基于小波分析的自动控制系统的故障诊断

The Detection of Fault in Control System Based on Wavelet Transform

工业控制系统常发生的故障是执行机构和检测装置的故障 ,且故障信号多是突变性的。传统的Fourier分析由于在时域缺乏空间局部性 ,只能确定一个函数奇异性的整体性质 ,而难以确定奇异点在空间的位置及分布情况 ,难以检测到突变信号。小波变换具有空间局部化性质 ,而且时域窗和频域窗的宽度可调节。对系统的输入、输出信号进行小波变换 ,利用该变换求出输入输出信号的奇异点。

The fault that often happens in industrial automatically controlled system is the error of executors and detectors. The faulty signal is often broken signal. Lacking space locality in time domain, Fourier analysis can only make certain of the integrity of a function's singularity. So it's hard to detect the spacial position and distribution of broken signal. Wavelet transform has the character of spacial locality, and its wideness of time's window and frequent window can be adjusted. Wavelet transform is exploited to get the input and output points of singularity. Simulating experiment proves the superiority of using wavelet transform