A two-stage scheduling method for intelligent warehouse material picking based on genetic algorithm with embedded neighborhood search strategy

Abstract

The present invention discloses a two-stage scheduling method for intelligent warehousing material picking based on a genetic algorithm embedded with a neighborhood search strategy, which relates to the field of intelligent logistics scheduling and is used to solve the warehousing material picking scheduling problem in the production logistics link of an enterprise. The present invention includes: considering the heterogeneous characteristics of production lines, establishing a two-stage optimization model for order task batching and group batch path optimization with the minimum total cost of picking order delay as the objective function, calling a corridor-type warehousing picking path strategy, proposing 4 kinds of neighborhood search strategies, and improving the traditional genetic algorithm for problem solving by embedding the neighborhood search strategy. This method realizes the automation of batching and group batch path planning of orders to be picked, minimizes the total cost of order picking delay, and can solve the two-stage scheduling problem of "order batching-path planning" of warehousing picking in the process of enterprise production material supply, thereby improving the comprehensive benefits of production logistics and achieving the purpose of reducing costs and increasing efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent logistics scheduling, and in particular to a two-stage scheduling method for intelligent warehousing material picking based on a genetic algorithm with an embedded neighborhood search strategy. Background Art

[0002] As an indispensable and important link in the production process of modern enterprises, material supply plays a vital role in improving the overall production efficiency of enterprises. The efficiency of warehouse material supply is directly related to the efficiency of workshop production lines. If the efficiency of warehouse material supply is not high, it will cause major production accidents such as production line shutdown. Therefore, how to improve the efficiency of warehouse material supply in the material preparation stage is a pain point that needs to be solved urgently.

[0003] In the process of modern warehousing material supply, the most time-consuming link is warehouse picking, which accounts for 50%-65% of the operating costs of warehousing. In traditional man-to-man picking warehousing, pickers will spend more than 50% of their working time on repeated walking on the picking path. Due to the lack of efficient and scientific picking solutions, this is a huge waste and unreasonable use of resources such as manpower, equipment, and funds. Solving the warehouse picking problem includes two important modules: material order batching and picking path planning. Among them, each material demand order contains a variety of material types and corresponding quantities. Each type of material is stored on the designated shelves of the warehouse. The material order batching determines different combinations of shelves to be visited; due to the particularity of the layout of warehousing logistics, the methods adopted by existing research include "S" type, maximum gap type, return type path strategy, etc. The picking path strategy determines the order of shelf access paths under known shelf combinations. Picking order batching is the prerequisite for picking route planning, and picking route planning will affect the determination of picking order batch combinations. The two restrict each other. Therefore, to improve the efficiency of warehousing material supply in the production logistics stage, joint scheduling optimization of the two stages of "order batching-route planning" is required. Summary of the invention

[0004] In order to solve the technical problems existing in the above-mentioned prior art, the present invention proposes a two-stage scheduling method for intelligent warehousing material picking based on a genetic algorithm embedded with a neighborhood search strategy, which can optimize the warehousing material order picking method in the enterprise's production logistics link, improve the warehousing material supply efficiency, reduce warehousing operation costs, and thereby improve the comprehensive benefits of the enterprise's production logistics.

[0005] In order to achieve the above technical objectives, the present invention provides the following technical solutions:

[0006] A two-stage scheduling method for intelligent warehouse material picking based on genetic algorithm with embedded neighborhood search strategy, specifically including:

[0007] S1. Obtain material demand order information from each workstation of the production line in the manufacturing workshop;

[0008] S2. Establishing the objective function of the corresponding order picking task according to the acquired material demand order information;

[0009] S3. Considering the heterogeneity of production lines, a two-stage joint planning model of material picking "order batching-path planning" is established, and constraints are set;

[0010] S4, genetic algorithm embedded in neighborhood search strategy based on “S” type path strategy design;

[0011] S5. Solve the enterprise material picking simulation instance and obtain the material picking scheduling plan.

[0012] Furthermore, step S1 is specifically as follows: collecting material demand order information of each workstation of the production line in the enterprise manufacturing workshop within a specified time period, wherein the material demand order information includes: material delivery period information, production line category information to which the workstation belongs, material category and its demand quantity information.

[0013] Furthermore, step S2 is specifically as follows:

[0014] The material orders of different production line stations are set to be picked as a batch during warehouse material picking. At the same time, the heterogeneity of the production lines is taken into account. The total delay cost of material order picking is minimized as the optimization goal, and the objective function of the corresponding order picking task is established. Its formula is expressed as follows:

[0015]

[0016] Among them, z is the objective function of the corresponding order picking task; p k represents the unit time delay cost of the k-th production line order, t o represents the picking and delivery delay time of order o, g ok represents a 0-1 variable. If order o belongs to the k-th production line, g ok =1, otherwise 0; K is the set of all production line categories, and O is the set of all orders.

[0017] Furthermore, when establishing the two-stage joint planning model of material picking "order batching-path planning" in step S3, the following settings are made in advance:

[0018] 1) Each picker has the same speed and maximum load, regardless of the picker path congestion;

[0019] 2) Not considering the service time for each order;

[0020] 3) Materials belonging to the same order will not be split into different batches;

[0021] 4) A batch can only be assigned to one picker to complete the picking, and the picking process of each batch is not allowed to be interrupted;

[0022] 5) Ignoring the volume waste caused by the gaps between goods when they are placed, that is, assuming that the goods are placed closely together;

[0023] 6) The warehouse shelf storage strategy is centralized storage, that is, materials with the same material number are stored in the same storage location;

[0024] 7) Unexpected situations such as out-of-stock shelves or inserted orders are not taken into account.

[0025] Based on the above settings, more specifically, the constraints of the joint planning model set in step S3 specifically include:

[0026] Constraint 1: Order assignment principle, which limits each order to one and only one batch, is expressed as:

[0027]

[0028] Constraint 2: Batch continuity constraint, which stipulates that for any batch, if the batch is selected, the batch with the previous number is also selected. The formula is:

[0029]

[0030] Constraint 3: Batch load constraint, which limits the total weight of each order in the same batch to not exceed the maximum load of the batch. The formula is:

[0031]

[0032] Constraints 1-3 are used to solve the order batching problem; in the above three formulas, B represents the set of all feasible batches, B = {1, ..., b, ... |B|}, \{1} means that it does not include the value of 1, O represents the set of all orders, O = {1, ..., o, ... |O|}; q o represents the weight of the picking order, Q represents the maximum load of the picker batch; u ob is a binary variable, u ob =1 means order o is assigned to batch b for combined picking, otherwise it is 0;

[0033] Constraint 4: Batch allocation constraint, which limits each batch to be processed at most once in a certain picking sequence of a certain picker; if the batch is empty, it should not be allocated to any picking sequence of any picker; at the same time, it limits any picker to process at most one batch in any picking sequence; it is expressed by the following three formulas in sequence:

[0034]

[0035]

[0036]

[0037] Constraint 5: Batch uniqueness constraint, which stipulates that for any production line order, the order is processed only in a certain picking sequence of a certain picker. The formula is:

[0038]

[0039] Constraint 6: The picker’s picking order continuity constraint stipulates that for any picking order of any picker, if the current picking order is assigned a batch, the previous order of the order is also assigned a batch. The formula is expressed as:

[0040]

[0041] Constraints 4-6 are used to solve the batch assignment problem; in the above five equations, M is the set of all pickers, M = {1,…,m,…|M|}, N is the set of all picking orders of a single picker, N = {1,…,n,…|N|}; y bmn is a binary variable, y bmn =1 indicates that batch b is processed in the picking sequence n of picker m, otherwise it is 0;

[0042] Constraint 7: Shelf point access rule, which stipulates that for any order, all shelf points involved in the order must be visited in the batch path; if the batch visits a shelf, there must be at least one order in the batch that contains the required materials on the shelf; represented by the following two formulas:

[0043]

[0044]

[0045] Constraint 8: Batch path node association constraint, which stipulates that for any batch, if the batch visits a shelf point, then the shelf point must exist as a batch path node. The formula is expressed as:

[0046]

[0047] Constraint 9: Batch path node access constraint, which limits any node in each batch path to be visited at most once. The formula is:

[0048]

[0049] Constraint 10: Path elimination sub-loop rule, eliminate the sub-loop situation in path planning, the formula is expressed as:

[0050]

[0051] Constraints 7-10 are used to solve the batch routing problem. In the above five equations, V represents the set of all shelf points, V = {0, 1, ..., s ..., |V|}, V = {0} represents the sorting station, and \{0} means that it does not include the value of 0; z bs is a binary variable, z bs =1 means batch b visits shelf s\{0}, otherwise it is 0; h os is a known parameter, h os =1 means that order o contains items to be picked on shelf s, otherwise it is 0; x bij is a binary variable, x bij =1 means that batch b visits path node j immediately after visiting path node i, otherwise it is 0; |S| is the number of path nodes in the sub-circuit involved in the batch;

[0052] Constraint 11: The order constraint of batch picking. For any picker, the completion time of the first batch picked by the picker is later than or equal to the sum of the picking initialization time and the time consumed in the picking process; and for any batch of any picker, the completion time of the batch picking is later than or equal to the sum of the completion time of the previous batch picking and the time consumed in the batch picking. The formula is expressed as:

[0053]

[0054]

[0055] Constraint 12: Batch picking completion time constraint, which stipulates that for any batch, if the batch is assigned to a certain picking sequence of a certain picker, the batch completion time shall not be less than the batch picking completion time of the picker in the sequence. The formula is expressed as:

[0056]

[0057] Constraint 13: Production line order picking completion time constraint, which stipulates that for any order, if the order belongs to a certain batch, the order picking completion time shall not be less than the batch picking completion time. The formula is:

[0058]

[0059] Constraint 14: Calculate the delay time of each material demand order. The formula is:

[0060]

[0061] Constraints 11-14 are used to solve the problem of picking time continuity and delay time calculation; in the above five equations, d ij is a known parameter, representing the distance between path node i and path node j; β o represents the delivery date of the material corresponding to order o, A is a constant; w mn is a continuous variable, indicating the completion time of the batch picking of picker m on the picking batch sequence n, c b represents the picking completion time of batch b, e o represents the picking completion time of production line order o, t o Indicates the delay time of order o.

[0062] Furthermore, step S4 is specifically as follows:

[0063] An improved genetic algorithm embedded with a neighborhood search strategy is designed, and the "S" type warehouse path picking strategy is called to solve the two-stage joint planning model of material picking "order batching-path planning" established in step S3, including the following steps:

[0064] S41, population initialization; set the evolutionary generation counter t=0, the maximum evolutionary generation T, and randomly generate M chromosome individuals as the initial population P(0); the chromosome encoding method is real number encoding, the orders are numbered in sequence, each gene in the chromosome represents an order, and the gene number is the corresponding order number;

[0065] S42, individual evaluation; calculate the fitness value f(x) of each individual in the population P(t) to determine the quality of the individuals in the population; and adopt an "S" type path strategy for the batch path after chromosome decoding;

[0066] S43, determine whether the algorithm termination condition is met; if not, execute step S64; if it is met, output the current optimal chromosome and decode it into a batch allocation plan; when the current evolutionary generation t reaches the preset maximum evolutionary number T, the termination condition is considered to be met;

[0067] S44, selection operation; applying the selection operator to the group, and based on the fitness evaluation of the individuals in the group, adopting the tournament selection method to obtain the parent chromosome;

[0068] S45, crossover operation: applying the crossover operator to the population, using a sequential crossover method, retaining the better gene segments of the parent generation, and obtaining chromosomes of the offspring generation;

[0069] S46, mutation operation: applying the mutation operator to the population, changing the gene values ​​at certain gene positions of individual chromosomes in the population, and obtaining mutated offspring chromosomes;

[0070] S47, neighborhood search; perform neighborhood search operations on all individuals x in the current population through different neighborhood search operators; initialize the current individual x as the optimal individual, if the fitness f(x′) of individual x′ in the current neighborhood structure is better than f(x), then obtain a new optimal individual x←x′, otherwise the current optimal individual enters the next neighborhood structure for search until all neighborhood structures are traversed; the population P(t)^ is obtained after selection operation, crossover operation, mutation operation, and neighborhood search;

[0071] S48, perform environmental selection operation to form a new population; merge the parent and child populations and retain the better individuals; that is, merge the population P(t) with the offspring population P(t)^ to form a merged population with a population size of M*2, calculate the fitness of the individuals in the merged population, arrange them in descending order, select the first M chromosome individuals with smaller fitness values ​​to form a new population P(t)*, and enter the next iteration, and the number of iterations is increased by 1; repeat steps S43-S48 until the genetic algorithm iteration is completed.

[0072] More specifically, the different neighborhood search operators described in step S47 specifically include:

[0073] a) Order swap operator: randomly select two gene loci from the current chromosome, compare the delivery dates of the orders represented by the genes at the gene loci, and if the delivery date of the order represented by the gene at the front position is later than the delivery date of the order represented by the gene at the back position, swap the gene positions of the two;

[0074] b) Order insert operator: randomly select two batches from the decoding scheme of the current chromosome, extract the gene loci corresponding to the two batches in the chromosome and the minimum order weight in their batches respectively, if the minimum order weight in the batch with the later gene loci plus the batch weight of the batch with the earlier gene loci can meet the batch load constraint, then delete the minimum weight order from the batch with the later gene loci and insert it into the batch with the earlier gene loci to form a new batch;

[0075] c) Batch sort operator: for the decoding scheme of the current chromosome, traverse all its batches, obtain the orders and corresponding order delivery dates contained in each batch, and calculate the average delivery date of the batches; sort the batches arranged on the current chromosome in ascending order according to the delivery date to obtain a new chromosome;

[0076] d) Batch swap operator: select two batches in the current chromosome decoding scheme, and obtain the gene positions represented by the batches in the chromosome. The urgency of the batch is defined as the difference between the completion time of the batch and the average delivery time. If the urgency of the batch with the gene position in the front is less than the urgency of the batch with the gene position in the back, the gene position is swapped.

[0077] Furthermore, step S5 specifically includes:

[0078] S51. Obtain basic information on the enterprise's warehouse layout, shelf specifications, storage methods, and staffing;

[0079] S52, obtaining material demand order information for each workstation of the production line within a specified time period of the workshop;

[0080] S53, applying the genetic algorithm described in step S4 to the joint planning model established in step S3 to solve the case;

[0081] S54. Obtain a warehouse material picking scheduling solution solved by the example.

[0082] Based on the above technical solution, the method proposed by the present invention has the following beneficial effects:

[0083] The present invention proposes a two-stage scheduling method for intelligent warehousing material picking based on a genetic algorithm embedded with a neighborhood search strategy, and establishes a two-stage joint planning model of material picking "order batching-path planning" with the total delay cost of material order picking as the objective function. The model fully considers the heterogeneous characteristics of production lines and order delivery periods. Compared with the existing research technology that considers order batching or picking path planning alone, the model proposed by the present invention better meets the diversified manufacturing material picking needs of modern manufacturing enterprises, is more practical, and has higher comprehensive scheduling efficiency.

[0084] The present invention designs a genetic algorithm embedded with a neighborhood search strategy for the problem of warehouse material picking, and calls the "S" type warehouse path picking strategy to solve the model, wherein crossover, mutation, and neighborhood search strategies suitable for this problem are designed, which can improve the solution accuracy and efficiency, realize automatic batching and path planning of material orders, and minimize the total delay cost of material picking; the neighborhood search strategy screens and optimizes individuals in each generation of the population from four different angles of reducing order delay time (i.e., four neighborhood search operators) in the algorithm, which is more flexible, accurate, and efficient than the iterative optimization effect of traditional genetic algorithms, so that the algorithm as a whole can better solve the material picking problem in the production logistics supply process, thereby improving material supply efficiency, reducing warehouse operation costs, and realizing the improvement of the benefits of enterprise production logistics. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments will be briefly described below. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0086] Figure 1 It is a schematic diagram of the overall solution of the present invention;

[0087] Figure 2 A warehouse layout diagram of a specific example provided by an embodiment of the present invention;

[0088] Figure 3 A flow chart of the neighborhood search algorithm used in the present invention;

[0089] Figure 4 A path planning diagram of a specific example provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0090] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0091] Although the steps in the present invention are arranged with numbers, they are not used to limit the order of the steps. Unless the order of the steps is clearly stated or the execution of a certain step requires other steps as a basis, the relative order of the steps can be adjusted. It is understood that the term "and / or" used in this article involves and covers any and all possible combinations of one or more of the associated listed items.

[0092] In this embodiment, for the two-stage scheduling problem of "order batching-route planning" for warehouse material picking, an optimization model is established with the minimum total delay cost of picking orders as the optimization goal, and the gurobi solver is used to fully verify the effectiveness of the model; in the case of small-scale data, the gurobi solver can give an accurate solution, but as the data scale increases, the solution efficiency of the gurobi solver is far from meeting the actual production needs of the enterprise. Therefore, an improved genetic algorithm embedded with a neighborhood search strategy is designed to solve this problem, and automatic batching and path planning of material picking orders in large-scale scenarios are achieved while taking into account both solution efficiency and solution quality.

[0093] Please refer to Figure 1 , showing that the present invention proposes a two-stage scheduling method for intelligent warehousing material picking based on a genetic algorithm embedded with a neighborhood search strategy, which specifically includes the following steps:

[0094] S1. Obtain material demand order information from each workstation of the production line in the manufacturing workshop;

[0095] As a preferred implementation of step S1, step S1 is specifically:

[0096] The material demand order information of each workstation of the production line in the enterprise manufacturing workshop is collected within a specified time period, and the material demand order information includes: material delivery period information, production line category information to which the workstation belongs, material category and its demand quantity information.

[0097] S2. Establishing the objective function of the corresponding order picking task according to the acquired material demand order information;

[0098] As a preferred implementation of step S2, step S2 is specifically:

[0099] The material orders of different production line stations are set to be picked as a batch during warehouse material picking. At the same time, the heterogeneity of the production lines is taken into account. The total delay cost of material order picking is minimized as the optimization goal, and the objective function of the corresponding order picking task is established. Its formula is expressed as follows:

[0100]

[0101] Among them, z is the objective function of the corresponding order picking task; p k represents the unit time delay cost of the k-th production line order, t o represents the picking and delivery delay time of order o, g ok represents a 0-1 variable. If order o belongs to the k-th production line, g ok =1, otherwise 0; K is the set of all production line categories, and O is the set of all orders.

[0102] In this embodiment, the storage environment is as follows: Figure 2 The single-area aisle warehouse is set up as follows: It includes a single sorting station, 10 rows of shelves, 15 shelves in each row, a total of 150 shelves, each shelf is 0.8 meters long and 0.6 meters wide, and there are single-row shelves on the upper and lower sides of the warehouse. In addition, they are all back-to-back shelves, and there is no difference between each shelf; there are 7 aisles, including horizontal aisles and longitudinal aisles. The width of the horizontal aisle is 1.2 meters, and the width of the longitudinal aisle is 1.5 meters. There is a longitudinal aisle in front and behind the warehouse, which is convenient for pickers to walk between the horizontal aisles. In this example, there are 3 pickers, the picking speed is 1m / s, and their picking load is fixed at 23kg. There is no difference in their work skill level and walking picking speed, and the impact of their picking action time on the delay time of the picking order delivery is not considered. The picking path adopts the "S" path strategy. Based on this warehousing environment setting, the material demand order information from each workstation of the workshop production line in S1 includes material type, material quantity, order weight, and material demand delivery time.

[0103] S3. Considering the heterogeneity of production lines, a two-stage joint planning model of material picking "order batching-path planning" is established, and constraints are set;

[0104] As a preferred implementation of step S3, step S3 specifically includes:

[0105] When establishing the two-stage joint planning model of "order batching-route planning" for material picking, the following settings are made in advance:

[0106] 1) Each picker has the same speed and maximum load, regardless of the picker path congestion;

[0107] 2) Not considering the service time for each order;

[0108] 3) Materials belonging to the same order will not be split into different batches;

[0109] 4) A batch can only be assigned to one picker to complete the picking, and the picking process of each batch is not allowed to be interrupted;

[0110] 5) Ignoring the volume waste caused by the gaps between goods when they are placed, that is, assuming that the goods are placed closely together;

[0111] 6) The warehouse shelf storage strategy is centralized storage, that is, materials with the same material number are stored in the same storage location;

[0112] 7) Do not consider unexpected situations such as out-of-stock shelves or inserted orders;

[0113] Based on the above settings, the following 14 constraints are further imposed on the two-stage joint planning model:

[0114] Constraint 1: Order assignment principle, the formula is expressed as:

[0115]

[0116] Constraint 2: Batch continuity constraint, the formula is expressed as:

[0117]

[0118] Constraint 3: Batch load constraint, the formula is expressed as:

[0119]

[0120] Constraint 4: Batch allocation constraint, the formula is expressed as:

[0121]

[0122]

[0123]

[0124] Constraint 5: Batch uniqueness constraint, the formula is:

[0125]

[0126] Constraint 6: The picker's picking order continuity constraint, the formula is expressed as:

[0127]

[0128] Constraint 7: Shelf point access rule, the formula is expressed as:

[0129]

[0130]

[0131] Constraint 8: Batch path node association constraint, the formula is expressed as:

[0132]

[0133] Constraint 9: Batch path node access constraint, the formula is expressed as:

[0134]

[0135] Constraint 10: Path elimination sub-loop rule, the formula is expressed as:

[0136]

[0137] Constraint 11: The order constraint of batch picking is expressed as:

[0138]

[0139]

[0140] Constraint 12: Batch picking completion time constraint, the formula is expressed as:

[0141]

[0142] Constraint 13: Production line order picking completion time constraint, the formula is expressed as:

[0143]

[0144] Constraint 14: Calculate the delay time of each material demand order. The formula is:

[0145]

[0146] In this embodiment, the above 14 constraints address different issues and have different functions, specifically:

[0147] Constraints 1-3 mainly solve the problem of order batching; in the formula, B represents the set of all feasible batches, B = {1, ..., b, ... | B |}, \ {1} means that it does not include the value of 1, O represents the set of all orders, O = {1, ..., o, ... | O |}; q o represents the weight of the picking order, Q represents the maximum load of the picker batch; u ob is a binary variable, u ob =1 means order o is assigned to batch b for combined picking, otherwise it is 0.

[0148] Constraint 1 ensures that each picking order will be assigned to a certain batch; Constraint 2 ensures the serial number continuity of the order allocation batches to avoid the occurrence of empty batches; Constraint 3 ensures that the total weight of the orders contained in each batch does not exceed the maximum load capacity of the picker.

[0149] Constraints 4-6 mainly solve the batch assignment problem; in the formula, M is the set of all pickers, M = {1,…,m,…|M|}, N is the set of all possible batch orders of a single picker, N = {1,…,n,…|N|}; y bmn is a binary variable, y bmn =1 indicates that batch b is accepted in batch picking sequence n of picker m, otherwise it is 0.

[0150] Constraint 4 ensures that each batch can only be accepted once in a batch picking sequence of a picker, and ensures that no empty batches are allowed to be processed by the picker, causing the picker's picking task to be idle, and stipulates that each picker can only process at most one batch in each batch picking sequence at the same time; Constraint 5 ensures that non-empty combined order batches need to be processed by the picker and only processed once; Constraint 6 ensures the continuity of the picker's batch processing to avoid the occurrence of empty batch numbers.

[0151] Constraints 7-10 mainly solve the batch path problem. V represents the set of all shelf points, V = {0, 1, ..., s ..., |V}, V = {0} represents the sorting station, \{0} means that it does not include the value of 0; z bs is a binary variable, z bs =1 means batch b visits shelf s\{0}, otherwise it is 0; h os is a known parameter, h os =1 means that order o contains items to be picked on shelf s, otherwise it is 0; x bij is a binary variable, x bij =1 means that batch b visits path node j immediately after visiting path node i, otherwise it is 0.

[0152] Constraint 7 ensures that all shelf points included in the orders in a non-empty batch are visited in a certain batch, and that all shelf nodes in the batch access path can be found in the included orders; Constraint 8 ensures that the shelf points to be visited in each batch exist as actual access path nodes; Constraint 9 limits any node in each batch path to be visited at most once; Constraint 10 eliminates sub-loops in path planning.

[0153] Constraints 11-14 mainly solve the problems of picking time continuity and delay time calculation. ij is a known parameter, representing the distance between path node i and path node j; w mn is a continuous variable, indicating the completion time of the batch picking of picker m on the picking batch sequence n, c b represents the picking completion time of batch b, e o represents the picking completion time of production line order o, t o Indicates the delay time of order o.

[0154] Constraint 11 ensures the time sequence of the picker's batch picking process; Constraint 12 ensures that the completion time of each batch picking is consistent with the picker's sequential picking time of each batch; Constraint 13 ensures that the completion picking time of the orders included in the batch is consistent with the batch picking completion time; Constraint 14 calculates the delay time of each material demand order.

[0155] In addition, the present embodiment also provides the value range of each parameter in the above formula, as shown below:

[0156]

[0157]

[0158]

[0159] c b ≥0,b∈B;

[0160]

[0161]

[0162]

[0163]

[0164] Since the warehouse material picking scheduling problem is an NP-Hard problem, as the scale of the problem increases, it is difficult for the exact solution method to obtain the optimal solution in a short time, so it is necessary to design a fast and effective approximate solution algorithm. This embodiment provides an improved genetic algorithm embedded in a neighborhood search strategy. The algorithm solution process is as follows: Figure 3 As shown, that is, step S4 of the method proposed by the present invention:

[0165] S4, genetic algorithm embedded in neighborhood search strategy based on “S” type path strategy design;

[0166] As a preferred implementation of step S4, S4 specifically includes:

[0167] S41, population initialization; set the evolutionary generation counter t = 0, the maximum evolutionary generation T, randomly generate M chromosome individuals as the initial population P(0); the chromosome encoding method is real number encoding, and |O| orders are encoded from 1 to |O|, for example, chromosome x = [1, 4, 6, 3, 8, 9, 10, 5, 2, 7], where each gene represents an order and the gene number is the corresponding order number;

[0168] S42, individual evaluation; calculate the fitness value f(x) of each individual in the population P(t) to determine the quality of the individuals in the population; and adopt an "S" type path strategy for the batch path after chromosome decoding;

[0169] S43, determine whether the algorithm termination condition is met; if not, execute step S64; if it is met, output the current optimal chromosome and decode it into a batch allocation plan; when the current evolutionary generation t reaches the preset maximum evolutionary number T, the termination condition is considered to be met;

[0170] S44, selection operation; the selection operator is applied to the group, and based on the fitness evaluation of the individuals in the group, the parent chromosome is obtained by the tournament selection method; this operation directly inherits the optimized individuals to the next generation or generates new individuals through pairing and crossover and then inherits them to the next generation;

[0171] S45, crossover operation: applying a crossover operator to the population. In this embodiment, the crossover operator adopts an order crossover method (OX), randomly selecting a gene segment from the parent chromosome A, copying the gene in the region of the parent chromosome A to the same position of the offspring, and then filling the missing genes in the offspring 1 in order on the parent chromosome B, and obtaining another offspring in a similar manner;

[0172] For example, a chromosome segment [4,5,6,7] is randomly selected from the parent chromosome A = [1,2,3,4,5,6,7,8,9,10], and the parent chromosome segment is copied to the same chromosome position of the offspring. Then, the genes missing from the offspring on the parent chromosome B = [9,5,8,4,2,1,3,7,6,10] are filled in order to obtain the offspring [9,8,2,4,5,6,7,1,3,10]. This operator can retain the better gene segments of the parent while retaining the diversity of the population, thus enhancing the algorithm effect.

[0173] S46, mutation operation: applying the mutation operator to the population, changing the gene values ​​at certain gene positions of individual chromosomes in the population, and obtaining mutated offspring chromosomes; in this embodiment, the crossover operator adopts the inversion mutation method (IM); randomly selecting a gene segment on the original chromosome, and reversing the gene order of the sequence;

[0174] For example, if we randomly select a chromosome segment [9,10,8] from the parent chromosome C = [2,1,4,6,7,9,10,8,5,3], then we can flip the gene sequence of this segment to get the offspring [2,1,4,6,7,8,10,9,5,3];

[0175] S47, neighborhood search; perform neighborhood search operations on all individuals x in the current population through different neighborhood search operators; initialize the current individual x as the optimal individual, if the fitness f(x′) of individual x′ in the current neighborhood structure is better than f(x), then obtain a new optimal individual x←x′, otherwise the current optimal individual enters the next neighborhood structure for search until all neighborhood structures are traversed; the population P(t)^ is obtained after selection operation, crossover operation, mutation operation, and neighborhood search;

[0176] S48, perform environmental selection operation to form a new population; merge the parent and child populations and retain the better individuals; that is, merge the population P(t) with the offspring population P(t)^ to form a merged population with a population size of M*2, calculate the fitness of the individuals in the merged population, arrange them in descending order, select the first M chromosome individuals with smaller fitness values ​​to form a new population P(t)*, and enter the next iteration, and the number of iterations is increased by 1; repeat steps S43-S48 until the genetic algorithm iteration is completed.

[0177] In addition, more specifically, four neighborhood search operators are used in this embodiment, including:

[0178] a) Order swap operator: randomly select two gene loci from the current chromosome, compare the delivery dates of the orders represented by the genes at the gene loci, and if the delivery date of the order represented by the gene at the front gene position is later than the delivery date of the order represented by the gene at the back gene position, swap the gene positions of the two, so that the orders with more urgent delivery dates can be picked in the earlier batches in the picking sequence, reducing the probability of delay. Otherwise, no operation is performed. For example, select genes 2 and 7 in chromosome [2,3,5,7,9,1,6,10,8,4], compare the delivery dates of the orders represented by the two, and if the delivery date of the order represented by gene 7 is earlier, the chromosome is changed to [7,3,5,2,9,1,6,10,8,4].

[0179] b) Order insert operator: randomly select two batches from the decoding scheme of the current chromosome, extract the gene loci corresponding to the two batches in the chromosome and the minimum order weight in their batches respectively, if the minimum order weight in the batch with the later gene loci plus the batch weight of the batch with the earlier gene loci can meet the batch load constraint, then delete the minimum weight order from the batch with the later gene loci, and insert the minimum weight order into the batch with the earlier gene loci to form a new batch. The optimization direction of this operation is to insert small weight orders into the batch with the earlier gene loci where the batch capacity still has surplus, improve its batch loading rate, and make the order picking completion time as early as possible to meet its delivery date. For example, the chromosome [1,2,3,4,5,6,7,8,9,10] is decoded in three batches: [1,2,3], [4,5,6], and [7,8,9,10]. Two batches [1,2,3] and [4,5,6] are randomly selected. If the genes representing the minimum order weights of each are 1 and 6 respectively, and if the batch [1,2,3,6] satisfies the batch load constraint, the chromosome is changed to [1,2,3,6,4,5,7,8,9,10].

[0180] c) Batch sort operator: In the decoding scheme of the current chromosome, traverse all its batches, obtain the orders and corresponding order delivery dates contained in each batch, and calculate the average delivery date of its batches based on the above parameters. Arrange the batches arranged on the current chromosome in ascending order according to the delivery date to obtain a new chromosome. The optimization idea of ​​this operation is to enable orders with more urgent delivery dates in batches to obtain a better picking batch position, so that their picking end time is as early as possible to meet their order delivery dates, so that the chromosome obtains a better fitness value and obtains a better order picking scheduling plan after decoding. For example, the chromosome is [2,4,6,1,5,10,8,3,7,9], the decoding batching scheme is [2,4,6], [1,5,10], [8,3,7,9], and the order of the average delivery time of the batches is [1,5,10], [8,3,7,9], [2,4,6], then the chromosome changes to [1,5,10,8,3,7,9,2,4,6].

[0181] d) Batch swap operator: randomly select two batches in the current chromosome decoding scheme, and obtain the gene positions represented by the batches in the chromosome. Define the urgency of the batch as the difference between the completion time and the average delivery time of the batch. If the urgency of the batch with the gene position in the front is less than the urgency of the batch with the gene position in the back, then swap the gene positions of the chromosome fragments represented by the two batches in the chromosome. The optimization idea of ​​this operation is to enable the chromosome fragments with higher batch urgency to effectively change their chromosome gene positions, so that they can be processed by the pickers in their earlier picking order during decoding, thereby reducing their delay time, and then obtaining a smaller batch delay cost, and obtaining a better picking scheduling plan. For example, the chromosome is [10,5,3,4,7,2,9,1,8,6], and the decoding batching scheme is [10,5], [3,4,7], [2,9,1], [8,6]. Two batches [10,5] and [2,9,1] are randomly selected. If the urgency of the batch [10,5] is less than that of the batch [2,9,1], the gene positions of the two on the chromosome are swapped, and the original chromosome is changed to [2,9,1,3,4,7,10,5,8,6].

[0182] S5. Solve the enterprise material picking simulation instance and obtain the material picking scheduling plan.

[0183] As a preferred implementation of step S5, step S5 is specifically:

[0184] S51. Obtain basic information on the enterprise's warehouse layout, shelf specifications, storage methods, and staffing;

[0185] S52, obtaining material demand order information for each workstation of the production line within a specified time period of the workshop;

[0186] S53, applying the genetic algorithm described in step S4 to the joint planning model established in step S3 to solve the case;

[0187] S54. Obtain a warehouse material picking scheduling solution solved by the example.

[0188] Based on the above technical solution, the optimization simulation example in the scenario of "order batching-path planning" two-stage scheduling planning for warehouse material picking in this embodiment is as follows:

[0189] First, the specific parameter settings of the warehouse are given, as shown in Table 1 below:

[0190] Table 1 Specific parameters of the warehouse

[0191]

[0192] The order picking simulation parameter settings are given, as shown in Table 2 below:

[0193] Table 2 Order picking simulation parameters

[0194]

[0195] This simulation example selects the material preparation order data required for the production line workstations in the manufacturing workshop during the time period of 2:00-2:30 pm for simulation experiments. There are two production lines and 20 orders in total. The weight of each order contains 2-8 materials, and the goods in each order follow the "centralized storage" strategy, each with different storage shelves; the weight of each material in production line category 1 is 0.2-0.5kg, and the weight of each material in production line category 2 is 0.1-0.2kg. The weight of the material demand order for each production line workstation ranges from 5kg to 11.2kg, and corresponds to the workstation one by one. The specific information of the simulation order is shown in Table 3 below:

[0196] Table 3 Simulation order details

[0197]

[0198] In this embodiment, step S4 is to set the simulation parameters of the improved genetic algorithm embedded in the neighborhood search strategy as shown in Table 4 below:

[0199] Table 4 Simulation parameters of genetic algorithm embedded in neighborhood search strategy

[0200]

[0201] The improved genetic algorithm with embedded neighborhood search strategy proposed in the present invention is used to bring in the case simulation order data, and the picking results are shown in Table 5 below; the picking results show that under the condition that there are three pickers (no difference between the pickers), each picker needs to undertake the picking tasks of three batches, and each picking batch is picked according to the "S" type path strategy according to the specific picking batch sequence (such as Figure 4 As shown in the figure, black represents the shelf to be picked, and the arrow is an "S"-shaped picking path; this embodiment gives the picking path of the picking order of picker No. 0, and the access order of other paths is the same), so that all picking tasks can be completed in a shorter time, so that the completion time of each batch of material order tasks does not exceed its specified delivery period, and on-time delivery is achieved, thereby minimizing the picking delay cost to 0 (the sum of the delay costs of different production line orders multiplied by the order picking delay time), effectively solving the problems of no scientific picking plan and serious order picking timeout in the material picking stage of the manufacturing plant, ensuring the normal production of the production line. The supply of storage materials provides a scientific and effective solution for optimizing the efficiency of factory production logistics and reducing costs and increasing efficiency.

[0202] Table 5 Picking results

[0203]

[0204] In summary, the present invention provides a two-stage scheduling method for intelligent warehousing material picking based on a genetic algorithm embedded with a neighborhood search strategy, which fully considers the heterogeneous characteristics of production lines and order delivery periods. Compared with the technology that only considers order batching or picking path planning in existing research, it better meets the diversified manufacturing material picking needs of modern manufacturing enterprises, is more practical and has higher comprehensive scheduling efficiency; the genetic algorithm embedded with a neighborhood search strategy realizes automatic batching and path planning of material orders, minimizing the total delay cost of material picking; at the same time, it is more flexible, accurate and efficient.

[0205] In this specification, the terms "one embodiment", "some embodiments", "examples", "specific examples" or "some examples" refer to at least one embodiment or example described in combination with specific features, structures, materials or characteristics. These described specific features, structures, materials or characteristics can be combined in an appropriate manner in one or more embodiments or examples. In addition, the technician can combine and combine different embodiments or examples and their features described in this specification without contradiction.

[0206] The logic and / or steps shown in the flowchart or described in other ways can be regarded as a sequence of executable instructions for implementing logical functions. These instructions can be implemented in any computer-readable medium for use by instruction execution systems, devices or equipment. These systems, devices or equipment include processor systems or other systems capable of receiving and executing instructions.

[0207] The above embodiments have described the principles and implementations of the present invention in detail, and have used specific examples to illustrate the working principles thereof. These examples are only used to help understand the method of the present invention and its core concept. Meanwhile, according to the concept of the present invention, the actual implementation methods and application scope may vary. Therefore, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A two-stage scheduling method for intelligent warehousing material picking based on genetic algorithm embedded with neighborhood search strategy, characterized in that: The specific steps include: S1. Obtain material demand order information from each workstation of the production line in the manufacturing workshop; S2. Establishing the objective function of the corresponding order picking task according to the acquired material demand order information; S3. Considering the heterogeneity of production lines, that is, the different types of production lines have significantly different material types and quantity requirements, a two-stage joint planning model of material picking "order batching-path planning" is established, and constraints are set; When establishing the two-stage joint planning model of "order batching-route planning" for material picking, the following settings are made in advance: 1) Each picker has the same speed and maximum load, regardless of the picker path congestion; 2) Not considering the service time for each order; 3) Materials belonging to the same order will not be split into different batches; 4) A batch can only be assigned to one picker to complete the picking, and the picking process of each batch is not allowed to be interrupted; 5) Ignoring the volume waste caused by the gaps between goods when they are placed, that is, assuming that the goods are placed closely together; 6) The warehouse shelf storage strategy is centralized storage, that is, materials with the same material number are stored in the same storage location; 7) Do not consider unexpected situations such as out-of-stock shelves or inserted orders; The constraints set include: Constraint 1: Order assignment principle, which limits each order to one and only one batch, is expressed as: Constraint 2: Batch continuity constraint, which stipulates that for any batch, if the batch is selected, the batch with the previous number is also selected. The formula is: Constraint 3: Batch load constraint, which limits the total weight of each order in the same batch to not exceed the maximum load of the batch. The formula is: Constraints 1-3 are used to solve the order batching problem; in the above three formulas, B represents the set of all feasible batches, B = {1, ..., b, ... |B|}, \{1} means that it does not include the value of 1, O represents the set of all orders, O = {1, ..., o, ... |O|}; q o represents the weight of the picking order, Q represents the maximum load of the picker batch; u ob is a binary variable, u ob =1 means order o is assigned to batch b for combined picking, otherwise it is 0; Constraint 4: Batch allocation constraint, which limits each batch to be processed at most once in a certain picking sequence of a certain picker; if the batch is empty, it should not be allocated to any picking sequence of any picker; at the same time, it limits any picker to process at most one batch in any picking sequence; it is expressed by the following three formulas in sequence: Constraint 5: Batch uniqueness constraint, which stipulates that for any production line order, the order is processed only in a certain picking sequence of a certain picker. The formula is: Constraint 6: The picker’s picking order continuity constraint stipulates that for any picking order of any picker, if the current picking order is assigned a batch, the previous order of the order is also assigned a batch. The formula is expressed as: Constraints 4-6 are used to solve the batch assignment problem; in the above five equations, M is the set of all pickers, M = {1,…,m,…|M|}, N is the set of all picking orders of a single picker, N = {1,…,n,…|N|}; y bmn is a binary variable, y bmn =1 indicates that batch b is processed in the picking sequence n of picker m, otherwise it is 0; Constraint 7: Shelf point access rule, which stipulates that for any order, all shelf points involved in the order must be visited in the batch path; if the batch visits a shelf, there must be at least one order in the batch that contains the required materials on the shelf; represented by the following two formulas: Constraint 8: Batch path node association constraint, which stipulates that for any batch, if the batch visits a shelf point, then the shelf point must exist as a batch path node. The formula is expressed as: Constraint 9: Batch path node access constraint, which limits any node in each batch path to be visited at most once. The formula is: Constraint 10: Path elimination sub-loop rule, eliminate the sub-loop situation in path planning, the formula is expressed as: Constraints 7-10 are used to solve the batch routing problem. In the above five equations, V represents the set of all shelf points, V = {0, 1, ..., s ..., |V|}, V = {0} represents the sorting station, and \{0} means that it does not contain the value of 0; z bs is a binary variable, z bs =1 means batch b visits shelf s\{0}, otherwise it is 0; h os is a known parameter, h os =1 means that order o contains items to be picked on shelf s, otherwise it is 0; x bij is a binary variable, x bij =1 means that batch b visits path node j immediately after visiting path node i, otherwise it is 0; |S| is the number of path nodes in the sub-circuit involved in the batch; Constraint 11: The order constraint of batch picking. For any picker, the completion time of the first batch picked by the picker is later than or equal to the sum of the picking initialization time and the time consumed in the picking process; and for any batch of any picker, the completion time of the batch picking is later than or equal to the sum of the completion time of the previous batch picking and the time consumed in the batch picking. The formula is expressed as: Constraint 12: Batch picking completion time constraint, which stipulates that for any batch, if the batch is assigned to a certain picking sequence of a picker, the batch completion time shall not be less than the picking completion time of the picker in the picking sequence. The formula is expressed as: Constraint 13: Production line order picking completion time constraint, which stipulates that for any order, if the order belongs to a certain batch, the order picking completion time shall not be less than the batch picking completion time. The formula is: Constraint 14: Calculate the delay time of each material demand order. The formula is: Constraints 11-14 are used to solve the problem of picking time continuity and delay time calculation; in the above five equations, d ij represents the distance between path node i and path node j, β o Indicates the delivery date of the material corresponding to order o, A is a constant, w mn represents the batch picking completion time of picker m on his picking batch sequence n, c b represents the picking completion time of batch b, e o represents the picking completion time of production line order o, t o Indicates the delay time of order o; S4, designing a genetic algorithm embedded with a neighborhood search strategy based on the "S" type path strategy; Step S4 specifically includes the following sub-steps: S41, population initialization; set the evolutionary generation counter t=0, the maximum evolutionary generation T, and randomly generate M chromosome individuals as the initial population P(0); the chromosome encoding method is real number encoding, the orders are numbered in sequence, each gene in the chromosome represents an order, and the gene number is the corresponding order number; S42, individual evaluation; calculate the fitness value f(x) of each individual in the population P(t) to determine the quality of the individuals in the population; and adopt an "S" type path strategy for the batch path after chromosome decoding; S43, determine whether the algorithm termination condition is met; if not, execute step S44; if satisfied, output the current optimal chromosome and decode it into a batch allocation plan; when the current evolutionary generation t reaches the preset maximum evolutionary number T, the termination condition is considered to be met; S44, selection operation; applying the selection operator to the group, and based on the fitness evaluation of the individuals in the group, adopting the tournament selection method to obtain the parent chromosome; S45, crossover operation: applying the crossover operator to the population, using a sequential crossover method, retaining the better gene segments of the parent generation, and obtaining chromosomes of the offspring generation; S46, mutation operation: applying the mutation operator to the population, changing the gene values ​​at certain gene positions of individual chromosomes in the population, and obtaining mutated offspring chromosomes; S47, neighborhood search; through different neighborhood search operators, perform neighborhood search operations on all individuals x in the current population; initialize the current individual x as the optimal individual, if the fitness f(x′) of individual x′ in the current neighborhood structure is better than f(x), then obtain a new optimal individual x←x′, otherwise the current optimal individual enters the next neighborhood structure for search until all neighborhood structures are traversed; the population P(t)^ is obtained after selection operation, crossover operation, mutation operation, and neighborhood search; S48, perform environmental selection operation to form a new population; merge the parent and child populations and retain the better individuals; that is, merge the population P(t)^ with the offspring population P(t)^ to form a merged population with a population size of M*2, calculate the fitness of the individuals in the merged population, arrange them in descending order, select the first M chromosome individuals with smaller fitness values ​​to form a new population P(t)*, and enter the next iteration, and the number of iterations is increased by 1; repeat steps S43-S48 until the genetic algorithm iteration is completed; S5. Solve the enterprise material picking simulation instance and obtain the material picking scheduling plan.

2. According to claim 1, a two-stage scheduling method for intelligent warehouse material picking based on genetic algorithm embedded with neighborhood search strategy is characterized in that: Step S1 is specifically as follows: The material demand order information of each workstation of the production line in the enterprise manufacturing workshop is collected within a specified time period, and the material demand order information includes: material delivery period information, production line category information to which the workstation belongs, material category and its demand quantity information.

3. According to claim 1, a two-stage scheduling method for intelligent warehouse material picking based on genetic algorithm embedded with neighborhood search strategy is characterized in that: Step S2 is specifically as follows: The material orders of different production line stations are set to be picked as a batch during warehouse material picking. At the same time, the heterogeneity of the production lines is taken into account. The total delay cost of material order picking is minimized as the optimization goal, and the objective function of the corresponding order picking task is established. Its formula is expressed as follows: Among them, z is the objective function of the corresponding order picking task; p k represents the unit time delay cost of the k-th production line order, t o represents the delay time of order o, g ok represents a 0-1 variable. If order o belongs to the k-th production line, g ok =1, otherwise 0; K is the set of all production line categories, and O is the set of all orders.

4. According to claim 1, a two-stage scheduling method for intelligent warehouse material picking based on a genetic algorithm embedded with a neighborhood search strategy is characterized in that: Step S5 specifically includes: S51. Obtain basic information on the enterprise's warehouse layout, shelf specifications, storage methods, and staffing; S52, obtaining material demand order information for each workstation of the production line within a specified time period of the workshop; S53, applying the genetic algorithm described in step S4 to the joint planning model established in step S3 to solve the case; S54. Obtain a warehouse material picking scheduling solution solved by the example.

Images