comparison of PCA and energy graph-based visualisation FDI on a heated two-tank process

In this study, the well-known Principal Component Analysis (PCA) Fault Detection and Isolation (FDI) method is compared against a recently developed hybrid FDI method, namely Energy Graph-Based Visualisation (EGBV) using twenty different process faults from a practical heated two-tank process. It was found that PCA is the more robust method as it has a lower overall False Alarm Rate (FAR), while EGBV is the superior fault isolation method. The best method for sensitive fault detection, determined using the True Alarm Rate (TAR), is dependent on the detection philosophy followed for EGBV.