Neural Control and Fault Diagnosis for 6/4 Switched Reluctance Motor

Prompt detection and diagnosis of faults in industrial plants are essential to minimize the production losses and increase the safety of the operator and equipment. Several conventional techniques are available in the literature to achieve these objectives. Neural networks are increasingly employed for fault diagnosis and control purposes. This paper presents neural based control and detection for a 6/4 switched reluctance motor. A neural network based optimal speed controller is designed with good robustness and performances are compared with fuzzy logic and conventional PI control. Two different structures of neural networks like back propagation (BP) and self-organizing map (SOM) neural networks have been used for detecting the faults for SR motor. Four different types of faults are introduced in the simulated system and detected using these networks. The simulation result is presented to demonstrate remarkable performance of the proposed controller and diagnosis scheme for the switched reluctance motor.