Fault detection of tank-system using ChangeFinder and SVM

Abstract

Tank-system is used in the process industry and needs to detect fault early to increase productivity. In recent years, research on fault detection by processing real-time sensing data has been actively conducted. The early fault detection method makes the system dependable. In this paper, a fault detection method using Support Vector Machine (SVM) combined with ChangeFinder, both are machine learning methods, is proposed, and the effectiveness is confirmed by simulation and experiment.