SHAP를 활용한 중요변수 파악 및 선택에 따른 잔여유효수명 예측 성능 변동에 대한 연구
A Study on the Remaining Useful Life Prediction Performance Variation based on Identification and Selection by using SHAP
1
조회
0
다운로드
4
인용
0
참고문헌
키워드
초록
Recently, the importance of preventive maintenance has been emerging since failures in a complex system are automatically detected due to the development of artificial intelligence techniques and sensor technology. Therefore, prognostic and health management (PHM) is being actively studied, and prediction of the remaining useful life (RUL) of the system is being one of the most important tasks. A lot of researches has been conducted to predict the RUL. Deep learning models have been developed to improve prediction performance, but studies on identifying the importance of features are not carried out. It is very meaningful to extract and interpret features that affect failures while improving the predictive accuracy of RUL is important. In this paper, a total of six popular deep learning models were employed to predict the RUL, and identified important variables for each model through SHAP (Shapley Additive explanations) that one of the explainable artificial intelligence (XAI). Moreover, the fluctuations and trends of prediction performance according to the number of variables were identified. This paper can suggest the possibility of explainability of various deep learning models, and the application of XAI can be demonstrated. Also, through this proposed method, it is expected that the possibility of utilizing SHAP as a feature selection method.
Abstract
Recently, the importance of preventive maintenance has been emerging since failures in a complex system are automatically detected due to the development of artificial intelligence techniques and sensor technology. Therefore, prognostic and health management (PHM) is being actively studied, and prediction of the remaining useful life (RUL) of the system is being one of the most important tasks. A lot of researches has been conducted to predict the RUL. Deep learning models have been developed to improve prediction performance, but studies on identifying the importance of features are not carried out. It is very meaningful to extract and interpret features that affect failures while improving the predictive accuracy of RUL is important. In this paper, a total of six popular deep learning models were employed to predict the RUL, and identified important variables for each model through SHAP (Shapley Additive explanations) that one of the explainable artificial intelligence (XAI). Moreover, the fluctuations and trends of prediction performance according to the number of variables were identified. This paper can suggest the possibility of explainability of various deep learning models, and the application of XAI can be demonstrated. Also, through this proposed method, it is expected that the possibility of utilizing SHAP as a feature selection method.
참고문헌 (22)
- Adadi, A., "Peeking inside the black-box : a survey on explainable artificial intelligence(XAI)", IEEE Access, 6, pp.52138~52160 (2018)
- Ahn, G., "Explainable convolutional neural networks for multisensor data", Journal of the Korean Institute of Industrial Engineers, 45(2) (2019)
- Babu, G.S, "Deep convoultional neural network based regression approach for estimation of remaining useful life", International Conference on Database Systems for Advanced Applications, pp.214~228 (2016)
- 최요환, "배터리 잔존 유효 수명 예측을 위한 전기화학 모델 기반 고장 예지 및 건전성 관리 기술", 한국통신학회논문지, 42(4), pp.939~949 (2017), 한국통신학회[DOI]
- Hancer, E., "Differential evolution for filter feature selection based on information theory and feature ranking", Knowledge-based systems, 140, pp.103~119 (2018)
- Hong, C. W., "Remaining useful life prognosis for turbofan engine using explainable deep neural networks with dimensionality reduction", Sensors, 20(22), pp.6626~ (2020)
- Hong, C. W., "Prognosis of the remaining useful life of a turbofan engine using deep neural network and pearson correlation coefficient", Journal of the KNST, 4(1), pp.078~083 (2021)
- 정찬일, "프로세스 분석을 위한 설명 가능한 인공지능 기법 비교 연구", 전자공학회논문지, 57(8), pp.51~59 (2020), 대한전자공학회
- 정상진, "딥러닝을 이용한 리튬이온 배터리 잔여 유효수명 예측", 한국기계가공학회지, 19(12), pp.21~27 (2020), 한국기계가공학회[DOI]
- 고기혁, "설명가능한 인공지능을 위한 특성기여도 분석 방법론 조사", 정보과학회논문지, 47(12), pp.1181~1191 (2020), 한국정보과학회[DOI]
- 권혁건, "RNN(Recurrent Neural Network)을 이용한 기업부도예측모형에서 회계정보의 동적 변화 연구", 지능정보연구, 23(3), pp.139~153 (2017), 한국지능정보시스템학회
- Laguna, J.O, "A dynamic sliding window approach for activity recognition", International conference on user modeling, adaptation, and personlization, pp.219~230 (2011)
- Li, J., "A directed acyclic graph network combined with CNN and LSTM for remaining useful life prediction", IEEE Access, 7, pp.75464~75475 (2019)
- Wang, J, "Remaining useful life estimation in prognostics using deep bidrectional lstm neural network", IEEE, In 2018 prognostics and system health management conference(PHM-chongqing), pp.1037~1042 (2018)
- Marcilio, W.E., "From explanations to feature selection: accessing SHAP values as feature selection mechanism", 2020 33rd SIBGRAPI conference on Graphics, Patterns and Images(SIBGRAPI), pp.340~347 (2020)
- 오재영, "인공지능 해석 기법을 이용한 태양광 발전량 예측 성능 향상", 전기학회논문지, 69(7), pp.1112~1117 (2020), 대한전기학회
- Seijo-Pardo, B., "Ensemble feature selection : homogeneous and heterogeneous approaches", Knowledge-Based Systems, 118, pp.124~139 (2017)
- 위지원, "DNN 모델을 이용한 기계 학습 기반 k-최근접 질의 처리 최적화 기법", 한국콘텐츠학회 논문지, 20(10), pp.715~725 (2020), 한국콘텐츠학회[DOI]
- Yu, S, "A domain adaptive convolutional LSTM model for prognostic remaining useful life estimation under variant conditions", IEEE, In 2019 Prognostics and System Health Management Conference(PHM-Paris), pp.130~137 (2019)
- 윤유리, "C-MAPSS 데이터를 이용한 항공기 엔진의 신경 회로망 기반 건전성관리", 항공우주시스템공학회지, 13(6), pp.17~25 (2019), 항공우주시스템공학회[DOI]
- Zhang, A., "Transfer learning with deep recurrent neural networks for remaining useful life estimation", Applied Sciences, 8(12), pp.2416~ (2018)
- Zhang, X., "Embedded feature-selection support vector machine for driving pattern recognition", Journal of the Franklin Institute, 352(2), pp.669~685 (2015)