飞机舱门收放系统CPSO-BP神经网络故障仿真与诊断  

CPSO-BP neural network fault simulation and diagnosis of aircraft door retracting system

作　　者：王强[1] 吴伟[2] 刘东[1] 娄华语 王良模[2] WANG Qiang;WU Wei;LIU Dong;LOU Huayu;WANG Liangmo(Shenyang Aircraft Design&Research Institute,Shenyang 110000,China;School of Mechanical Engineering,Nanjing University of Science and Technology,Nanjing 210014,China)

机构地区：[1]航空工业沈阳飞机设计研究所,沈阳110000 [2]南京理工大学机械工程学院,南京210014

出　　处：《重庆理工大学学报(自然科学)》2023年第11期293-299,共7页Journal of Chongqing University of Technology:Natural Science

基　　金：科工局基础科研项目(JCKY2020205B014)。

摘　　要：针对民机舱门收放系统故障模拟代价大、故障数据少、故障诊断精度低的问题,提出基于CPSO-BP神经网络的飞机舱门收放系统故障诊断方法。根据民机舱门系统工作特性和高发故障的情况,确定流量控制阀磨损、液压马达泄漏、液压油污染和节流阀阻塞4种典型故障模式;建立飞机舱门AMESim收放系统仿真模型,通过典型故障的仿真分析获得120组故障数据,构建包含29520个样本的故障数据集;采用BP神经网络进行故障诊断,其平均诊断正确率仅为85.36%。采用混沌粒子群算法(CPSO)优化BP神经网络的初始权重和阈值,故障诊断正确率达到93%,提高了飞机舱门收放系统的故障诊断正确率。In this paper,a fault diagnosis method based on CPSO-BP neural network is proposed for civil aircraft cabindoor retracting system to address such issues as high simulation cost,limited fault data,and low fault diagnosis accuracy.Considering the operating characteristics and frequent faults of the civil aircraft cabindoor system,four typical failures,namely,flow control valve wear,hydraulic motor leakage,hydraulic oil pollution and throttle valve blockage,are identified.Based on AMESim,an aircraft cabindoor system simulation model is built,120 sets of fault data are obtained by simulation analysis of the typical faults,and a fault data set containing 29520 samples is built.A fault diagnosis by BP neural network reveals the average accuracy rate of fault diagnosis only stands at 85.36%.In comparison,the chaotic particle swarm optimization algorithm(CPSO),which optimizes both the initial weight and threshold of BP neural network,increases the accuracy rate to 93%,demonstrating better fault diagnosis performance.

关 键 词：故障诊断 AMESIM 飞机舱门收放系统 BP神经网络 混沌粒子群优化算法(CPSO) 

分 类 号：V37[航空宇航科学与技术—航空宇航推进理论与工程] V328.5