Fault Detection And Diagnosis Based On Wavelet Analysis

As the modern technology develops, the engineering control systems become more and more complicated and advanced in practical application, therefore the reliability and the security of those systems will play a critical role to both the society and the economy. More and more attention has been paid in the reliability and security in the last two decades. Fault diagnosis technology has therefore become a hot issue in the area of control theory. Chemical production process is a high degree of automation, in case of fault, it will cause huge economic losses, so the fault diagnosis of chemical process is of practical significance.Wavelet analysis is a kind of method to process non-stationary signal in these years and has good time-frequency localization characters. Fault diagnosis based on wavelet analysis does not need the precise mode of system. However, due to the presence of noise, there is a high rate of false alarm in the wavelet fault diagnosis methods. It severely limits the wavelet fault diagnosis method in practice. While the denoising effect of existing wavelet denoising methods can not meet demand of modern fault diagnosis.First, we will briefly introduce the state of fault diagnosis. The major aspects on its theory and techniques will be summarized. The applications of the wavelet analysis in fault diagnosis will also be introduced.Second, we will compare wavelet analysis and fourier analysis and introduce wavelet transform. Wavelt multiscale theory and Mallat arithmetic will also be introduced particularly. We will use wavelet multiscale theory to research feature extraction and detect fault more accurately via extracting high frequency and low frequency feature on multiscale.Third, according to the status of wavelet threshold de-noising, we will bring forward a new wavelet threshold de-noising method based on wavelet entropy. Results of simulation test show the effectiveness of the method.At last, the proposed method in the dissertation is applied into the simulation of faults diagnosis of CSTR, results of simulation test show the effectiveness of the method.