Fault detection and identification using a hierarchical neural network-based system

A new approach to detect and identify faults in complex processes is proposed. The approach is based on a hierarchical neural network structure. Other neural network applications in process fault diagnostics provide only fault detection and isolation. Through the proposed scheme, fault detection, isolation, and identification (recognizing the size of fault) can be achieved. This is due to the higher learning ability of the hierarchical structure. The performance of the suggested fault detector and identifier is evaluated via an industrial case study. The results show a satisfactory level of accuracy.<
>