功率谱在调速阀故障诊断中的性能分析

摘要

By means of wavelet packet, this paper first extracts features from the AR power spectrum of signals in normal and fault state respectively, then the features are input to a LSSVM as parameters to diagnose faults, and the result is satisfactory. Furthermore, the result is compared with that of primary signals, and the causes are interpreted.%利用小波包分别提取调速阀振动时的正常信号和故障信号的AR功率谱的特征值,之后利用提取的特征值作为LSSVM的输入,对调速阀的故障进行诊断,取得了良好效果,并与原始信号的效果进行了对比,并阐明了原因.