CN-121998556-A - Multi-commodity order fulfilling method, device and equipment based on dynamic computing power distribution

CN121998556ACN 121998556 ACN121998556 ACN 121998556ACN-121998556-A

Abstract

The invention provides a method, a device and equipment for fulfilling a multi-commodity order based on dynamic computing power distribution, and relates to the technical field of supply chain management. The method comprises the steps of obtaining commodity order and inventory information, establishing a multi-objective optimization algorithm model based on actual requirements of multi-commodity order fulfillment in the order and inventory information, generating an initial population by adopting a heuristic method based on priority and inventory feasibility, performing iterative optimization on the initial population by adopting a multi-objective optimization algorithm based on dynamic computing power distribution to obtain a Pareto non-dominant solution set, and selecting a final order fulfillment scheme according to the Pareto non-dominant solution set. The method can effectively solve the problems of low optimization efficiency and easiness in sinking into local optimization of multi-commodity order fulfillment in a high-dimensional space, balance a plurality of core operation indexes under complex constraint conditions, and remarkably improve the scientificity, global search efficiency and distribution fairness of decisions.

Inventors

ZHANG HAN
ZHANG XIAOPING
WANG CHENG
Pan Chongdi
CHEN ZHIWEI
WU SHUIGEN
LIN HUANGXING
LIU CHENG
Gou jin

Assignees

华侨大学
智立通(厦门)科技有限公司

Dates

Publication Date: 20260508
Application Date: 20260410

Claims (10)

1. A multi-commodity order fulfillment method based on dynamic computing power distribution, comprising: S1, acquiring commodity orders and inventory information, wherein the commodity orders and the inventory information comprise a current order set, a commodity set and a warehouse inventory set; S2, based on actual requirements of multi-commodity order fulfillment in order and inventory information, establishing a multi-objective optimization algorithm model comprising profit margin, order satisfaction rate, inventory utilization rate and order priority violation of a fulfillment scheme; s3, generating an initial population by adopting a heuristic method based on priority and stock feasibility; S4, carrying out iterative optimization on the initial population by a multi-objective optimization algorithm based on dynamic computing power distribution to obtain a Pareto non-dominant solution set, wherein the multi-objective optimization algorithm based on dynamic computing power distribution adopts a dynamic computing power distribution strategy, and distributes computing resources to all the sub-problems through decision space sub-region division and decision density evaluation; and S5, selecting a final order fulfillment scheme according to the Pareto non-dominant solution set.
2. The method for fulfilling a multiple commodity order according to claim 1, wherein said profit margin is the ratio of the actual total profit margin allocated to fulfill the order to the theoretical maximum total profit margin, expressed as: ; Wherein, the Is profit margin; For orders Medium commodity A variable of whether the demand of (2) is fully satisfied; For orders Medium commodity Is a profit of (2); Collecting current orders; is a commodity collection; indicating that the maximum value is taken; the order satisfaction rate is the ratio of the number of fully satisfied orders to the total order number, and the expression is: ; Wherein, the The order meeting rate is the order meeting rate; The value of (2) represents the total commodity quantity of all orders; Is a minimum value; the stock utilization is the ratio of the number of goods allocated to the order to the total available stock, and the expression is: ; Wherein, the Inventory utilization; For orders Medium commodity Is a required amount of (a); Is a warehouse Medium commodity Is a stock quantity of (a); Inventory sets for a warehouse; the order priority violation degree is calculated stepwise based on the commodity priority number, the identification of whether the commodity meets the priority, the commodity arrival time sequence number and the first commodity sequence number which is not allocated to be met, and is used for measuring a punishment value generated when the order is not fulfilled, and the expression is as follows: ; ; Wherein, the Order priority violation degree; is the theoretical maximum penalty value; Is a theoretical minimum penalty value; Is that Punishment value of group commodity; representing to take the minimum value; Is that Priority number of group commodity; Representation of Commodity in group commodity Whether or not to meet priority ; Representation of Order in group commodity Medium commodity An arrival time number; Representation of The first product serial number which is not allocated and satisfied in the group products.
3. The method for fulfilling multiple commodity orders based on dynamic computing force distribution according to claim 1, wherein said multiple objective optimization algorithm model is further provided with technical constraints, said technical constraints comprising: inventory allocation constraints, i.e., specifying that the total amount of a commodity allocated to an order must not exceed its required amount, are expressed as: ; Wherein, the To be allocated to an order Certain commodity Is a total amount of (2); For orders Medium commodity Is a required amount of (a); Collecting current orders; is a commodity collection; Inventory sets for a warehouse; Warehouse inventory capability constraints, i.e., specifying that the total amount of items dispensed from a warehouse must not exceed its existing inventory, are expressed as: ; ; Wherein, the Is a warehouse Medium commodity Is a stock quantity of (a); Indicating individual, the value is 1 or 0, if A value of 1, representing an order Commodity in (a) From a warehouse If it meets A value of 0, representing an order Commodity in (a) Not from warehouse Satisfying the following requirements; and (3) constraint of warehouse processing capacity, namely setting an upper limit of the number of orders which can be processed by each warehouse for each commodity by limiting the accumulated value of the decision variable in the order dimension, wherein the expression is as follows: ; Wherein, the An upper limit on the number of orders that can be processed for each commodity for each warehouse.
4. A multi-item order fulfillment method based on dynamic computing power distribution according to claim 3, wherein the specific step of generating an initial population is: Firstly, setting up a priority rule, namely, sorting the priority of commodity demands according to the arrival time of orders, wherein the priority of the orders arriving earlier is higher, and sequentially trying to match the stock from each warehouse from the demand with high priority in each group of commodities until the stock of the commodities is used or the commodity demands in all orders are met; then the three-dimensional binary matrix is adopted for the individual Coding, i.e. each individual corresponds to a three-dimensional binary matrix For representing the distribution relationship of orders, commodities and warehouses, The dimension of (c) is the product of the number of orders, the number of categories of goods and the number of warehouses, ; Followed by heuristic rules following the feasibility principle Generating an initial population of individuals, wherein the expression is: ; Wherein, the Collecting current orders; is a commodity collection; Inventory sets for a warehouse; Priority rules are used for determining the sequence of commodity distribution; Inventory feasibility for the warehouse, which is used for restricting the distribution quantity not to exceed the existing inventory of the warehouse; Is a disturbance factor; the initial population is obtained based on the initial population of individuals.
5. The method for fulfilling the multi-commodity order according to claim 1, wherein the iterative optimization of the initial population by the multi-objective optimization algorithm based on dynamic computing power distribution specifically comprises the following steps: First, the multi-objective problem is decomposed into Each sub-question Correspondingly associating a weight vector which is uniformly distributed, wherein the weight vector is generated through uniform sampling; For each sub-problem Selecting the first N weight vectors closest to the Euclidean distance of the weight vectors to form a neighbor set ; For each sub-problem Generating Initial decisions, each corresponding to a three-dimensional binary matrix, the total number of decisions is ; All initial decisions form a decision space, and the decision space is divided into K mutually exclusive subareas; Expanding each decision in the decision space into a binary sequence according to rows, and converting the binary sequence into a corresponding decimal integer value, wherein the expression is as follows: ; Wherein, the Flattening the decision and converting the decision into decimal integer values; To flatten the decision into a string of binary digits Values at the individual positions; Collecting current orders; is a commodity collection; Inventory sets for a warehouse; Performing modular operation on the decimal integer values to map each decision to a unique subarea in K mutually exclusive subareas, and completing division of the subareas of the decision space so as to simplify the three-dimensional binary matrix with high dimension into subarea index numbers with low dimension, wherein the expression is: ; Wherein, the For making decisions Index number of the sub-region; dividing the total number of subareas for the decision space; for the remainder operation; Then go through Counting decision numbers of all sub-regions, calculating decision density of all sub-regions, and introducing shannon entropy indexes as measurement values for evaluating distribution uniformity of solution sets in a decision space to establish a decision density evaluation model, wherein the expression is as follows: ; ; ; Wherein, the Is the first Decision number of sub-regions; Is a decision; Is a decision set; Is the first Decision density of the sub-region; For decision distribution uniformity, i.e. shannon entropy, when When in use, the distribution is completely uniform; then, for each sub-problem, counting the maximum density value in the region set where the decision is located in the population as the decision density weight, wherein the expression is as follows: ; ; Wherein, the Is a sub-problem A collection of sub-regions covered by a population; is a sub-problem A corresponding population; Is the first A decision space subregion; is a sub-problem Is a decision density weight of (1); Indicating the presence; normalizing the decision density weight to obtain the selection probability of the sub-problem, wherein the expression is: ; Wherein, the Is a sub-problem Is selected according to the selection probability of (1); is a sub-problem Decision density weights of (2) Based on the selection probability, calculating the dynamic calculation force distribution times of each sub-problem, wherein the expression is as follows: ; Wherein, the Is a sub-problem Dynamic calculation force distribution times of (a); the total calculation power of each generation, namely the total times of information exchange between population individuals and neighbors; and selecting a parent individual from a neighbor set of each sub-problem to carry out evolution operation according to the dynamic calculation power distribution times, and generating a offspring decision, namely a Pareto non-dominant solution set.
6. The method for fulfilling multiple commodity orders based on dynamic computing power distribution according to claim 5, wherein when each child problem performs evolution operation, two parent individuals are selected from the neighbor set to perform intersection and mutation operation until the uniformity of decision distribution tends to be uniform or a preset maximum evolution algebra is reached.
7. The method of claim 5, further comprising comparing the offspring decisions with decisions in the current population after each sub-problem completes the current evolutionary operation to generate offspring decisions, and if the offspring decisions are better, replacing corresponding decisions in the original population to update the population.
8. The method of claim 2, wherein when selecting a final order fulfillment plan based on the Pareto non-dominant solution set, a composite objective function value is calculated for each of the solutions expressed as: ; Wherein, the Is a comprehensive objective function value; 、、 Is a weight coefficient; is a penalty coefficient; Is profit margin; The order meeting rate is the order meeting rate; Inventory utilization; Order priority violation degree; and selecting an optimal solution as a final order fulfilling scheme according to the comprehensive objective function value, and outputting a corresponding order, commodity and warehouse allocation scheme.
9. A multi-commodity order fulfilling device based on dynamic computing force distribution for implementing a multi-commodity order fulfilling method based on dynamic computing force distribution as set forth in any one of claims 1-8, comprising: The information acquisition module is used for acquiring commodity orders and inventory information, and comprises a current order set, a commodity set and a warehouse inventory set; the modeling module is used for establishing a multi-objective optimization algorithm model comprising profit margin, order satisfaction rate, inventory utilization rate and order priority violation of a fulfillment scheme based on actual requirements of multi-commodity order fulfillment in the order and inventory information; The population generation module is used for generating an initial population by adopting a heuristic method based on priority and stock feasibility; the iterative module is used for carrying out iterative optimization on the initial population by a multi-objective optimization algorithm based on dynamic computing power distribution to obtain a Pareto non-dominant solution set; and the decision module is used for selecting a final order fulfillment scheme according to the Pareto non-dominant solution set.
10. A multi-item order fulfillment device based on dynamic computing power distribution, comprising a processor and a memory, the memory having stored therein a computer program executable by the processor to implement a multi-item order fulfillment method based on dynamic computing power distribution as claimed in any one of claims 1 to 8.

Description

Multi-commodity order fulfilling method, device and equipment based on dynamic computing power distribution Technical Field The invention relates to the technical field of supply chain management, in particular to a method, a device and equipment for fulfilling multi-commodity orders based on dynamic computing power distribution. Background With the rapid development of the supply chain industry, enterprises are required to process massive orders containing multiple commodity demands in daily operations. Multi-item order fulfillment involves making allocation decisions within a three-dimensional decision space of orders, items, and warehouses, i.e., determining whether each item in an order is offered by a particular warehouse and a particular offered quantity. Since the available inventory is typically scattered across multiple different warehouses, how to achieve optimal allocation of resources has become a core topic in the field of supply chain management. Existing multi-commodity order fulfillment scheme formulation methods have mostly focused on a single objective optimization or rule-of-experience-dependent simplified framework, such as heuristic allocation with cost minimization or satisfaction rate maximization alone as the sole guide, or through a fixed fulfillment order. These methods often fail to adequately account for the fine-grained coupling relationships of multiple commodities between multiple warehouses and the competing interactions between different targets. When facing complex scenes of multiple commodity types, wide stock distribution and coexistence of conflicting targets, the prior art is difficult to balance among core operation indexes such as profit, order satisfaction rate, stock utilization rate, order priority and the like. Meanwhile, due to the extremely high solution space dimension of the problem, the problem that the traditional algorithm is low in efficiency or falls into local optimum easily occurs in the optimization process, and high-quality global solutions are difficult to search in a limited time. In addition, actual business operations also require that fairness of allocation be maintained under rules such as first come first served, which typically requires the introduction of commodity priorities based on commit time and corresponding penalty mechanisms. However, the conventional formulation method has obvious insufficient overall capacity for such fairness constraint and multi-objective coordination, and it is difficult to stably give a scheme for considering global benefit and individual feasibility under real-time changing order and inventory data. With the expansion of service scale and the enhancement of constraint conditions, the defects of reduced solving efficiency, feasible solution shrinkage, scheme bias and the like are more likely to occur in the prior art, and the operation requirements of modern supply chains on high profit, high order satisfaction rate, high inventory utilization and high order priority adherence degree cannot be continuously met. In view of this, the present application has been made. Disclosure of Invention The invention aims to provide a multi-commodity order fulfilling method, device and equipment based on dynamic computing power distribution, which are used for solving the defects that in the prior art, when a three-dimensional space decision consisting of an order, a commodity and a warehouse is processed by a multi-commodity order fulfilling scheme, core indexes such as profit, order satisfaction rate, inventory utilization, order priority and the like are difficult to balance, and the defects of low optimization efficiency, easiness in trapping in local optimization, difficulty in considering distribution fairness and the like exist in a high-dimensional decision space. In order to solve the technical problems, the invention is realized by the following technical scheme: a multi-commodity order fulfillment method based on dynamic computing power distribution, comprising: S1, acquiring commodity orders and inventory information, wherein the commodity orders and the inventory information comprise a current order set, a commodity set and a warehouse inventory set; S2, based on actual requirements of multi-commodity order fulfillment in order and inventory information, establishing a multi-objective optimization algorithm model comprising profit margin, order satisfaction rate, inventory utilization rate and order priority violation of a fulfillment scheme; s3, generating an initial population by adopting a heuristic method based on priority and stock feasibility; S4, carrying out iterative optimization on the initial population by a multi-objective optimization algorithm based on dynamic computing power distribution to obtain a Pareto non-dominant solution set, wherein the multi-objective optimization algorithm based on dynamic computing power distribution adopts a dynamic computing power distribution strategy, and distributes