Gearboxes are one of the complex machinery used widely in all process industries as speed reducers. Condition monitoring and fault diagnosis play an important role in increasing availability of machinery. In order to increase the reliability of fault diagnosis an effort has been made in this work to develop an ANN based diagnosis system with two prominent fault conditions of gears worn- out and broken tooth are being simulated. Vibration signal is collected and five feature parameters are extracted based on vibration signals and used as input features to the ANN diagnosis system developed in MATLAB, a three layered feed forward network using back propagation algorithm. The ANN system has been trained and tested with the learning rate, number of hidden layer neurons is varied one after the other and fixed optimal training parameters are identified.

Key Words: Gearbox Fault Diagnosis, Vibration signal, Artificial Neural Networks, Training data, optimization, Back propagation Algorithm.