摘要
为了解决航空发动机液压管路系统中管路故障诊断困难的问题,提出了一种基于深度置信网络(Deep Belief Networks,DBN)的航空液压管路智能故障诊断方法。首先,将采集的液压管路振动数据进行处理,提取出时频域特征参数,其次,将时频域特征参数作为输入样本,输入到深度置信网络模型中,利用深度置信网络模型进行液压管路故障的识别;最后,将本方法应用于航空液压管路模拟故障实验数据中,同时将本文方法与BPNN和SVM等方法进行对比分析,结果表明:本方法对液压管路故障的总体准确率达到99.27%,平均AUC值达到0.9937,同时表明本文建立的分类模型不仅能够实现航空液压管路状态的准确分类,而且对于管路单一故障和多故障并发情况也能精准识别。
In order to solve the problem of difficult pipeline fault diagnosis in the aero-engine hydraulic pipeline system,an intelligent fault diagnosis method of aero-hydraulic pipeline based on deep belief networks(DBN)is proposed.Firstly,the collected hydraulic line vibration data is processed to extract the time-frequency domain feature parameters.Secondly,the time-frequency domain feature parameters are used as input samples and input to the deep belief networks model established in this paper,and the deep belief networks model is used to identify the hydraulic line faults.Finally,the method of this paper is applied to the simulated fault experimental data of the aviation hydraulic line,and the method of this paper is compared with the methods of BPNN and SVM.The results show that the overall accuracy of this method for hydraulic line faults reaches 99.27%and the average AUC value reaches 0.9937.It also shows that the classification model established in this paper can not only achieve the accurate classification of aviation hydraulic line status,but also accurately identify single faults and multi-fault concurrent cases of the line.
作者
黄续芳
杨雪银
张小波
冯铃
HUANG Xufang;YANG Xueyin;ZHANG Xiaobo;FENG Ling(Sichuan Vocational College of Chemical,TechnologyInstitute of Intelligent Manufacturing,Luzhou Sichuan 646099,China;Linyi University Institute of Mechanical and Vehicle Engineering,Linyi Shandong 276000,China;Southwest UniversityInstitute of Electronics and Information,Chongqing 400715,China)
出处
《机械设计与研究》
CSCD
北大核心
2021年第5期94-98,105,共6页
Machine Design And Research
基金
中央高校基本科研业务费项目(XDJK2020D018)。
关键词
故障诊断
航空液压管路
深度置信网络
智能识别
fault diagnosis
aviation hydraulic pipeline
deep confidence network
intelligent recognition
