Past Issues

Studies in Informatics and Control
Vol. 4, No. 2, 1995

A Data Based Approach To On-line Process Fault Diagnosis

Jie Zhang
Abstract

This paper describes a data based technique for on-line process fault detection and diagnosis. The only knowledge required in this approach is process measurement data covering the events of various faults. The data can be obtained from the recorded operating history of a process or from simulation studies. They are usually the easiest available knowledge about a process since various process variables are measured during operations and those measurements can be easily collected and stored by a computer. Through multi-variable statistical data analysis, the features of various faults can be discovered and used in fault detection and diagnosis. In the technique presented here, principal component analysis is performed for the data corresponding to each fault and the loading vector of the first principal component is taken as the direction of the associated fault in the measurement space. During process supervision, principal component analysis is performed for the current on-line measurements whose direction is taken as the loading vector of the first principal component. Fault diagnosis is performed by comparing the direction of the current on-line measurements with that of various faults. The fault whose direction is very aligned with the current data direction is a plausible fault and is taken as the diagnosis result. The technique is very easy to implement and can be used to complement current fault diagnosis techniques. Applications of the proposed technique to the on-line fault diagnosis of a CSTR (continuous stirred tank reactor) system demonstrate that the technique is robust to measurement noise and effective in diagnosing faults.

Keywords

Fault diagnosis, process supervision, statistical data analysis, principal component analysis.

View full article