PROGNOSIS KERUSAKAN BANTALAN GELINDING DENGAN MENGGUNAKAN METODE SUPPORT VECTOR REGRESSION (SVR)

Abstract

This research has been performed on the prognosis of rolling bearing damage using by Support Vector Regression (SVR). This research was performed using rolling bearings object as a rolling bearing component capable of making a machine continues to spin or work. Prognosis for rolling bearing damage can optimize maintenance costs because it can determine the remaining life of rolling bearings function before rolling bearing is damaged. Data derived from the extraction of some statistical features of the data trend rolling bearing vibration signal. The resulting feature data used in the process of learning and testing processes with SVR method, as it will result in rolling bearing damage prognosis approaching the ideal value of Root Mean Square Error (RMSE) and Coefisien Correlation (R). The results of this research is RMS feature is a good feature to perform the prognosis. RMSE and R values for RMS feature is close to the ideal value. RMSE values for RMS feature is 0.0129, while the value of R for RMS feature is 0.9709.