Diagnosis of motor aging through CNN model using signal correlation

Unlike motor failure diagnosis, motor failure prediction is more difficult to collect data and develop algorithms. The classic motor failure prediction is a method of determining the characteristics of a motor failure and finding a signal accordingly. However, this requires specialized knowledge of the motor, and this results in poor versatility. Also, it is not easy to classify the signal change due to the aging of the motor and the signal change due to the failure. To solve this problem, a correlation analysis algorithm is proposed between the motor steady-state signal and the collected signal. The proposed algorithm is composed of deep learning algorithms using the values rather than simple correlation analysis. It is data-driven, unlike the classic method based on prior knowledge. In addition, this algorithm can detect the change of the motor signal by FFT the motor vibration signal by 100 ms. Through the proposed algorithm, the classification of the signal of aging and the failure signal of the motor will proceed.