汽轮机转子典型振动故障诊断模型训练系统

Research of Turbine Rotor Vibration Fault Diagnosis for Typical Model Training System

文中基于支持向量机分类理论,运用二叉树算法建立了汽轮机转子典型故障的多分类诊断模型训练系统。通过小波包分析、经验模态分解(EMD)和傅里叶变换(FFT)三种信号处理方法训练出的诊断模型训练系统对测试样本分类的正确率、比较三种训练方法的优劣。

Based on the theory of support vector machine （SVM） classification, steam turbine rotor was established. In order to build a typical fault classification diagnosis model of training system, the binary tree algorithm is applied. Through wavelet packet analysis, empirical mode decomposition （EMD） and Fourier transform （FFT） of three kinds of signal processing methods based on the statistics of the diagnosis model for training system are analyzed. In a certain extent, the pros and cons of three kinds of training methods can be compared.