Fault Classification of Pump Using Support Vector Machine (SVM) Method

Abstract

A machine is mechanical or electrical equipment that converts energy to aid human activities or manufacture specific items. A machine's condition should be maintained and checked in proper working order. As a result, the machine's state must be determined before major harm occurs. The goal of this research is to use machine learning to detect the type of pump damage. The focus of this investigation was a Panasonic GP-129 water pump. The purpose of this research is to classify three types of pump faults: misalignment, imbalance, and bearing fault. Based on the results obtained, the classification of pump fault using Support Vector Machine (SVM) methods had average accuracy of 98.35% on the Linear SVM and Cubic SVM models, and average accuracy of 100% on the Quadratic SVM model.