Canonical correlation analysis and key performance indicator based fault detection scheme with application to marine diesel supercharging systems

The main objective of this paper is to propose a fault detection scheme for marine diesel supercharging systems based on the canonical correlation analysis and key performance indicator approach. To be specific, the key performance indicator-based system model is first constructed with measurable variables. Then, the canonical correlation analysis is used to identify the established system model. By applying the T2 test statistic, a fault detection is presented for process monitoring. In this end, a simulation on the marine diesel supercharging system is given to illustrate the effectiveness of the developed results.