컴퓨터지능 및 지능시스템] 심층 학습 및 IoT 기법을 이용한 유도 전동기의 실시간 고장 진단과 자기 보완 시스템

Abstract

Existing fault diagnosis using deep Learning, has experimented by collecting data from a controlled environment. However, it is not easy to diagnose motor faults in various environments, becuase input data are measured with various disturbances together. For this reason, in this paper, the verification and learning process are separated and used in each system so that motor data of various environments can be considered. In the verification process, a data preprocessing process is added to verify and collect necessary data. A CNN-based in-depth learning algorithm is implemented and data is stored in real time. Since the data is re-learned based on the collected data, a model considering both existing features and newly input data is created. Even the continuous disturbance is also used as learning data, it is easier to cope with disturbance than the conventional method. As a result, a system applicable to an industrial field is proposed considering various environments.