, represents the retail cabinet replenishment decision in solution s, represents the replenishment quantity decision in solution s, represents the inventory level decision in solution s, represents the path decision in solution s, Respectively represent the solution s After obtaining y(s), q(s), I(s), and R(s), we can get the set of retail cabinet replenishment points in period t: Collection of retail cabinet i replenishment period Replenishment And replenishment line The most critical components of the solution to the retail cabinet inventory routing problem are y(s) and R(s). Under the OU strategy, once y(s) is determined, q(s) and I(s) can be further calculated.

[0094] The initial solution construction method specifically includes:

[0095] (3.2.1) Adopting the greedy idea, the initial solution is constructed based on the principle of minimizing inventory cost. In this step, the influence of path cost is temporarily ignored, and the solution that minimizes inventory-related cost is sought. This is a problem of inventory allocation with deterministic demand. The following MIP model is solved directly by Cplex:

[0096]

[0097] (3.2.2) Through the heuristic method, according to the obtained Find the time when the delivery cost is the lowest At this time As a feasible solution. The purpose of this step is to If the vehicle path variables are already determined, Keeping the delivery cost as low as possible is equivalent to Considering that the MIP model itself is an NP-Hard problem, in order to ensure efficiency, the exact solution of CVRP is not sought during the construction process. Through heuristic methods and iterations, high-quality path decision quantities can be quickly obtained.

[0098] (3.2.3) The feasible solution is improved, and the improved solution is used as the initial solution. The above feasible solution is essentially a greedy solution. In (3.2.1), in order to ensure the minimum inventory cost, the replenishment frequency of the retail cabinets is very high, resulting in an extremely high path cost in (3.2.2). In the solution improvement stage, each retail cabinet in the replenishment plan is evaluated to estimate the impact on the total cost if they are deleted; at the same time, each retail cabinet that is not in the replenishment plan is evaluated to estimate the impact on the total cost if they are added to the replenishment plan. For example, deleting the replenishment of retail cabinet i in period t will reduce the total cost, then in the improved solution Then the paths are redistributed. The specific steps of solution promotion are described in detail below in step (3.7).

[0099] (3.3) Determine whether the termination condition is met, that is, whether the number of iterations is greater than the maximum number of iterations. If so, output the historical optimal solution; if not, execute (3.4).

[0100] (3.4) Use 4 different neighborhood operators to calculate the neighborhood solutions of the current solution s, and select the best solution from the domain solutions as the candidate solution according to the contempt rule. The contempt rule means that if the quality of the neighborhood solution is better than the historical optimal solution, and the operator operation is forbidden by the taboo table, then the taboo table is "disdained" and the current solution is updated to the neighborhood solution.

[0101] The four different neighborhood operators are Remove, Insert, Relocate and Replace. The four operators are as follows: Remove: Selects the retail counter i and the planning period t∈T i (s), order Then according to renew and R(s). Insert operator: select retail counter i and planning period make Then according to renew and R(s). Relocate operator: select retail counter i and planning period t1∈T i (s), planning period make Then according to renew and R(s). Replace operator: select retail counters i and j, planning period t1∈T i (s) and make Then according to renew and R(s). The schematic diagrams of the four operators are as follows Figure 2 shown.

[0102] After the neighborhood operator is used for calculation, the calculation of the change in total cost includes two parts: the change in inventory-related costs and the change in path costs.

[0103] The change in inventory-related costs is caused by the change in the replenishment period of retail cabinet i. The change in the replenishment status of retail cabinet i in period t will directly affect the inventory level of both retail cabinet i and the warehouse in each period after period t, thereby affecting the inventory-related costs. In the process of calculating cost changes, Recalculate changes in inventory levels at retail counters and warehouses, and thereby calculate changes in inventory-related costs.

[0104] The change of the path is caused by two reasons: first, because the retail cabinet i is deleted (inserted) in the planning period t, the path of the planning period t is shortened (extended); second, because the retail cabinet i is deleted (inserted) in the planning period t, the replenishment quantity after the t period changes, which may cause the original feasible path to become infeasible. Therefore, if we want to accurately calculate the change of the path cost after the retail cabinet i is deleted (inserted), it is equivalent to solving each CVRP after the t period, which is very time-consuming. In the actual calculation, an approximate cost scheme is adopted. If the retail cabinet i needs to be deleted from the planning period t, the predecessor node and the successor node of the retail cabinet i in the path are connected. If the retail cabinet i needs to be inserted into the planning period t, the CheapestInsert strategy of the retail cabinet i is executed on the path of the planning period t, that is, traverse each position of the path in the planning period t, so that the path cost increases the least after the retail cabinet i is inserted.

[0105] In addition to calculating the total cost change of the current solution s after the neighborhood operator operation, it is also necessary to determine the feasibility of the neighborhood operator. When executing the operator operation, the main consideration is whether the warehouse inventory level is negative and whether the retail cabinet i exceeds the vehicle capacity limit after being inserted into a certain path, but the vehicle capacity limit after the planning period t is not considered for the time being. This is because there will be solution improvement mechanism and disturbance mechanism in the iteration process, both of which will cause all paths to be replanned.

[0106] The Remove operator is described by a five-tuple opt1, which contains the retail counter i, the planned period t, the path number, the position in the path, and the approximate cost change after Remove. The Insert operator is described by a five-tuple opt2, which contains the retail counter i, the planned period t, the path number, the position in the path, and the approximate cost change after Insert. The Relocate operator is described by an eight-tuple opt3, which contains the retail counter i, the planned period t1, the path number, the position in the path, the planned period t2, the path number, the position in the path, and the approximate cost change after Relocate. The Replace operator is described by a six-tuple opt4, which contains the retail counter i, the planned period t1, the path number, the position in the path, the retail counter j that replaces the retail counter i, and the approximate cost change after Replace.

[0107] (3.5) Determine whether the candidate solution is better than the historical optimal solution. If not, execute (3.6). If yes, execute (3.7).

[0108] (3.6) Randomly perturb the candidate solution according to the perturbation length and execute (3.7).

[0109] De-perturbation is a common method in heuristic algorithms to jump out of local optimal solutions and increase the breadth of search. If there has been no improvement in the historical optimal solution for many generations during the algorithm iteration process, it means that the current solution is likely to be trapped in the dilemma of local optimality. In addition to increasing the taboo length, you can also directly perturb (Shake) the current solution to make a larger "jump" to the current solution. The perturbation method in this chapter requires the perturbation length. When the number of iterations in which the current solution has not been updated is equal to a multiple of the perturbation length, the perturbation step is entered. Generally, the perturbation direction in metaheuristic algorithms is random, but the de-perturbation of the taboo search algorithm in this chapter is to transfer the retail cabinet from the replenishment time of the current solution to the time when it is usually not replenished.

[0110] The de-perturbation can be achieved through the taboo table. and In fact, it stores the information of historical operator operations. Each element is initialized to 0. When the i-th retail counter in the k-th iteration changes from not replenishing to replenishing in the t-th period, the taboo table The (i, t) element of is assigned a value of k. In this way, when the disturbance needs to be resolved, only the taboo table needs to be By taking the minimum value for each row, we can obtain the time when the retail cabinet is least frequently visited.

[0111] (3.7) The candidate solution or the solution after random perturbation of the candidate solution is improved, including the improvement of the replenishment period of each retail counter and the improvement of the path plan, and the solution after the improvement is taken as the optimal solution and execute (3.8).

[0112] Solution boosting is applied to the three links of initial solution generation, neighborhood optimal solution improvement (when the neighborhood solution is better than the historical optimal solution), and solution perturbation. At these three links, the current solution can be considered to have entered a "new area", and applying solution boosting at this time will have a better effect. The purpose of solution boosting is to estimate some directions of improvement of the current solution on the basis of a feasible solution by analyzing the neighborhood of the current solution, and to improve the performance of the solution as much as possible. In terms of effect, solution boosting is more like a combination of multiple neighborhood search operators, which is a "leap" improvement of the current solution. According to the example experiments, the design of the solution boosting program can greatly accelerate the convergence of the taboo search algorithm to the optimal solution.

[0113] Solution improvement includes two parts: improving the replenishment period of each retail counter and improving the route plan.

[0114] Improve the replenishment period of each retail counter: add a two-dimensional array DeltaCost to record the change in the replenishment period to the total cost. Then the element DeltaCost[i][j] of DeltaCost records if The estimated change in total cost when Then DeltaCost[i][j] records if Then, we get the element with the smallest negative number from each row of DeltaCost, extract the value of the index number (i, t) of the element, and record it as (i0, t0). If the solution before the improvement is but The value after promotion If the previous solution is not improved but The value after promotion Finally, according to the OU policy Synchronously promote and get the promoted value The OU strategy means that each time you replenish the inventory of retail cabinet i, you directly replenish the inventory of retail cabinet i to the maximum inventory level C. i ;

[0115] (3.7.2) According to the improved Re-plan the path and get the upgrade Value After the improvement As the solution after the current solution s is improved. Because the solution improvement may change the replenishment period of multiple retail counters at the same time, making the path cost estimation method using the previous operator no longer accurate, it is necessary to re-plan the path R(s) for the new y(s) in the solution improvement. Another function of re-planning the path is to re-estimate the path cost of the current solution, because the approximate path cost estimate is used when executing the operator operation. After multiple iterations, it is likely to cause a large deviation in the path cost. Re-planning the path solves this problem.

[0116] (3.8) Update the taboo table based on the current solution and the optimal solution.

[0117] The specific steps of updating are: the optimal solution of solution s is adopted as solution s , Indicates that if retail cabinet i is replenished in the tth planning period in solution s, but in solution s , If retail counter i is not replenished in the tth planning period, the tuple (i, t) is added to the taboo table In the solution s, it means that retail cabinet i is prohibited from replenishing in the tth period in the next several iterations; if retail cabinet i is not replenished in the tth planning period in the solution s, but in the solution s , In the tth planning period, retail cabinet i is replenished, then the tuple (i, t) is added to the taboo table , which means that in the next several iterations, retail cabinet i will remain in the tth period for replenishment; among them, the taboo lengths of the two taboo tables are Where L and λ are fixed parameters, n is the number of retail cabinets, |R(s)| is the number of feasible paths in solution s, is 0 to Randomly pick a number between.

[0118] (3.9) Increase the number of iterations by 1 and return to execute step (3.3).

[0119] The embodiment of the present invention provides a method for optimizing retail cabinet inventory path decision. By establishing a mathematical model of the inventory path problem and jointly optimizing the replenishment decision and path planning, the operating cost can be reduced. The taboo search algorithm proposed in the present invention can greatly improve the solution efficiency. In addition, the results of the present invention can be extended to other actual IRP scenarios, promote enterprise supply chain optimization, and further reduce enterprise operating costs.

[0120] Embodiment 2

[0121] The embodiment of the present invention provides a computer device, which provides services for implementing the method of the first embodiment of the present invention. The device may include: a memory storing a computer executable program; a processor coupled to the memory 301; the processor calls the computer executable program stored in the memory to execute the steps in the method described in the first embodiment.

[0122] The memory may include computer system readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory. The device may further include other removable / non-removable, volatile / non-volatile computer system storage media. By way of example only, the memory may be used to read and write non-removable, non-volatile magnetic media (commonly referred to as "hard drives"). A program / utility having a set (at least one) of program modules may be stored in, for example, the memory, such program modules including but not limited to an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment. The computer executable program of the program module typically performs the functions and / or methods in the embodiments described in the present invention.

[0123] The processor executes various functional applications and data processing by running the program stored in the memory, such as implementing the method provided in the first embodiment of the present invention.

[0124] The code of the computer executable program can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" language or similar programming languages.

[0125] It should be understood that the above embodiments and descriptions only describe the principles, main features and advantages of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, and these changes and improvements all fall within the scope of protection of the present invention.

Claims

1. A retail cabinet inventory path decision optimization method, characterized in that: The steps include: (1) Establishing a retail counter inventory path optimization model, wherein the objective function of the inventory path optimization model is to minimize the sum of inventory cost, out-of-stock cost and vehicle path cost in the current planning period, and the constraints are replenishment constraint, maximum inventory level constraint, vehicle capacity constraint, retail counter flow balance constraint, vehicle closed loop constraint and sub-loop elimination constraint; wherein the objective function of the inventory path optimization model is: In the formula, h i represents the unit inventory cost, l i Indicates the unit out-of-stock loss, Respectively represent inventory levels The positive and negative parts of represents the remaining inventory level of retail counter i at the end of the t-th planning period, and serves as the initial inventory level of the t+1-th planning period. express The positive part, represents the remaining inventory level of retail counter i at the end of the t-1 planning period, c ij represents the path cost of edge (i,j), represents a 0-1 variable indicating whether edge (i, j) is the route of vehicle k in period t. , it means that edge (i, j) is the route of vehicle k in period t, otherwise it is not; represents a 0-1 variable indicating whether retail cabinet i is replenished in the tth planning period. When , it means that retail cabinet i is replenished in the t planning period, otherwise it means that it is not; represents a 0-1 variable indicating whether vehicle k replenishes retail cabinet i in the tth planning period. When , it means that vehicle k replenishes retail counter i in the tth planning period, otherwise it means no; represents a set of points, where vertex 0 represents the warehouse, n represents the number of retail cabinets, Represents a collection of retail counter locations. represents a collection of vehicles, represents the set of planning periods, represents the demand for retail cabinet i in the tth planning period, represents the replenishment quantity received by retail cabinet i in the tth planning period, represents the replenishment quantity of vehicle k to retail cabinet i in the tth planning period; (2) Obtaining information related to the inventory path, including the warehouse location, the location of each retail counter, the initial inventory level of each retail counter in the current planning period, the maximum inventory level of each retail counter, the unit out-of-stock loss of each retail counter, and the demand and vehicle capacity of each retail counter in the current planning period; (3) Based on the inventory path related information, the taboo search method is used to solve the inventory path optimization model to obtain the optimized inventory decision and path decision.

2. The retail cabinet inventory path decision optimization method according to claim 1, characterized in that: The replenishment constraints are specifically: In the formula, C i represents the maximum inventory level of retail cabinet i.

3. The retail cabinet inventory path decision optimization method according to claim 1, characterized in that: The maximum inventory level constraint is specifically: In the formula, They represent the remaining inventory levels of the warehouse at the end of the t and t-1 planning periods, respectively. t represents the warehouse's incoming inventory in each period, C i represents the maximum inventory level of retail cabinet i; The vehicle capacity constraint is specifically: In the formula, Q represents the capacity of the vehicle, represents a 0-1 variable indicating whether the k-th vehicle visits the warehouse in period t. If the vehicle visits the warehouse, then otherwise 4. The retail cabinet inventory path decision optimization method according to claim 1, characterized in that: The retail counter flow balance constraint is specifically: In the formula, represents a 0-1 variable indicating whether the edge (j,i) is the route of vehicle k in period t. When , it means that edge (j,i) is the route of vehicle k in period t, otherwise it is not; The vehicle closed-loop constraints are specifically: In the formula, They are 0-1 variables used to indicate whether the edge (0,j) and (j,0) are the routes of vehicle k in period t. When the variable is 1, it means that the edge is its route, otherwise it is not. The 0 in (0,j) and (j,0) indicates a warehouse. represents a 0-1 variable indicating whether the k-th vehicle visits the warehouse in period t. If the vehicle visits the warehouse, then otherwise 5. The retail cabinet inventory path decision optimization method according to claim 1, characterized in that: The elimination of sub-ring constraints is specifically as follows: In the formula, represents a subset of the retail cabinet set, represents a 0-1 variable indicating whether vehicle k replenishes retail cabinet m in the tth planning period. , it means that vehicle k replenishes retail cabinet m in the tth planning period, otherwise it means no.

6. The retail cabinet inventory path decision optimization method according to claim 1, characterized in that: Step (3) specifically includes: (3.1) Construct two empty taboo tables and They are used to store the retail cabinet and replenishment period tuples (i, t) respectively, and initialize the perturbation length and the maximum number of iterations, where i represents the retail cabinet number and t represents the number of planning periods; (3.2) Construct an initial solution and use the initial solution as the current solution, where the solution structure is represents the retail cabinet replenishment decision in solution s, represents the replenishment quantity decision in solution s, represents the inventory level decision in solution s, represents the path decision in solution s, Respectively represent the solution s The value of (3.3) Determine whether the termination condition is met, that is, whether the number of iterations is greater than the maximum number of iterations. If so, output the historical optimal solution; if not, execute (3.4); (3.4) Calculate the neighborhood solution of the current solution using four different neighborhood operators, and select the best solution from the domain solutions as the candidate solution according to the contempt criterion; wherein the four different domain operators are specifically Remove operator, Insert operator, Relocate operator and Replace operator; (3.5) Determine whether the candidate solution is better than the historical optimal solution. If not, execute (3.6). If yes, execute (3.7). (3.6) Randomly perturb the candidate solution according to the perturbation length and execute (3.7); (3.7) Improve the candidate solution or the solution after random perturbation of the candidate solution, including the improvement of the replenishment period of each retail cabinet and the improvement of the path plan, and take the improved solution as the optimal solution and execute (3.8); (3.8) Update the taboo table according to the current solution and the optimal solution; (3.9) Increase the number of iterations by 1 and return to execute step (3.3).

7. The retail cabinet inventory path decision optimization method according to claim 6, characterized in that: Step (3.2) specifically includes: (3.2.1) Without considering the impact of path cost, solve the following model to obtain the minimum inventory cost (3.2.2) Through the heuristic method, according to the obtained Find the time when the delivery cost is the lowest At this time As a feasible solution; (3.2.3) The feasible solution is improved, and the improved solution is used as the initial solution.

8. The retail cabinet inventory path decision optimization method according to claim 6, characterized in that: Step (3.7) specifically includes: (3.7.1) Improve the replenishment period of each retail counter: Add a two-dimensional array DeltaCost to record the change in the total cost caused by the change in the replenishment period. Then the element DeltaCost[i][t] of DeltaCost records if The estimated change in total cost from 0 to 1 is Then DeltaCost[i][t] records if Estimate the change in total cost from 1 to 0; then get the element with the smallest negative number from each row of DeltaCost, extract the value of the index number (i, t) of the element, and record it as (i0, t0). If the solution before the improvement is represents a 0-1 variable in solution s that indicates whether retail cabinet i0 is replenished in the planning period t0, then The value after promotion If the previous solution is not improved but The value after promotion Finally, according to the OU policy Synchronously promote and get the promoted value (3.7.2) According to the improved Re-plan the path and get the upgrade Value After the improvement As the solution after the solution s is improved.

9. The retail cabinet inventory path decision optimization method according to claim 6, characterized in that: Step (3.8) specifically includes: The optimal solution of solution s is adopted as solution s , Indicates that if retail cabinet i is replenished in the tth planning period in solution s, but in solution s , If retail counter i is not replenished in the tth planning period, the tuple (i, t) is added to the taboo table , which means that retail cabinet i is prohibited from replenishing in period t in the following iterations; If in solution s, retail cabinet i is not replenished in the tth planning period, but in solution s , In the tth planning period, retail cabinet i is replenished, then the tuple (i, t) is added to the taboo table , which means that retail cabinet i will maintain replenishment in period t in the next several iterations; Among them, the taboo lengths of the two taboo tables are where L and λ are fixed parameters, |R(s)| is the number of feasible paths in solution s, is 0 to Randomly pick a number between.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: The processor executes the computer program to implement the method according to any one of claims 1 to 9.

Images