Online fault detection and isolation using unbalanced residuals

This paper presents a FDI method based on standardized unbalanced residuals which are generated from the nodal imbalances of a process modeled by balance equations. The fault detection task is performed using a statistical test. The fault isolation decision maker proposed here is based on fault signatures and has the goal to reduce the rate of false alarms without increasing the rate of non-detections. The method is appropriate for online FDI because it is simple to implement, requires no special computations, no estimations of true values, and implies no iterations. The performance of the method was evaluated by computer simulation. Finally, the algorithm and an example are given.