PERANCANGAN SISTEM DIAGNOSIS GETARAN MOTOR MENGGUNAKAN JARINGAN SARAF TIRUAN PROPAGASI MUNDUR

Abstract

Expert system design is an effective and sophisticated way of diagnosing a fault in a 12 kW DC Motor. This study aims to design an ANN system to determine damage to the motor. The research method uses spectrum data from the vibration analyzer which is collected based on different types of damage. The training data patterns from the spectrum characteristics to be used in the system, the goal is that the systems can recognize the patterns that have been made. The training data patterns that have been successfully recognized by the system are then tested. The results of training and ANN testing are quite good, with the greatest Cross-Entropy value of 9.94, having 0% error value, the largest Mean Square Error value 8.33e-6 and the smallest regression 0.998. A testing of 8 new spectrums resulted in accurate predictions.