웨이블릿 변환을 이용한 주기 신호의 실시간 이상 탐지에 대한 연구

A real-time fault detection method using wavelet tranform is proposed in this paper. First, de-noised signals are obtained by wavelet noise reduction method. Next, wavelet coefficients that well reflect the process condition are selected by comparing de-noised signals to original signals and statistical process control (SPC) charts of each coefficient and Hotelling’ T2 chart are constructed for monitoring the process condition. With these charts, faults are detected during the process is in activation and after completion of the process. In addition, identification of the location and the type of the detected fault is possible. For performance evaluation, various types of abnormal process conditions are simulated and the proposed algorithm is compared with other methodologies.