LAUSR

Accurate detection and isolation of an incipient fault is crucial for efficient and optimal operation of industrial processes. In the area of statistical process monitoring (SPM), reconstruction-based methods are widely used for fault isolation purposes. Though possessing several merits, most existing conventional reconstruction methods are not efficient in isolating incipient faults that have small magnitudes. This paper proposes a new method called exponential smoothing reconstruction (ESR) for incipient fault isolation. The proposed ESR introduces the exponentially weighted moving average (EWMA) technique to conventional reconstruction methods. Fault isolability analysis of the ESR is provided, which demonstrates its efficiency in isolating incipient faults, compared with conventional reconstruction-based methods. Several remarks on the use of ESR, such as selection of the smoothing parameter and some practical issues, are discussed as well. Finally, simulation studies on a numerical example and a cont...