Diagnosis Kerusakan Bantalan Gelinding Pada Sistem Industri Dengan Metode Self Organizing Map (SOM)

Abstract

This research is discussing about the usage of data mining which addressed to damage diagnosis of rolling bearings. Data input was obtained from signal frequency feature extraction which taken from calibration against rolling bearings. The diagnosis was extremely important to industrial machines since this diagnosis can help to discover damages that occurred so that total failure of cessation of the machines can be avoided and industrial machines treatment costs can be optimized. Method used in this research is Self Organizing Map (SOM), SOM method on this research was done by sequence: signal frequency data that have been through the process of acquisition and preprocessing, feature extraction, Principal Component Analysis (PCA), then come into the process of SOM so that accuracy of the diagnosis process can be discovered. The result of this research is a software that can diagnose rolling bearings damage on industrial system. From tests result, software that has been produced was able to diagnose rolling bearings damage. Accuracy result shown 87.5% success, this software can be developed further to help technicians in diagnosing rolling bearings damage. This research method can be developed further to detect other damages in industrial systems.

Keywords: Data mining; PCA; SOM; Diagnosis; Rolling bearings; Statistic feature extraction