CN-121998219-A - Logistics vehicle path planning method, system, terminal and storage medium of supply chain system

CN121998219ACN 121998219 ACN121998219 ACN 121998219ACN-121998219-A

Abstract

The invention relates to the technical field of combination optimization and discloses a logistics vehicle path planning method, a logistics vehicle path planning system, a logistics vehicle path planning terminal and a logistics vehicle path planning storage medium of a supply chain system, wherein the logistics vehicle path planning method comprises the steps of obtaining a vehicle dispatching task of a logistics end, analyzing and processing to obtain a task analysis result, and extracting information to obtain dispatching information; and carrying out symmetrical self-distillation processing on the mixed expert strategy to obtain a target path planning strategy, and carrying out path planning on the scheduling information to obtain a target planning path. According to the invention, through a mixed expert collaborative decision mechanism based on dynamic weight distribution and the introduction of a two-stage knowledge fusion strategy, the vehicle path planning efficiency is improved, and the optimal vehicle planning path can be obtained.

Inventors

PAN YINGHUI
LIU HUI
MING ZHONG

Assignees

深圳大学

Dates

Publication Date: 20260508
Application Date: 20260408

Claims (10)

1. A logistics vehicle path planning method of a supply chain system, characterized in that the logistics vehicle path planning method of the supply chain system comprises the following steps: Acquiring a vehicle dispatching task of a logistics terminal, analyzing and processing the vehicle dispatching task to obtain a task analysis result, and extracting information from the task analysis result to obtain dispatching information; Constructing a planning path expert collaborative framework, performing strategy selection on the scheduling information according to the planning path expert collaborative framework to obtain a plurality of path planning strategies, and performing mixed processing on all the path planning strategies to obtain a mixed expert strategy; And carrying out symmetrical self-distillation processing on the mixed expert strategy to obtain a target path planning strategy, and carrying out path planning on the scheduling information according to the target path planning strategy to obtain a target planning path.
2. The logistics vehicular path planning method of a supply chain system of claim 1, wherein said constructing a planned path expert collaborative framework comprises: Building a standard expert module according to a deterministic greedy strategy, building a disturbance expert module according to a random graph structure disturbance attention mechanism, and building a regret expert module according to a historical state backtracking mechanism; And setting modes of the standard expert module, the disturbance expert module and the remorse expert module to obtain a parallel processing mode, and constructing frames of the standard expert module, the disturbance expert module and the remorse expert module according to the parallel processing mode to obtain a planned path expert collaborative frame.
3. The supply chain system logistics vehicular path planning method of claim 2, wherein the path planning strategy comprises a first strategy, a second strategy, and a third strategy; The strategy selection is carried out on the scheduling information according to the planning path expert collaborative framework to obtain a plurality of path planning strategies, and all the path planning strategies are mixed to obtain a mixed expert strategy, which comprises the following steps: Encoding the scheduling information to obtain a high-dimensional embedded representation, wherein the scheduling information comprises coordinate information, road network distance matrix information and cargo demand information; Carrying out path planning on the high-dimensional embedded representation according to a standard expert module to obtain a first path joint probability distribution, and obtaining a first strategy according to the first path joint probability distribution; Setting a corresponding disturbance mode according to the disturbance expert module, carrying out path planning on the high-dimensional embedded representation according to the disturbance mode to obtain a second path joint probability distribution, and obtaining a second strategy according to the second path joint probability distribution, wherein the disturbance mode comprises a random discarding mode, a random adding mode and a mixed mode; Setting a cancellation operation according to the regret expert module, planning a path for the high-dimensional embedded representation according to the cancellation operation to obtain a third path joint probability distribution, and obtaining a third strategy according to the third path joint probability distribution; and carrying out mixed processing on the first strategy, the second strategy and the third strategy to obtain a mixed expert strategy.
4. A logistics vehicular path planning method of a supply chain system of claim 3, wherein said path planning said high-dimensional embedded representation according to said perturbation pattern results in a second path joint probability distribution, comprising in particular: performing compatibility score calculation on the high-dimensional embedded representation to obtain an initial compatibility score, and performing disturbance calculation on the initial compatibility score according to the disturbance mode to obtain a target compatibility score matrix; performing weight aggregation on the target compatibility score matrix to obtain node context characterization, and performing probability distribution calculation on the node context characterization to obtain second path joint probability distribution; the compatibility score calculation is performed on the high-dimensional embedded representation, specifically: ; And performing disturbance calculation on the initial compatibility score according to the disturbance mode, wherein the disturbance calculation comprises the following specific steps: ; the probability distribution calculation is carried out on the node context representation, specifically: ; Wherein, the For the initial compatibility score to be the same, In order to query the vector of the vector, For the transposition of the key vector, For the dimension of each key vector, For the target compatibility score matrix, For the multiplication on an element-by-element basis, In the form of a random mask matrix, For the purpose of node context characterization, The probability distribution is combined for the second path.
5. A logistics vehicular path planning method of a supply chain system in accordance with claim 3, wherein said path planning of said high-dimensional embedded representation in accordance with said undoing operation is in particular: ; Wherein, the For the third path joint probability distribution, In order to be a complete path, For the final regret recording, In order for the scheduling information to be available, For the total length of the path sequence, In order to be a time step, the time step, In-state for policy network Down selection action Is a function of the probability of (1), For the original predicted probability of the policy network pair being remorsed, For remorse path Is used for the prefix path of (a).
6. The logistics vehicle path planning method of a supply chain system of claim 1, wherein the performing symmetric self-distillation processing on the hybrid expert strategy to obtain a target path planning strategy specifically comprises: Symmetrically transforming the mixed expert strategy to obtain a symmetrical transformation set, wherein the symmetrical transformation comprises complete reverse order transformation, random replacement transformation and cyclic displacement transformation; And carrying out weighted aggregation on the mixed expert strategy to obtain a plurality of strategy dynamic weights, and carrying out strategy prediction according to the symmetrical transformation set and all strategy dynamic weights to obtain a target path planning strategy.
7. The logistics vehicular path planning method of a supply chain system of claim 6, wherein said symmetrically transforming said hybrid expert strategy comprises: ; The step of carrying out weighted aggregation on the mixed expert strategy is specifically as follows: ; Wherein, the In order to transform the set symmetrically, In order to be a symmetric transformation operator, Is the first The path that the individual expert generates, A set of paths generated for all the parallel experts, In order to solve the quality evaluation function, Is the first The action sequences corresponding to the paths of the individual experts, Is the first The dynamic weights assigned by the individual experts are, Is the first The recent average cost of the individual specialists, As a function of the temperature parameter(s), Is the first Average cost of the individual experts in the near term.
8. A logistics vehicular path planning system of a supply chain system, wherein the logistics vehicular path planning system of a supply chain system comprises: The task analysis module is used for acquiring a vehicle dispatching task at the logistics end, analyzing and processing the vehicle dispatching task to obtain a task analysis result, and extracting information from the task analysis result to obtain dispatching information; The expert processing module is used for constructing a planning path expert collaborative framework, carrying out strategy selection on the scheduling information according to the planning path expert collaborative framework to obtain a plurality of path planning strategies, and carrying out mixed processing on all the path planning strategies to obtain a mixed expert strategy; And the strategy distillation module is used for carrying out symmetrical self-distillation processing on the mixed expert strategy to obtain a target path planning strategy, and carrying out path planning on the scheduling information according to the target path planning strategy to obtain a target planning path.
9. A terminal comprising a memory, a processor and a program stored on the memory and executable on the processor, which when executed by the processor, implements the steps of the logistics vehicle path planning method of the supply chain system in accordance with any one of claims 1-7.
10. A computer readable storage medium, characterized in that it has stored thereon a computer program, which computer readable storage medium has stored thereon a logistics vehicular path planning program of a supply chain system, which logistics vehicular path planning program of a supply chain system, when executed by a processor, implements the steps of the logistics vehicular path planning method of a supply chain system as claimed in any one of claims 1-7.

Description

Logistics vehicle path planning method, system, terminal and storage medium of supply chain system Technical Field The invention relates to the technical field of combination optimization, in particular to a logistics vehicle path planning method, a logistics vehicle path planning system, a logistics vehicle path planning terminal and a logistics vehicle path planning computer readable storage medium of a supply chain system. Background In a complex supply chain system, the core decision link of logistics vehicle path planning of the supply chain system essentially belongs to a combination optimization problem, and the combination optimization problem aims to search a scheme with the lowest cost or highest efficiency from mass feasible solutions. However, this combinatorial optimization problem increases exponentially with the size of the client nodes. In industrial practice, how to rapidly plan a vehicle path close to global optimum under the constraint of a limited time window and computational resources is a technical problem to be solved. Traditional accurate algorithms and heuristic rules perform well on small-scale problems, but when facing large-scale, high-dimensional real industrial scenes, often take too long to calculate or are difficult to jump out of local optimum. While conventional approaches require reliance on problem-specific heuristics, there are many limitations. Secondly, the dynamic adaptability of symmetry learning and multi-strategy collaborative optimization is insufficient, the exploration depth in a complex scene is influenced, the efficiency of vehicle path planning is low, and an optimal vehicle planning path cannot be obtained. Accordingly, the prior art is still in need of improvement and development. Disclosure of Invention The invention mainly aims to provide a logistics vehicle path planning method, a logistics vehicle path planning system, a logistics vehicle path planning terminal and a logistics vehicle path planning storage medium, and aims to solve the problems that in the prior art, the vehicle path planning efficiency is low and an optimal vehicle planning path cannot be obtained. In order to achieve the above object, the present invention provides a logistics vehicle path planning method of a supply chain system, the logistics vehicle path planning method of the supply chain system comprising the steps of: Acquiring a vehicle dispatching task of a logistics terminal, analyzing and processing the vehicle dispatching task to obtain a task analysis result, and extracting information from the task analysis result to obtain dispatching information; Constructing a planning path expert collaborative framework, performing strategy selection on the scheduling information according to the planning path expert collaborative framework to obtain a plurality of path planning strategies, and performing mixed processing on all the path planning strategies to obtain a mixed expert strategy; And carrying out symmetrical self-distillation processing on the mixed expert strategy to obtain a target path planning strategy, and carrying out path planning on the scheduling information according to the target path planning strategy to obtain a target planning path. Optionally, the method for planning a logistics vehicle path of a supply chain system, wherein the constructing a planning path expert collaborative framework specifically includes: Building a standard expert module according to a deterministic greedy strategy, building a disturbance expert module according to a random graph structure disturbance attention mechanism, and building a regret expert module according to a historical state backtracking mechanism; And setting modes of the standard expert module, the disturbance expert module and the remorse expert module to obtain a parallel processing mode, and constructing frames of the standard expert module, the disturbance expert module and the remorse expert module according to the parallel processing mode to obtain a planned path expert collaborative frame. Optionally, the logistics vehicle path planning method of the supply chain system, wherein the path planning strategy comprises a first strategy, a second strategy and a third strategy; The strategy selection is carried out on the scheduling information according to the planning path expert collaborative framework to obtain a plurality of path planning strategies, and all the path planning strategies are mixed to obtain a mixed expert strategy, which comprises the following steps: Encoding the scheduling information to obtain a high-dimensional embedded representation, wherein the scheduling information comprises coordinate information, road network distance matrix information and cargo demand information; Carrying out path planning on the high-dimensional embedded representation according to a standard expert module to obtain a first path joint probability distribution, and obtaining a first strategy according to the first pa