基于改进聚类和神经网络的多准则备件分类

Multi-criteria Spare Parts Classification Based on Improved Clustering and Neural Networks

传统的备件分类方法存在主观性强、补偿性差的问题,且在分类完成后缺少持续更新的能力。首先,本文考虑备件的经济、供给和重要性三个方面建立了备件分类指标体系,并结合否定规则改进聚类算法,解决多指标的互补性问题。其次,提出了一种改进粒子群优化BP神经网络的备件分类模型,对新入库备件分类有持续更新能力。最后,结合S公司的备件数据,利用改进的算法对备件进行分类和更新。结果表明,所提出的改进聚类算法更符合实际需求,且所提出的BP神经网络平均准确率为96.77%,优于其他传统的算法,可为企业的备件分类管理提供有力的支持。

The traditional method of spare parts classification suffers from issues of subjectivity and poor compensatory ability,and lacks the capability for continuous updating after the classification is completed.In this paper,a scheme was proposed for spare parts classification.Firstly,a spare parts classification index system was established considering three aspects of spare parts:economy,supply,and importance.To address the compensatory problem of multiple indicators,the clustering algorithm was improved by incorporating the negation rule.Secondly,an improved particle swarm optimization BP neural network was proposed as a spare parts classification model.This model allowed for continuous updating when new spare parts were introduced.Finally,the improved algorithms were applied to classify and update the spare parts using the spare parts data from S Company.The results demonstrate that the proposed improved clustering algorithm better meets practical needs.Additionally,the proposed BP neural network achieves an average accuracy of 96.77%,surpassing traditional algorithms,and provides robust support for the classification and management of spare parts in enterprises.