基于多智能体的备件库存网络系统仿真研究  

Simulation Research on Spare Parts Inventory Network System Based on Multi-Agent

作　　者：张伟钢 张彧雄 ZHANG Wei-gang;ZHANG Yu-xiong(Civil Aviation University of China,Tianjin 300000,China)

机构地区：[1]中国民航大学,天津300000

出　　处：《航空计算技术》2025年第2期12-17,22,共7页Aeronautical Computing Technique

基　　金：中央高校基本科研业务项目资助(3122022084)。

摘　　要：备件库存优化作为库存系统中的重要内容,当前研究大多基于数学规划,通过设立目标与约束函数进行求解,但此方法在面对多节点、多约束情况时存在一定求解复杂度。针对以上问题,以库存系统中多点库存网络为对象,结合智能体建模方法,搭建由多仓库构成的备件库存网络多智能体仿真平台,设计基于库存系统的需求保障算法与基于经验决策的库存优化算法。通过仿真结果分析,发现建立的模型收敛并达到了较好的库存优化效果,且在相同保障目标下,系统收敛结果不受初始配置影响,从而认为能够作为解决相关库存优化问题的方法。最后,在备件放置数量分析中发现一定自组织特性。为相关问题研究提供参考,并作出进一步研究展望。Spare parts inventory optimization,as a critical aspect of inventory systems,has been predominantly studied using mathematical programming,which involves setting objective and constraint functions for solutions.However,this approach faces certain complexities when dealing with multi-node and multi-constraint scenarios.To address these issues,this study focuses on a multi-node inventory network within the inventory system,integrating agent-based modeling methods to construct a multi-agent simulation platform composed of multiple warehouses.The platform incorporates a demand guarantee algorithm based on the inventory system and an inventory optimization algorithm based on empirical decision-making.Through the analysis of simulation results,it is found that the established model converges and achieves effective inventory optimization.Moreover,under the same guarantee objectives,the system′s convergence results are not influenced by initial configurations,suggesting its potential as a method for solving related inventory optimization problems.Finally,a certain degree of self-organizing characteristics is observed in the analysis of spare parts placement quantities.This study provides a reference for related research and offers prospects for further investigation.

关 键 词：备件库存网络 多智能体建模 横向转运 经验决策 

分 类 号：V268[航空宇航科学与技术—航空宇航制造工程]