1. An article ex-warehouse scheduling method comprises the following steps:

constructing a multi-target article delivery optimization model; wherein the multi-objective item delivery optimization model comprises: the targets are three target functions and constraint functions with the minimum total number of pieces to be disassembled, the minimum total distribution time and the minimum total distribution cost respectively;

acquiring stock minimum unit SKU and SKU demand quantity corresponding to the items in the item order data;

solving a pareto solution set corresponding to the item order data by using a multi-objective genetic algorithm and based on the multi-objective item delivery optimization model; wherein each pareto solution of the set of pareto solutions comprises: shipment warehouse information and a shipment quantity corresponding to the SKU and meeting the quantity required by the SKU;

and selecting an optimal pareto solution from the pareto solution set based on a preset selection criterion, and determining an article delivery scheduling scheme according to the optimal pareto solution.

2. The method of claim 1, the solving a set of pareto solutions corresponding to the item order data using a multi-objective genetic algorithm and based on the multi-objective item delivery optimization model comprising:

obtaining a scheduling solution corresponding to the item order data;

generating three populations respectively corresponding to the three objective functions based on the scheduling solution;

solving the pareto solution sets corresponding to the three populations using a multi-objective genetic algorithm and based on the multi-objective item delivery optimization model.

3. The method of claim 2, the solving the set of pareto solutions corresponding to the three populations comprising:

updating the three populations by using a multi-target genetic algorithm and based on the multi-target article delivery optimization model to obtain a pareto solution;

screening according to the crowding distance of the pareto solutions, and placing the pareto solutions after screening as elite individuals into an elite population;

updating the elite population by using a multi-target genetic algorithm and based on the multi-target article delivery optimization model, and screening based on the crowding distance of the elite individual;

and if a termination condition is met, outputting the pareto solution set.

4. The method of claim 2, further comprising:

performing neighborhood search on the original scheduling solutions in the three populations to generate a first new scheduling solution corresponding to the original scheduling solutions in the three populations;

wherein the objective function value corresponding to the first new scheduling solution is less than the objective function value corresponding to the original scheduling solution.

5. The method of claim 2, further comprising:

performing distributed estimation on the original scheduling solutions in the three populations to generate a second new scheduling solution corresponding to the original scheduling solutions in the three populations;

and the quantity of the delivery warehouses corresponding to the second new scheduling solution is less than that of the delivery warehouses corresponding to the original scheduling solution.

6. The method of claim 2, the obtaining a scheduling solution corresponding to the item order data comprising:

acquiring warehouse inventory information and cost information of a warehouse corresponding to each item order; wherein the cost information includes: time cost information and distribution cost information;

generating the scheduling solution according to the warehouse inventory information and the cost information;

wherein the scheduling solution comprises: shipment warehouse information corresponding to each item order and a corresponding shipment volume.

7. The method of claim 6, the generating the scheduling solution from the warehouse inventory information and the cost information comprising:

determining shipment warehouse information and corresponding shipment quantity of each SKU of each item order according to the warehouse inventory information and the cost information so as to generate the scheduling solution; and/or the presence of a gas in the gas,

and determining shipment warehouse combination information corresponding to all the article orders and corresponding shipment quantity according to the warehouse inventory information and the cost information so as to generate the scheduling solution.

8. The method of claim 7, the generating three populations corresponding to the three objective functions, respectively, based on the scheduling solution comprising:

acquiring a maximum target value and a minimum target value of the target function;

calculating a target value of the scheduling solution corresponding to the objective function, and calculating a first difference between the target value and the maximum target value, and a second difference between the maximum target value and the minimum target value;

determining a ratio of the first difference to the second difference as a population distance between the scheduling solution and a population corresponding to the objective function;

assigning the scheduling solution to one of the three populations based on the population distance.

9. The method of claim 8, the assigning the scheduling solution to one of the three populations based on the population distance comprising:

acquiring three population distances between the scheduling solution and the three populations respectively, and determining the shortest population distance in the three population distances;

and allocating the scheduling solution to the population corresponding to the shortest population distance.

10. The method of claim 1, wherein said selecting an optimal pareto solution from the set of pareto solutions based on a preset selection criterion comprises:

calculating three target values corresponding to the three target functions respectively for each pareto solution in the set of pareto solutions;

and calculating the product of the three target values, and taking the pareto solution corresponding to the minimum product as the optimal pareto solution.

11. The method of any one of claims 1 to 10,

the multi-target genetic algorithm comprises the following steps: NSGA-II algorithm.

12. An article out-of-warehouse scheduling device, comprising:

the model setting module is used for constructing a multi-target article delivery optimization model; wherein the multi-objective item delivery optimization model comprises: the targets are three target functions and constraint functions with the minimum total number of pieces to be disassembled, the minimum total distribution time and the minimum total distribution cost respectively;

the order acquisition module is used for acquiring stock minimum unit SKU and the SKU demand quantity corresponding to the items in the item order data;

the scheduling solving module is used for solving a pareto solution set corresponding to the item order data by using a multi-objective genetic algorithm and based on the multi-objective item delivery optimization model; wherein each pareto solution of the set of pareto solutions comprises: shipment warehouse information and a shipment quantity corresponding to the SKU and meeting the quantity required by the SKU;

and the scheduling determination module is used for selecting an optimal pareto solution from the pareto solution set based on a preset selection criterion, and determining an article delivery scheduling scheme according to the optimal pareto solution.

13. An article out-of-warehouse scheduling device, comprising:

a memory; and a processor coupled to the memory, the processor configured to perform the method of any of claims 1-11 based on instructions stored in the memory.

14. A computer-readable storage medium having stored thereon computer instructions for execution by a processor to perform the method of any one of claims 1 to 11.

Background

With the continuous development of logistics and the higher requirement of customers on timeliness, many merchants choose to build logistics or use third-party logistics warehouse services. Due to the ever-increasing size of SKUs, warehouse area, and cost effectiveness, each SKU is typically stored in multiple warehouses. At present, a warehouse delivery strategy for orders is generally adopted, which brings great difficulty to supply chain cooperative stock preparation, is difficult to fully utilize logistics network resources, and has the defects of high order removal rate, high logistics cost, high operation cost and the like.

Disclosure of Invention

In view of the above, an object of the present invention is to provide a method, an apparatus and a storage medium for dispatching an article out of a warehouse.

According to one aspect of the present disclosure, there is provided an article ex-warehouse scheduling method, including: constructing a multi-target article delivery optimization model; wherein the multi-objective item delivery optimization model comprises: the targets are three target functions and constraint functions with the minimum total number of pieces to be disassembled, the minimum total distribution time and the minimum total distribution cost respectively; acquiring stock minimum unit SKU and SKU demand quantity corresponding to the items in the item order data; solving a pareto solution set corresponding to the item order data by using a multi-objective genetic algorithm and based on the multi-objective item delivery optimization model; wherein each pareto solution of the set of pareto solutions comprises: shipment warehouse information and a shipment quantity corresponding to the SKU and meeting the quantity required by the SKU; and selecting an optimal pareto solution from the pareto solution set based on a preset selection criterion, and determining an article delivery scheduling scheme according to the optimal pareto solution.

Optionally, the solving a pareto solution set corresponding to the item order data using a multi-objective genetic algorithm and based on the multi-objective item delivery optimization model includes: obtaining a scheduling solution corresponding to the item order data; generating three populations respectively corresponding to the three objective functions based on the scheduling solution; solving the pareto solution sets corresponding to the three populations using a multi-objective genetic algorithm and based on the multi-objective item delivery optimization model.

Optionally, the solving the pareto solution sets corresponding to the three populations comprises: updating the three populations by using a multi-target genetic algorithm and based on the multi-target article delivery optimization model to obtain a pareto solution; screening according to the crowding distance of the pareto solutions, and placing the pareto solutions after screening as elite individuals into an elite population; updating the elite population by using a multi-target genetic algorithm and based on the multi-target article delivery optimization model, and screening based on the crowding distance of the elite individual; and if a termination condition is met, outputting the pareto solution set.

Optionally, performing neighborhood search on the original scheduling solutions in the three populations to generate a first new scheduling solution corresponding to the original scheduling solutions in the three populations; wherein the objective function value corresponding to the first new scheduling solution is less than the objective function value corresponding to the original scheduling solution.

Optionally, performing distributed estimation on the original scheduling solutions in the three populations, and generating a second new scheduling solution corresponding to the original scheduling solutions in the three populations; and the quantity of the delivery warehouses corresponding to the second new scheduling solution is less than that of the delivery warehouses corresponding to the original scheduling solution.

Optionally, the obtaining a scheduling solution corresponding to the item order data includes: acquiring warehouse inventory information and cost information of a warehouse corresponding to each item order; wherein the cost information includes: time cost information and distribution cost information; generating the scheduling solution according to the warehouse inventory information and the cost information; wherein the scheduling solution comprises: shipment warehouse information corresponding to each item order and a corresponding shipment volume.

Optionally, the generating the scheduling solution according to the warehouse inventory information and the cost information includes: determining shipment warehouse information and corresponding shipment quantity of each SKU of each item order according to the warehouse inventory information and the cost information so as to generate the scheduling solution; and/or determining shipment warehouse combination information corresponding to all the article orders and corresponding shipment quantity according to the warehouse inventory information and the cost information so as to generate the scheduling solution.

Optionally, the generating three populations respectively corresponding to the three objective functions based on the scheduling solution includes: acquiring a maximum target value and a minimum target value of the target function; calculating a target value of the scheduling solution corresponding to the objective function, and calculating a first difference between the target value and the maximum target value, and a second difference between the maximum target value and the minimum target value; determining a ratio of the first difference to the second difference as a population distance between the scheduling solution and a population corresponding to the objective function; assigning the scheduling solution to one of the three populations based on the population distance.

Optionally, the assigning the scheduling solution to one of the three populations based on the population distance includes: acquiring three population distances between the scheduling solution and the three populations respectively, and determining the shortest population distance in the three population distances; and allocating the scheduling solution to the population corresponding to the shortest population distance.

Optionally, the selecting an optimal pareto solution from the set of pareto solutions based on a preset selection criterion includes: calculating three target values corresponding to the three target functions respectively for each pareto solution in the set of pareto solutions; and calculating the product of the three target values, and taking the pareto solution corresponding to the minimum product as the optimal pareto solution.

Optionally, the multi-objective genetic algorithm comprises: NSGA-II algorithm.

According to another aspect of the present disclosure, there is provided an article delivery scheduling apparatus, including: the model setting module is used for constructing a multi-target article delivery optimization model; wherein the multi-objective item delivery optimization model comprises: the targets are three target functions and constraint functions with the minimum total number of pieces to be disassembled, the minimum total distribution time and the minimum total distribution cost respectively; the order acquisition module is used for acquiring stock minimum unit SKU and the SKU demand quantity corresponding to the items in the item order data; the scheduling solving module is used for solving a pareto solution set corresponding to the item order data by using a multi-objective genetic algorithm and based on the multi-objective item delivery optimization model; wherein each pareto solution of the set of pareto solutions comprises: shipment warehouse information and a shipment quantity corresponding to the SKU and meeting the quantity required by the SKU; and the scheduling determination module is used for selecting an optimal pareto solution from the pareto solution set based on a preset selection criterion, and determining an article delivery scheduling scheme according to the optimal pareto solution.

According to still another aspect of the present disclosure, there is provided an article out-of-warehouse scheduling apparatus, including: a memory; and a processor coupled to the memory, the processor configured to perform the method as described above based on instructions stored in the memory.

According to yet another aspect of the present disclosure, a computer-readable storage medium is provided, which stores computer instructions for execution by a processor to perform the method as described above.

The article ex-warehouse scheduling method, the article ex-warehouse scheduling device and the storage medium provide a solution for the problem of order sourcing based on total odd number removal, total timeliness and total cost minimization in a multi-order scene, multi-objective optimization of total odd number removal, total delivery time minimization and total delivery cost minimization can be achieved in the delivery scheduling, optimal utilization of warehousing resources can be achieved, working efficiency is improved, and operation cost is reduced.

Drawings

In order to more clearly illustrate the embodiments of the present disclosure or the technical solutions in the prior art, the drawings needed to be used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present disclosure, and other drawings can be obtained by those skilled in the art without inventive exercise.

FIG. 1 is a schematic flow chart diagram illustrating one embodiment of an item ex-warehouse scheduling method according to the present disclosure;

FIG. 2 is a schematic flow chart diagram for solving a pareto solution set in one embodiment of an item ex-warehouse scheduling method according to the present disclosure;

FIG. 3 is a schematic flow chart diagram illustrating a procedure for obtaining a pareto solution set in an embodiment of an item ex-warehouse scheduling method according to the present disclosure;

FIG. 4 is a schematic flow chart diagram illustrating a procedure for obtaining a pareto solution set in another embodiment of an item ex-warehouse scheduling method according to the present disclosure;

FIG. 5 is a block schematic diagram of one embodiment of an item ex-warehouse scheduling apparatus according to the present disclosure;

FIG. 6 is a block diagram illustration of a dispatch resolution module in one embodiment of an item ex-warehouse dispatch device according to the present disclosure;

FIG. 7 is a block diagram illustration of another embodiment of an item ex-warehouse scheduling apparatus according to the present disclosure;

fig. 8 is a block diagram of yet another embodiment of an item ex-warehouse scheduling device according to the present disclosure.

Detailed Description

The present disclosure now will be described more fully hereinafter with reference to the accompanying drawings, in which exemplary embodiments of the disclosure are shown. The technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the drawings in the embodiments of the present disclosure, and it is obvious that the described embodiments are only a part of the embodiments of the present disclosure, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments disclosed herein without making any creative effort, shall fall within the protection scope of the present disclosure.

The terms "first", "second", and the like are used hereinafter only for descriptive distinction and not for other specific meanings.

Fig. 1 is a schematic flow chart of an embodiment of an article warehouse-out scheduling method according to the present disclosure, as shown in fig. 1:

and 101, constructing a multi-target goods delivery optimization model.

The multi-objective item delivery optimization model may be of various types. For example, the multi-objective goods delivery optimization model comprises three objective functions and a constraint function, wherein the three objective functions respectively aim to minimize the total quantity of disassembled goods, minimize the total delivery time and minimize the total delivery cost.

Step 102, receiving the item order data, and acquiring stock minimum unit SKU (stock locating unit) and SKU demand quantity corresponding to the items in the item order data.

103, solving a pareto solution set corresponding to the item order data by using a multi-objective genetic algorithm and based on a multi-objective item delivery optimization model, wherein each pareto solution in the pareto solution set comprises: shipment warehouse information and a shipment volume corresponding to the SKU and satisfying the quantity required by the SKU.

Pareto solutions are Pareto solutions, also known as non-dominated solutions or non-dominated solutions. When there are multiple targets, one solution is best on one target and may be worst on the other target, due to collisions and incomparable phenomena between targets. These solutions that, while improving any objective function, necessarily weaken at least one other objective function are called non-dominant solutions or Pareto solutions.

The multi-target genetic algorithm comprises an NSGA-II algorithm and the like. The core of the genetic algorithm is a genetic three-operator: selection, crossover and mutation. And randomly crossing and randomly mutating the offspring genes through the randomly selected parent, so as to form a new population. And sequencing the population individuals in the population by using a fitness function, and selecting the individuals with good fitness to perform next generation iteration.

NSGA-II (genetic algorithm with non-dominated sorting of elite strategy) is a multi-objective optimization algorithm based on genetic algorithm, and is multi-objective optimization based on pareto optimal solution discussion. The multi-objective optimization problem is different from the single-objective optimization problem, when a plurality of targets exist, because of conflicts among the targets, it is difficult to find a solution to make all objective functions optimal at the same time, therefore, for the multi-objective optimization problem, a solution set usually exists, the solutions cannot be relatively good and bad with respect to all objective functions, and cannot improve any objective function while not weakening at least one other objective function, and the solution is a Pareto solution (Pareto optimal solutions).

The NSGA-II algorithm reduces the complexity of calculation on one hand, and on the other hand, combines the parent population with the offspring population, so that the next generation population is selected from double space, and all the most excellent individuals are reserved; an elite strategy is introduced to ensure that certain excellent population individuals cannot be discarded in the evolution process, so that the precision of an optimization result is improved; and a crowding degree and crowding degree comparison operator is adopted as a comparison standard among individuals in the population, so that the individuals in the quasi-Pareto domain can be uniformly expanded to the whole Pareto domain, and the diversity of the population is ensured.

And 104, selecting an optimal pareto solution from the pareto solution set based on a preset selection criterion, and determining an item ex-warehouse scheduling scheme corresponding to the item order data according to the optimal pareto solution. A plurality of pareto solutions are obtained based on a multi-objective genetic algorithm strategy, the selection criteria can be various, and an article ex-warehouse scheduling scheme is flexibly selected according to the pareto solution set.

In one embodiment, the location, storage, age, cost information, etc. of the warehouse are known, the SKU and the number of pieces of the order are also known, and the optimization goals for constructing the multi-objective item delivery optimization model are to minimize total quantity of pieces removed, minimize total delivery time, and minimize total delivery cost. The parameters used in the multi-objective goods delivery optimization model are shown in Table 1 below:

TABLE 1 parameter Table used

Setting a decision variable:

x_i,j,kindicating the amount of delivery of sku j of order k to warehouse i.

Setting auxiliary variables:

T_k: the time required for order k to be fulfilled.

The multi-objective goods delivery optimization model (mixed integer linear programming model) is constructed as follows:

x_i，j，kis a non-negative integer variable, y_i，k，z_kIs a 0-1 variable (7)

Wherein, the formula (1) is three objective functions, which are total number of demolished orders (total number of demolished orders), total delivery time (total delivery timeliness) and total delivery cost (total cost); equation (2-7) is a constraint function, where equation (2) means that the number of units required per SKU for all orders is not greater than the number allocated; equation (3) means that for each SKU in any warehouse, the sum of the number of SKUs allocated to all orders does not exceed the inventory level of that warehouse.

Equation (4) means that if there are more than 1 warehouse to satisfy a certain order, the order is a disassembled order; (5) the definition refers to that if one warehouse at least meets one commodity of a certain order, the value of y is 1, namely the variable of y; (6) indicating that the age of an order depends on all maximum times that the order is fulfilled; the formula (7) is a variable value range.

Fig. 2 is a schematic flow chart of solving the pareto solution set in an embodiment of the item ex-warehouse scheduling method according to the present disclosure, as shown in fig. 2:

step 201, obtaining a scheduling solution corresponding to the item order data.

Step 202, three populations respectively corresponding to the three objective functions are generated based on the scheduling solution.

And step 203, solving a pareto solution set corresponding to the three populations by using a multi-objective genetic algorithm and based on a multi-objective article delivery optimization model.

There may be a variety of ways to obtain a scheduling solution corresponding to the item order data. And acquiring warehouse inventory information and cost information of the warehouse corresponding to each item order, wherein the cost information comprises time cost information, distribution cost information and the like. Generating a scheduling solution according to the warehouse inventory information and the cost information, wherein the scheduling solution comprises: shipment warehouse information corresponding to each item order and a corresponding shipment volume.

In one embodiment, there may be two ways to generate a scheduling solution based on warehouse inventory information and cost information: acquiring shipment warehouse information and a corresponding shipment amount of each SKU of each item order by using a roulette algorithm according to warehouse inventory information and cost information to generate a scheduling solution; and secondly, acquiring shipment warehouse combination information corresponding to all the article orders and corresponding shipment quantities by using a roulette algorithm according to the warehouse inventory information and the cost information so as to generate a scheduling solution.

The item order data includes the required pieces count for all SKUs for each item order; the warehouse inventory information includes the number of pieces of each SKU stored, the time cost information and the delivery cost information of the warehouse corresponding to each item order are the delivery time and the delivery cost of the warehouse corresponding to the address of each item order. Based on the item order data and the warehouse inventory, warehouses that do not store SKUs for the item order may be deleted to reduce the amount of computation. If the number of warehouses is too large, the warehouse that is under control may be deleted, i.e., if one warehouse has a SKU inventory, a time cost, and a distribution cost that are all inferior to another warehouse.

An item order data is shown in table 2 below:

order numbering	SKU numbering	Number of SKUs	Delivery address
				D01	SKU01	3	Shandong university of Shandong province Jinan City
D02	SKU02	3	Qingdao university of Qingdao city, Shandong province
				D03	SKU03	2	Shandong University of Science and Technology, Qingdao, Shandong Province
D03	SKU01	1	Shandong University of Science and Technology, Qingdao, Shandong Province

TABLE 2 item order data sheet

The above three orders, the warehouse is assumed to have two, and the inventory information of the warehouse is shown in the following table 3:

storage houseNumbering	SKU	Number of
			01	SKU1	10
01	SKU2	1
			02	SKU2	10
02	SKU3	10

TABLE 3 inventory information Table for warehouse

The delivery cost and timeliness information for the warehouse is shown in table 4 below:

warehouse numbering	Order numbering	Time	Cost of
				01	D01	2	100
01	D02	10	200
				01	D03	9	150
02	D01	8	160
				02	D02	2	110
02	D03	1	40

TABLE 4 delivery cost and timeliness information Table

In one embodiment, a scheduling solution includes the shipment warehouse information for each order's SKU and the corresponding shipment volume. The shipment warehouse information for each order's SKU and the corresponding shipment volume (dispatch plan) may be represented encoded using two-dimensional dynamic arrays P1 and P2, P1 and P2 for each order having the same structure, P1 representing the shipment warehouse list for each SKU, and P2 representing the shipment pieces list for each SKU.

For example, the scheduling schemes for order a and order B are:

of order AThe first column is SKU1 information indicating that the number of ex-warehouse items in warehouse 1 is 2 and the number of ex-warehouse items in warehouse 2 is 3, and the second column is SKU2 information indicating that the number of ex-warehouse items in warehouse 2 is 1. For order B, there is only one SKU1, indicating warehouse 1 ex-warehouse count of 1.

For the scheduling scheme for order A and order B, three target values may be calculated relative to three objective functions. For example, since the order a is satisfied by two warehouses, the target value of the objective function corresponding to the total split number (total split number) of the scheduling solution is 1, and the target value with respect to the other two objective functions can be obtained according to the delivery time cost and the delivery cost information.

Generating an initial population based on a scheduling solution, and in order to obtain a solution of target value diversity, obtaining the scheduling solution by adopting the following two methods: method one, a warehouse available list for each SKU is obtained, and for each SKU in each order, the warehouse that satisfies it is obtained according to a roulette algorithm. The roulette algorithm is an existing roulette algorithm that implements that the probability of a warehouse being selected is proportional to its fitness function value. Two methods may be used to determine the probability of a warehouse being selected when using the roulette algorithm: a, according to the distribution cost, using the reciprocal of the distribution cost value and carrying out normalization to obtain the probability of each warehouse being selected; and b, according to the time required by delivery, using the reciprocal of the delivery time value and carrying out normalization to obtain the probability of each warehouse being selected.

For example, if method one is needed to get 10 scheduling solutions (initial solutions), then 5 scheduling solutions use method a to determine the probability of the warehouse being selected, and another 5 scheduling solutions use b to determine the probability of the warehouse being selected. One scheduling solution includes shipment warehouse information corresponding to all of the item orders and the corresponding shipment volume. The scheduling scheme of each order is a group of two-dimensional arrays, and the two-dimensional arrays of all orders form a scheduling solution.

And secondly, acquiring a SKU list of each warehouse capable of meeting the order, acquiring all warehouse combinations which have the minimum number of the warehouses and can meet the order, and acquiring shipment warehouse combination information corresponding to all the article orders and the corresponding shipment quantity by using a roulette algorithm. Two methods may be used to determine the probability of a warehouse being selected when using the roulette algorithm: calculating the distribution cost under each warehouse combination, and using the reciprocal of the distribution cost value and carrying out normalization to obtain the probability of each warehouse being selected; and b, calculating the time required for distribution under each warehouse combination, and using the reciprocal of the distribution time value and carrying out normalization to obtain the probability of each warehouse being selected.

For the roulette algorithm to determine the probability of the bin being selected, for example, order A is { SKU1:2, SKU2:1} and order B is { SKU1:2}, there are three bins, bin 1 storing SKU 1100 pieces, bin 2 and bin 3 storing SKU 2100 pieces. For order a, only warehouse 1 can be selected, the distribution cost is 20, and the timeliness (distribution time cost) is 3; for order B, the distribution cost of warehouse 2 and warehouse 3 is 10 and 20 respectively, the time efficiency is 3 and 2 respectively, the minimum warehouse combination is {1,2}, {1,3}, the distribution cost is 30 and 40, the time efficiency is 3 and 3, and under the method a, the two combination selection probabilities are 3/7 and 4/7 respectively, and under the method B, the probabilities are equal.

In one embodiment, the purpose of population partitioning is to dedicate different populations to exploring different directions, e.g., populations corresponding to objective functions for the total number of splits should be made as small as possible. The method comprises the steps of obtaining a maximum target value and a minimum target value of an objective function, calculating a target value corresponding to a scheduling solution and the objective function, calculating a first difference value between the target value and the maximum target value and a second difference value between the maximum target value and the minimum target value, and determining the ratio of the first difference value to the second difference value as a population distance between the scheduling solution and a population corresponding to the objective function. And acquiring three population distances between the scheduling solution and the three populations respectively, determining the shortest population distance in the three population distances, and distributing the scheduling solution to the population corresponding to the shortest population distance.

For example, the minimum target values (m1, m2, m3) and the maximum target values (mx1, mx2, mx3) of each objective function are obtained; for each scheduling solution, target values corresponding to three objective functions are calculated, namely, an objective function tuple (o1, o2, o3) is obtained, the population distance between the scheduling solution and the population corresponding to the objective function is calculated, namely, a ratio (r1, r2, r3) ((o1-m1)/(mx1-m1), (o2-m2)/(mx2-m2), (o3-m3)/(mx3-m3) is obtained), and each scheduling solution is allocated to the population with the minimum ratio, namely, the scheduling solution is allocated to the population corresponding to the minimum ratio. The scheduling solutions within the three populations may be adjusted using a variety of existing methods such that the number of scheduling solutions within each population is substantially equal.

Firstly, randomly generating an initial population with the size of N, and obtaining a first generation offspring population through three basic operations of selection, crossing and variation of a genetic algorithm after non-dominated sorting; secondly, from the second generation, merging the parent population and the offspring population, performing rapid non-dominant sorting, simultaneously performing crowding degree calculation on the individuals in each non-dominant layer, and selecting proper individuals according to the non-dominant relationship and the crowding degree of the individuals to form a new parent population; and finally, generating a new offspring population through the basic operation of the genetic algorithm, and so on until the condition of program end is met.

Fig. 3 is a schematic flow chart of obtaining a pareto solution set in an embodiment of an item ex-warehouse scheduling method according to the present disclosure, as shown in fig. 3:

and 301, updating the three populations by using a multi-objective genetic algorithm and based on a multi-objective goods delivery optimization model to obtain a pareto solution. The NSGA-II algorithm can be used, selection, intersection, variation and other processing are carried out on the basis of three objective functions, constraint functions and the like, and the three populations are updated to obtain pareto solutions.

And aiming at the population corresponding to each objective function, the non-pareto solution is firstly removed according to an updating principle until the upper limit requirement of the required quantity is met. If the upper limit is still exceeded after all non-pareto solutions are rejected, then deletion is performed based on the congestion distance.

For example, the number of scheduling solutions in the population is 3, and if 3 children are obtained after searching, there are six scheduling solutions at this time. If the number of the non-pareto solutions is more than or equal to 3, randomly eliminating 3 of the non-pareto solutions, and reserving 3 of the non-pareto solutions. If the number of the non-pareto solutions still exceeds the required pareto solution upper limit after all the non-pareto solutions are removed, the congestion distance is used for deleting, and the calculation formula is as follows:

wherein f is_iI is 1,2,3 is the objective function, f_i ^-Is the ratio f in other scheduling solutions_iSmall and closest. If fi is the minimum scheduling solution, the congestion distance is a sufficiently large number. And reserving the scheduling solution with large congestion distance. For example, there are four scheduling solutions in the population, with target values of [1,2,3, respectively]，[2,1,3]，[3,2,1]，[2,2,2](ii) a For scheduling solution 1, if f1 is 1 and is not smaller than f1, the congestion distance is infinite; for scheduling solution 2, if f2 is 1 and is not smaller than f2, the congestion distance is infinite; for scheduling solution 3, if f3 is 1 and is not smaller than f3, the congestion distance is infinite; for scheduling solution 4, if f1 is 2, 1 is smaller than f1, and the denominator is the maximum target value 3 — the minimum target value 1 is the congestion distance 1/2.

And 302, performing screening processing according to the crowding distance of the pareto solutions, and putting the pareto solutions after screening into an elite population as elite individuals.

And 303, updating the elite population by using a multi-target genetic algorithm and based on a multi-target article delivery optimization model, and screening based on the crowding distance of the elite individual.

Adding all the pareto solutions after screening into the elite population, deleting the non-pareto solutions of the elite population at the moment, then calculating the crowding distance, and reserving the pareto solutions with large crowding distances.

And step 304, if the termination condition is met, outputting a pareto solution set.

And (3) carrying out selection, crossing, mutation and other processes by using an NSGA-II algorithm based on three objective functions, constraint functions and the like, terminating population iteration after the iteration times reach an upper limit or the maximum time reaches an upper limit, and outputting a pareto solution. The parameters set for the NSGA-II algorithm are shown in Table 5 below:

TABLE 5 setting parameters table of algorithm

And calculating three target values corresponding to each pareto solution in the pareto solution set and three target functions respectively, calculating products of the three target values, and taking the pareto solution corresponding to the minimum product as the optimal pareto solution. For example, there are two pareto solutions (scheduling solutions) in the pareto solution set, and the three target values corresponding to the three objective functions in each pareto solution are (1,2,3), (1,2.5,2.5), respectively, and the product of the three target values is calculated, and the first scheduling solution with the smallest product is taken as the optimal pareto solution.

In one embodiment, a neighborhood search is performed on the original scheduling solutions in the three populations to generate a first new scheduling solution corresponding to the original scheduling solutions in the three populations; wherein the objective function value corresponding to the first new scheduling solution is less than the objective function value corresponding to the original scheduling solution. The neighborhood search may use existing multiple domain search algorithms.

For example, for a population corresponding to an objective function for the total number of splits, for each scheduling solution within that population, the warehouse is reselected for the order of the split to reduce the rate of the split. For example, for a scheduling solution, the SKUs for an order are satisfied by warehouse 1 and warehouse 2, and then a neighborhood search algorithm is used to determine whether there is a warehouse that satisfies the order, i.e., all warehouses are traversed, and it is determined whether only one warehouse is needed to satisfy the order, thereby generating a first new scheduling solution.

For a population corresponding to an objective function of total delivery time, randomly selecting k orders for each scheduling solution in the population, acquiring delivery processing time of a warehouse, judging whether the warehouse with the longest delivery time of each order can be replaced by the warehouse with short delivery time by using a neighborhood search algorithm, and selecting the warehouse (with shorter delivery time and meeting SKU requirements) to replace by using a roulette algorithm.

For example, for an order, the delivery time depends on the warehouse with the longest delivery time, and a neighborhood search algorithm is used to traverse the warehouse to determine if there are other warehouses that replace the warehouse and the delivery time is shorter, thereby generating a first new scheduling solution. Similarly, the same method may be used to generate a first new scheduling solution for the population corresponding to the objective function of the total delivery cost.

In one embodiment, distributed estimation is performed on the original scheduling solutions in the three populations to generate a second new scheduling solution corresponding to the original scheduling solutions in the three populations; and the quantity of the delivery warehouses corresponding to the second new scheduling solution is less than that of the delivery warehouses corresponding to the original scheduling solution. The distributed estimation algorithm is derived from a genetic algorithm, and various existing distributed estimation algorithms can be used.

For each population, a bin selection is calculated for each order, and if an order is satisfied by two bins, each bin index value is incremented 1/2 to obtain an index value for each bin, and the scheduling solution is updated again for that order using the roulette algorithm. For example, a population has three dispatch solutions, and there are 3 satisfaction modes for order 1, which are [1,2], [1], [1,2], respectively, then warehouse 1 has index 1/2+1+1/2, warehouse 2 has index 1/2+1/2, and the probability of using 1 and 2 for the order is 2/3, 1/3. A store is obtained by using a roulette algorithm to determine whether the order can be fulfilled and the next store can be selected if the roulette algorithm cannot be fulfilled.

Fig. 4 is a schematic flow chart of obtaining a pareto solution set in another embodiment of an item ex-warehouse scheduling method according to the present disclosure, as shown in fig. 4:

step 401, an initial population is generated.

Step 402, population division. Three populations corresponding to the three objective functions, respectively, are generated, and the number of times i that no new pareto solution is found is initialized to 0.

Step 403, judging whether the number i of times that no new pareto solution is found is less than L; if yes, go to step 404, if no, go to step 405.

At step 404, a neighborhood search algorithm is performed. And performing neighborhood search on the original scheduling solutions in the three populations by using a neighborhood search algorithm.

Step 405, a distributed estimation algorithm is executed. And performing distributed estimation on the original scheduling solutions in the three populations by using a distributed estimation algorithm.

At step 406, a new set of scheduling solutions is obtained via steps 404 and 405.

And 407,408, updating the population and the elite population. And updating the population and the elite population by using a multi-target genetic algorithm and based on the multi-target article delivery optimization model.

Step 409, judging whether the termination condition is reached, if yes, entering step 411, and if not, advancing step 410.

The number of times no new pareto solutions are found is updated, step 410. And judging whether a new pareto solution is obtained, and if not, i is i + 1.

Step 411, outputting the elite population.

In one embodiment, as shown in fig. 5, the present disclosure provides an article out-of-warehouse scheduling device 50, comprising: a model setup module 51, an order acquisition module 52, a schedule solution module 53, and a schedule determination module 54. The model setting module 51 constructs a multi-objective article delivery optimization model; the multi-target goods delivery optimization model comprises three target functions and a constraint function; the three objective functions are respectively aimed at minimum total quantity of disassembled pieces, minimum total delivery time and minimum total delivery cost.

The order taking module 52 receives the item order data and takes the SKU and the SKU demand quantity corresponding to the item in the item order data. The scheduling solving module 53 solves a pareto solution set corresponding to the item order data using a multi-objective genetic algorithm and based on a multi-objective item delivery optimization model; wherein each pareto solution of the set of pareto solutions comprises: shipment warehouse information and a shipment volume corresponding to the SKU and satisfying the quantity required by the SKU.

The scheduling determination module 54 selects an optimal pareto solution from the pareto solution set based on a preset selection criterion, and determines an item ex-warehouse scheduling scheme corresponding to the item order data according to the optimal pareto solution. For example, the schedule determination module 54 calculates three target values corresponding to three objective functions respectively for each pareto solution in the set of pareto solutions, calculates products of the three target values, and takes the pareto solution corresponding to the smallest product as the optimal pareto solution.

In one embodiment, as shown in FIG. 6, the schedule solution module 53 includes: a scheduling solution obtaining unit 531, a population generating unit 532, and a pareto solution obtaining unit 533. The scheduling solution obtaining unit 531 obtains a scheduling solution corresponding to the item order data. The population generating unit 532 generates three populations corresponding to the three objective functions, respectively, based on the scheduling solution. The pareto solution obtaining unit 533 solves the pareto solution sets corresponding to the three populations using the multi-objective genetic algorithm and based on the multi-objective item delivery optimization model.

The pareto solution obtaining unit 533 updates the three populations to obtain pareto solutions by using a multi-objective genetic algorithm and based on a multi-objective item delivery optimization model; the pareto solution obtaining unit 533 performs screening processing according to the crowding distance of the pareto solution, and places the pareto solution after screening as an elite individual into an elite population; the pareto solution obtaining unit 533 updates the elite population by using a multi-objective genetic algorithm and based on a multi-objective article delivery optimization model, and performs screening processing based on the crowding distance of the elite individual; the pareto solution obtaining unit 533 outputs the pareto solution set if the termination condition is satisfied.

The scheduling solution obtaining unit 531 obtains warehouse inventory information, cost information of the warehouse corresponding to each item order, the cost information including time cost information, delivery cost information, and the like. The scheduling solution obtaining unit 531 generates a scheduling solution according to the warehouse inventory information and the cost information, where the scheduling solution includes: shipment warehouse information corresponding to each item order and a corresponding shipment volume.

The scheduling solution obtaining unit 531 determines shipment warehouse information and a corresponding shipment amount of each SKU of each item order according to the warehouse inventory information and the cost information, so as to generate a scheduling solution; the scheduling solution obtaining unit 531 determines shipment warehouse combination information corresponding to all the article orders and corresponding shipment quantities according to the warehouse inventory information and the cost information, to generate a scheduling solution.

The population generating unit 532 acquires a maximum target value and a minimum target value of the target function, calculates a target value of the scheduling solution corresponding to the target function, and calculates a first difference between the target value and the maximum target value and a second difference between the maximum target value and the minimum target value; the population generating unit 532 determines a ratio of the first difference to the second difference as a population distance between the scheduling solution and a population corresponding to the objective function, and assigns the scheduling solution to one of the three populations based on the population distance. For example, the population generating unit 532 obtains three population distances between the scheduling solutions and the three populations, respectively, determines the shortest population distance of the three population distances, and assigns the scheduling solution to the population corresponding to the shortest population distance.

In one embodiment, as shown in fig. 6, the article warehouse-out scheduling device 50 further includes: a neighborhood search module 55 and a distribution estimation module 55. The neighborhood searching module 55 performs neighborhood searching on the original scheduling solutions in the three populations to generate a first new scheduling solution corresponding to the original scheduling solutions in the three populations, and an objective function value corresponding to the first new scheduling solution is smaller than an objective function value corresponding to the original scheduling solution.

The distribution estimation module 56 performs distributed estimation on the original scheduling solutions in the three populations to generate a second new scheduling solution corresponding to the original scheduling solutions in the three populations, and the number of shipment warehouses corresponding to the second new scheduling solution is smaller than the number of shipment warehouses corresponding to the original scheduling solutions.

Fig. 8 is a block diagram of another embodiment of an item ex-warehouse scheduling device according to the present disclosure. As shown in fig. 8, the apparatus may include a memory 81, a processor 82, a communication interface 83, and a bus 84. The memory 81 is used for storing instructions, the processor 82 is coupled to the memory 81, and the processor 82 is configured to execute the method for implementing the above-mentioned article ex-warehouse scheduling method based on the instructions stored in the memory 81.

The memory 81 may be a high-speed RAM memory, a non-volatile memory (non-volatile memory), or the like, and the memory 81 may be a memory array. The storage 81 may also be partitioned and the blocks may be combined into virtual volumes according to certain rules. The processor 82 may be a central processing unit CPU, or an application Specific Integrated circuit asic, or one or more Integrated circuits configured to implement the item ex-warehouse scheduling method of the present disclosure.

In one embodiment, the present disclosure provides a computer-readable storage medium storing computer instructions that, when executed by a processor, implement a method as in any one of the above embodiments.

The article ex-warehouse scheduling method, the article ex-warehouse scheduling device and the storage medium provided in the embodiments above construct a multi-objective article delivery optimization model and set three objective functions for total quantity removal, total delivery time and total delivery cost, obtain a plurality of pareto solutions based on a multi-objective genetic algorithm and a multi-objective article delivery optimization model, and select a final scheduling scheme according to a pareto solution set; the method provides a solution for the order sourcing problem based on the minimum total quantity of detached odd numbers, total timeliness and total cost in a multi-order scene, can achieve multi-objective optimization of the minimum total quantity of detached odd numbers, the minimum total delivery time and the minimum total delivery cost in delivery scheduling, can achieve optimal utilization of warehousing resources, improves working efficiency and reduces operation cost.

The method and system of the present disclosure may be implemented in a number of ways. For example, the methods and systems of the present disclosure may be implemented by software, hardware, firmware, or any combination of software, hardware, and firmware. The above-described order for the steps of the method is for illustration only, and the steps of the method of the present disclosure are not limited to the order specifically described above unless specifically stated otherwise. Further, in some embodiments, the present disclosure may also be embodied as programs recorded in a recording medium, the programs including machine-readable instructions for implementing the methods according to the present disclosure. Thus, the present disclosure also covers a recording medium storing a program for executing the method according to the present disclosure.

The description of the present disclosure has been presented for purposes of illustration and description, and is not intended to be exhaustive or limited to the disclosure in the form disclosed. Many modifications and variations will be apparent to practitioners skilled in this art. The embodiment was chosen and described in order to best explain the principles of the disclosure and the practical application, and to enable others of ordinary skill in the art to understand the disclosure for various embodiments with various modifications as are suited to the particular use contemplated.