Sistem Pendeteksian Kerusakan Mesin Sepeda Motor 4-Langkah Berbasis Suara Menggunakan Support Vector Machine (SVM)

Abstract

Detection process early towards motorcycle engine condition will be important matter especially for common user motorcycle. This detection can be used to estimate motorcycle engine condition (normal or damage), damage kind, how big damage influence towards motorcycle continuance, motorcycle duration can survive with damage and cost estimate that taked suppose will repair damage. In this research is built 4-stroke motorcycle engine damage detection system based on voice uses Support Vector Machine (SVM) multi class. In system that proposed, motorcycle engine voice is recorded and then cultivated so that produce feature shaped coefficient Linear Predictive Coding (LPC). Coefficient LPC that extracted from this motorcycle engine voice then become an input for SVM. Furthermore SVM will determine motorcycle engine condition. Engine condition detection system based on SVM this meant to detect three engine conditions that is normal condition, damage cham chain and damage ignition system. System applications that proposed show that motorcycle engine condition detection system based on voice uses SVM has good accuracy that is 100%.