DETEKSI MISALIGNMENT PADA MOTOR INDUKSI MENGGUNAKAN TRANSFORMASI WAVELET DISKRIT DAN FUZZY SUBSPACE CLUSTER

Abstract

Currently induction motors are widely used in industry because of their strong construction,high efficiency, and low maintenance. Machine maintenance is necessary to extend the life of the induction motor. Based on previous research, bearing faults can cause 42% - 50% of all motor failures. Generally this is caused by manufacturing errors, lack of lubrication and installation errors. Motor misalignment is one of the errors in installation. This research is concerned with descrete wavelete transform simulations to identify misalignment in induction motors. Modeling of motor operation is introduced in this paper as normal operation and two variations of misalignment. Haar and symlet wavelet transformations at the first level to the third level are used to extract the motor vibration signal into a high frequency signal. Then the energy signal and other signal extracts obtained form the high frequency signal are evaluated to analyze the condition of the motor. This evaluation process uses fuzzy logic of the fuzzy subspace cluster type. The results of research using a combination method of signal processing in the form of DWT and artificial intelligence methods of the fuzzy subspace cluster type. Then the occurrence of misalignment in three-phase induction motors can be detected early. So that maintenance and replacement can be anticipated before misalignment occurs. From the experimental result, it was obtained that motor and clutch endurance test for level 1 of the fuzzy subspace cluster method was 0,88% better than the fuzzy c-mean method of 0,75%.