Fault Diagnosis For Industrial Propylene Polymerization Based On MSPCA

In the light of special characteristics of industrial propylene polymerization, we raises an improved multi-scale principal component analysis method is proposed. Based on the theory of wavelet and principal component analysis, multi-scale principal component analysis is introduced which combines the ability of PCA to decorrelate the variables by extracting a linear relationship, with that of wavelet analysis to extract deterministic features and approximately decorrelate autocorrelated measurements. Main research work and contribution of this dissertation are as following:When applying conventional PCA to fault detection, it would lead to false-alarm of the system due to the measured data corrupted with noise and outliers. To overcome the 1 imitations of conventional PCA handling the data corrupted with noise and outliers, an approach is developed by combining the wavelets transform , wavelet threshold de-noising and PCA. This method utilizes the advantage of wavelets transform and wavelet threshold de-noising to preprocess the date to eliminate noise and outliers. Using wavelet transforms, the individual variable signals are decomposed into approximations and details at different scales. Coefficients from each scale are collected in separate matrices, and a PCA model is then constructed to extract correlation at each scale , and a PCA model is then constructed to extract correlation at each scale .At last, this method is applied to fault detection and has a good effect which proves the method is effective and feasible.The results of application on Propylene Polymerization process demonstrate that improved MSPCA is able to efficiently monitor performance changes of process, and accurately identify faults and diagnose the causes of faults in time. Comparing with PCA and MSPCA, improved MSPCA can be used to effectively detect different resolution variation, decrease false alarms, and increase the reliability of process monitoring.