基于卷积神经网络算法的飞机发电机故障诊断研究  

Fault diagnosis of aircraft generator based on convolution neural network algorithm

作　　者：陈星 Chen Xing(Aviation Maintenance Engineeri ng College,Xi′an Aviation Vocat ional and Technical College,Xi′an 710089,China)

机构地区：[1]西安航空职业技术学院航空维修工程学院,西安710089

出　　处：《国外电子测量技术》2020年第11期57-60,共4页Foreign Electronic Measurement Technology

摘　　要：飞机电源系统中发电机是一个耦合性强的复杂非线性系统,确保发电机的稳定性非常重要,针对飞机发电机故障诊断过程中存在性能下降的问题,以飞机发电机系统结构为基础,引入卷积神经网络算法,以多源传感器采集的系统故障信号为输入,快速提取故障诊断信号,已形成样本数据集,然后建立飞机发电机的算法诊断模型,对飞机发电机的系统故障进行可视化的处理,在MATLAB神经网络工具箱中验证该算法模型的准确性,结果表明,采用的诊断算法能够提高飞机发电机故障诊断性能和可靠性。The generator in aircraft power system is a complex nonlinear system with strong coupling,so it is very important to ensure the stability of the generator.Aiming at the problem of performance degradation in the process of fault diagnosis of aircraft generator,based on the system structure of aircraft generator,the convolution neural network algorithm is introduced,and the system fault signal collected by multi-source sensors is taken as the input to quickly extract the fault.The fault diagnosis signal has formed the sample data set,and then the algorithm diagnosis model of aircraft generator is established,and the system fault of aircraft generator is visually processed.The accuracy of the algorithm model is verified in MATLAB neural network toolbox.The results show that the fault diagnosis algorithm can improve the performance and reliability of aircraft generator fault diagnosis.

关 键 词：飞机发电机 卷积神经网络 传感器 系统故障 可视化 

分 类 号：TN2[电子电信—物理电子学]