Fault detection and diagnosis using empirical mode decomposition based principal component analysis

Abstract

This paper presents a new algorithm to identify and diagnose stochastic faults in Tennessee Eastman (TE) process. The algorithm combines Ensemble Empirical Mode Decomposition (EEMD) with Principal Component Analysis (PCA) and Cumulative Sum (CUSUM) to diagnose a group of faults that could not be properly detected and/or diagnosed with previously reported techniques. This algorithm includes three steps: measurements pre-filtering, fault detection, and fault diagnosis. Measured variables are first decomposed into different scales using the EEMD-based PCA, from which fault signatures can be extracted for fault detection and diagnosis (FDD). The T2 and Q statistics-based CUSUMs are further applied to improve fault detection, where a set of PCA models are developed from historical data to characterize anomalous fingerprints that are correlated with each fault for accurate fault diagnosis. The algorithm developed in this paper can successfully identify and diagnose both individual and simultaneous occurrences of stochastic faults.