Fault Detection And Diagnosis Of Continuous Process Based On Data-Driven Method

The fault detection and diagnosis of continuous process is very important for the production safety and product quality. Owing to its no need to know much about the process mechanism and exact process model, the data driven method, typically the principal component analysis (PCA) has attracted much attention of chemical researchers for monitoring process.This thesis researches the application of PCA on the fault detection and diagnosis of continuous process. The principle of PCA and corresponding statistics are introduced and applied to the benchmark Tennessee Eastman process (TE process) for simulation. The simulation results show that PCA is powerful in fault detection, however, there are some difficulties for PCA in diagnosing fault correctly and it can not find the fault position exactly in complex process.The MBPCA is presented for fault detection and diagnosis in continuous process. This method builds the integral PCA model of normal process data to detect fault and divides the data into function-independent block or site-independent block to build the block PCA model. Once a fault has been detected, we first judge which block it belongs to and then determines the fault location by the variables contribution of that block and process knowledge. The essence of this approach decomposes the large-scale system into several blocks to enhance the model's ability of explanation and fault diagnosis. The simulations on the TE process show that the proposed method can not only detect fault quickly, but also find the fault position exactly.This thesis also explores the different blocking method of MBPCA, compares the effect of fault diagnosis of TE process based on different blocking method and proposes some suggestions of blocking.