论文部分内容阅读
提出一种基于灰色关联度分析优化堆栈自动编码器的故障特征自适应提取方法,并用于航空发电机的旋转整流器二极管故障诊断中。首先,采集发电机交流励磁机励磁电流信号;其次,借助灰色关联度和深度学习理论对堆栈编码器网络进行训练学习,以确立其较优的网络结构,通过该网络可以自适应地从励磁电流信号中提取故障特征;训练完毕,借助于支持向量机(support vector machine,SVM)分类器实施故障诊断。对所提方法与快速傅里叶变换方法进行了仿真和物理实验,并对分类性能进行比较。结果表明,所提方法自动化程度高,自适应性能好,所提取的特征用SVM评估可以取得很好的分类效果。
This paper presents a method of adaptive fault feature extraction based on gray relational analysis and optimized stack automatic encoder. It is also used in fault diagnosis of rotating rectifier diode of aerogenerator. Firstly, the generator excitation current signal of AC exciter is collected. Secondly, the stack encoder network is trained by the gray relational degree and depth learning theory to establish its optimal network structure. From this network, adaptive excitation current The fault feature is extracted from the signal. After the training, the fault diagnosis is implemented by means of a support vector machine (SVM) classifier. The proposed method and the Fast Fourier Transform method were simulated and the physical experiments were carried out, and the classification performance was compared. The results show that the proposed method has a high degree of automation and good self-adaptability. The extracted features can be well classified by SVM.