Condition Monitoring and Faults Diagnosis for Synchronous Generator Using Artificial Neural Networks

Abstract

Early detection and diagnosis of incipient fault is desirable for on line condition assessment production quality assurance and improved operational efficiency of synchronous generator running of power supply. Artificial Intelligent techniques are increasly used for condition monitoring and fault diagnosis of machines. In this paper, Artificial Neural Network (ANN) approach employed for fault diagnosis in the generator, based on monitoring generator currents to give indication of the winding faults. Feed-forward Network, error back propagation training algorithm are used to perform the generator faults diagnosis and their values. NN which has been trained for all possible operating condition of the machine used to classify the incoming data. The inputs of the NN are the stator and rotor currents, and the output represents the running condition of the generator. The training of the NN achieved by the data through a mathematical model based approach to simulate the generator faults at various degree of severity.This paper evaluates through simulation line currents magnitude of the generator .The final results have been represented on a monitoring unit, built using matlab program, to give early warning of the generator failure. DOI : http://dx.doi.org/10.11591/telkomnika.v12i2.3809