航发喘振智能化故障诊断模型研究  

Research on Intelligent Fault Diagnosis Model of Aero-Engine Surge

作　　者：张柯欣 郑德生 吴欣隆 陈继鑫 ZHANG Ke-xin;ZHENG De-sheng;WU Xin-long;CHEN Ji-xin(School of Computer Science,Southwest Petroleum University,Sichuan Chengdu 610500,China)

机构地区：[1]西南石油大学计算机科学学院,四川成都610500

出　　处：《计算机仿真》2023年第3期25-30,484,共7页Computer Simulation

基　　金：四川省科技计划重点研发项目(2019YFG0424);西南石油大学科研“启航计划”项目(604)。

摘　　要：针对航空发动机的喘振故障严重影响飞机安全运行的问题,提出了一种基于深度学习的航发喘振智能化故障诊断模型。结合喘振的生成机理对航空发动机传感器数据进行喘振故障分析,采用基于滑动窗口的数据预处理算法构造数据集和标签集;集成卷积神经网络(Convolutional Neural Network, CNN)和长短期记忆(Long-Short Term Memory, LSTM)网络,设计出针对航发喘振故障诊断的深度神经网络模型(1D-CLSTM);在所构建的数据集上,对所提模型与当下流行的深度神经网络进行比较。实验结果表明,所提模型对喘振故障分类的F1分数(F1_score)、召回率(Recall)和精确度(Precision)分别达到了96.45%、95.48%、97.46%,优于其它网络模型。所提模型在时序信号处理与旋转机械智能化故障诊断方面有着较高的应用和推广价值。Aiming at the problem that aero-engine surge seriously affects the safe operation of aircraft,an intelligent fault diagnosis model of aero-engine surge based on deep learning was proposed.Firstly,combined with the generation mechanism of surge,the surge fault of aero-engine sensor data was analyzed,and the data set and label set were constructed by using the data preprocessing algorithm based on sliding window;Then,integrating convolutional neural network(CNN)and long short term memory(LSTM)network,a deep neural network model(1D-CLSTM)for Aero-engine surge fault diagnosis was designed;Finally,on the constructed data set,the proposed model was compared with the current popular deep neural network.Experimental results show that the F1 score,recall rate and accuracy of the proposed model for surge fault classification reach 96.45%,95.48%,and 97.46%,respectively,which are better than other network models.The proposed model has high application and promotion value in timing signal processing and intelligent fault diagnosis of rotating machinery.

关 键 词：航空发动机 深度学习 喘振 故障诊断 神经网络 

分 类 号：V271.4[航空宇航科学与技术—飞行器设计]