Fault Diagnosis Method Based on Independent Component Analysis and Dynamic Time Warping

A fault diagnosis method was proposed by combining independent component analysis (ICA) and dynamic time warping (DTW). Wavelet analysis was firstly used to preprocess process data while ICA was to abstract independent components as data feature. DTW method flexibly matched fault data and fault pattern using dynamic programming principle. The minimal distance between two types of data sets was calculated for fault pattern diagnosis. Simulation results on Tennessee Eastman process show that the proposed method can detect faults more effectively than traditional PCA method, identify fault pattern and recognize new fault pattern successfully.