Fault Detection in a Heat Exchanger, Comparative Analysis between Dynamic Principal Component Analysis and Diagnostic Observers

Abstract

Abstract. A comparison between the Dynamic Principal Component Analysis (DPCA) method and a set of Diagnostic Observers (DO) under the same experimental data from a shell and tube industrial heat exchanger is presented. The comparative analysis shows the detection properties of both methods when sensors and/or actuators fail online, including scenarios with multiple faults. Similar metrics are defined for both methods: robustness, quick detection, isolability capacity, explanation facility, false alarm rates and multiple faults identifiability. Experimental results show the principal advantages and disadvantages of both methods. DO showed quicker detection for sensor and actuator faults with lower false alarm rate. Also, DO can isolate multiple faults. DPCA required a minor training effort; however, it can not identify two or more sequential faults.