A HYBRID ANN-BASED FAULT DIAGNOSIS AND TOLERANCE METHOD AND ITS APPLICATION

In this study, a hybrid dynamic Artificial Neural Network (ANN)-based fault diagnosis and tolerance method is developed. The adopted hybrid ANN is a combination of feedforward ANN and recurrent ANN forming a dynamic identification model for the non-linear time-varying system. It has three work modes and can perform the fault and degradation diagnosis and tolerance by using these modes alternately. The result of its application in an Electro-Hydraulic Servomechanism in Hydroelectric Generation Unit shows its effectiveness and ability of online implementation without importing disturbance signals to the system.