개인구독
소속 기관이 없으신 경우, 개인 정기구독을 하시면 저렴하게
논문을 무제한 열람 이용할 수 있어요.
지원사업
학술연구/단체지원/교육 등 연구자 활동을 지속하도록 DBpia가 지원하고 있어요.
커뮤니티
연구자들이 자신의 연구와 전문성을 널리 알리고, 새로운 협력의 기회를 만들 수 있는 네트워킹 공간이에요.
논문 기본 정보
- 자료유형
- 학위논문
- 저자정보
- 지도교수
- 홍선기
- 발행연도
- 2021
- 저작권
- 호서대학교 논문은 저작권에 의해 보호받습니다.
이용수25
이 논문의 연구 히스토리 (11)
2022
2021
2020
2019
- 이 논문의 후속연구가 궁금하신가요?
- 연관 학술논문 또는 학술발표를 통해 보다 발전된 연구결과를 확인하실 수 있습니다.
- 이 논문의 연구 히스토리 확인하기
초록· 키워드
오류제보하기
The most problem when applying deep learning algorithms to motor fault diagnosis and prediction is outlier data. In the previous studies, it is assumed that all states data for target motor exist in the big data set. However, this is data that is difficult to collect. This assumption for convenience of experimentation reduces the versatility of the system. To solve this problem, a CNN-based motor fault diagnosis and prediction system considering outlier data is proposed.
In the proposed system, a diagnosis model using DT-CNN, an open set recognition technique capable of classifying outlier data, is used. Also, a prediction model that learns the RMSE pattern is used. The prediction model examines the similarity based on the outlier data classified in the diagnosis model. In order to develop the proposed system, a measurement environment based on LABVIEW is constructed. The possibility of using the CNN-based algorithm model is confirmed. Hyper-parameter optimization using genetic algorithms that help system learning and virtual data generation techniques using GAN algorithms are studied. The proposed system is constructed through the above studies. The robustness of the proposed system against transient disturbances is confirmed. In addition, the performance of motor aging diagnosis and failure prediction is evaluated. The proposed system is expected to help more general-purpose motor fault diagnosis and prediction.
In the proposed system, a diagnosis model using DT-CNN, an open set recognition technique capable of classifying outlier data, is used. Also, a prediction model that learns the RMSE pattern is used. The prediction model examines the similarity based on the outlier data classified in the diagnosis model. In order to develop the proposed system, a measurement environment based on LABVIEW is constructed. The possibility of using the CNN-based algorithm model is confirmed. Hyper-parameter optimization using genetic algorithms that help system learning and virtual data generation techniques using GAN algorithms are studied. The proposed system is constructed through the above studies. The robustness of the proposed system against transient disturbances is confirmed. In addition, the performance of motor aging diagnosis and failure prediction is evaluated. The proposed system is expected to help more general-purpose motor fault diagnosis and prediction.
목차
등록된 정보가 없습니다.
최근 본 자료
댓글(0)
0