现代制造工程 ›› 2026, Vol. 553 ›› Issue (10): 128-135.doi: 10.16731/j.cnki.1671-3133.2026.10.016

• 仪器仪表/检测/监控 • 上一篇    下一篇

基于SSA-GA-BP神经网络的多方法检测视情维修预测优化模型*

张伟钢1, 于雷2, 肖俊贤2   

  1. 1 中国民航大学交通科学与工程学院,天津 300300;
    2 中国民航大学航空工程学院,天津 300300
  • 收稿日期:2025-07-28 出版日期:2026-10-18 发布日期:2026-11-05
  • 作者简介:张伟钢,副教授,博士,主要研究方向为可靠性与系统工程。E-mail:wgzhang@cauc.edu.cn。于雷,硕士研究生,主要研究方向为维修理论与可靠性。E-mail:1833809886@qq.com
  • 基金资助:
    *中央高校基本科研项目(3122022084)

Multi-method detection condition-based maintenance prediction optimization model based on SSA-GA-BP neural network

Zhang Weigang1, Yu Lei2, Xiao Junxian2   

  1. 1 College of Transportation Science and Engineering,Civil Aviation University of China,Tianjin 300300,China;
    2 College of Aeronautical Engineering,Civil Aviation University of China,Tianjin 300300,China
  • Received:2025-07-28 Online:2026-10-18 Published:2026-11-05

摘要: 针对多方法检测视情维修(Multi-Method detection Condition-Based Maintenance,MMCBM)策略的理论解可能不适应实际情况的问题,需要对策略进行调整,这种调整往往难以准确预测调整后策略的总成本与可靠性。为此,通过基于麻雀搜索算法-遗传算法-反向传播(Sparrow Search Algorithm-Genetic Algorithm-Back Propagation, SSA-GA-BP)神经网络的方法,对调整后策略中的总成本与可靠性进行预测,以获得较为精确的解决方案。首先,建立了基于不完美维修多方法检测视情维修模型,并用枚举算法得到某部件基于多方法视情维修的最优策略与多方法检测数据集。其次,构建基于SSA-GA-BP神经网络的预测模型,使用麻雀搜索算法(Sparrow Search Algorithm,SSA)对BP神经网络中的超参数进行优化,使用遗传算法(Genetic Algorithm,GA)对BP神经网络中的隐藏层的权值和偏置进行优化,将寻优结果代入BP神经网络,最终预测出给定条件下的总成本与可靠性。结果表明,优化后的神经网络在预测总成本与可靠度的精度上,相比未优化的神经网络有了不同程度的提升。

关键词: 多方法检测, 视情维修, BP神经网络, 不完美维修, 麻雀搜索算法, 遗传算法

Abstract: The theoretical solution of Multi-Method detection Condition-Based Maintenance (MMCBM) strategy may not adapt to the actual situation,so it is necessary to adjust the strategy. This adjustment is often difficult to accurately predict the total cost and reliability of the adjusted strategy. To this end,the total cost and reliability in the adjusted strategy are predicted by the method based on Sparrow Search Algorithm-Genetic Algorithm-Back Propagation (SSA-GA-BP) neural network to obtain a more accurate solution. Firstly,the MMCBM model based on imperfect maintenance was established,and the enumeration algorithm was used to obtain the optimal MMCBM strategy and multi-method detection data set of a certain component. Secondly,a prediction model based on SSA-GA-BP neural network was constructed,Sparrow Search Algorithm (SSA) was used to optimize the hyperparameters in BP neural network,Genetic Algorithm (GA) was used to optimize the weights and bias of the hidden layer in BP neural network,and the optimization results were substituted into BP neural network. Finally,the total cost and reliability under the given conditions are predicted. The results show that the optimized neural network has different degrees of improvement in the accuracy of predicting the total cost and reliability compared with the unoptimized neural network.

Key words: multi-method detection, condition-based maintenance, BP neural network, imperfect maintenance, sparrow search algorithm, genetic algorithm

中图分类号: 

版权所有 © 《现代制造工程》编辑部 
地址:北京市东城区东四块玉南街28号 邮编:100061 电话:010-67126028 电子信箱:2645173083@qq.com
本系统由北京玛格泰克科技发展有限公司设计开发 技术支持:support@magtech.com.cn