The Fault Diagnosis Of Continuous Industrial Process Based On Wavelet Theory And Principal Component Analysis

With the development of modernized wholesale manufacture and theprogress of science and technology, the structures of continuous industrialprocess are more complex, the functions are more consummate and theautomatic degree is higher. For the existence of inevitable factors, there aremany faults in these industrial systems. The results of these faults maydebase or invalidate the functions, or cause catastrophic accidents. So it hasimportant academic and practical meanings to research industrial processfault diagnosis. According to the status pro of the theories and industrial applicationsof industrial process fault diagnosis technology at home and abroad, thispaper studies the process and essential of fault diagnosis, analyzes somefault diagnosis technologies, and emphasizes on the methods of waveletanalysis, autoassociative neural network and principal component analysis.For the advantages of wavelet analysis and PCA, this paper proposes a II北 京 化 工 大 学 学 位 论 文 用 纸continuous industrial process fault diagnosis solution using these twomethods. The solution has been applied to the fault diagnosis system ofethylene pyrolysis furnace. The validity and effectivity of the solution isproved and the fault diagnosis results are satisfied. At last, this paper pointsout the problems to be further researched and the development trends infault diagnosis fields.