文献类型：期刊文献

英文题名：Owfd-Up: An Open-World Fault Diagnosis Scheme Based on Uncertainty and Prototypes

作者：Gao, Fulin[1]; Zhong, Weimin[1]; Jiang, Qingchao[1]; Peng, Xin[1]; Li, Zhi[1]

机构：[1] Key Laboratory of Smart Manufacturing in Energy Chemical Process, Ministry of Education, East China University of Science and Technology, Shanghai, 200237, China

年份：2023

外文期刊名：SSRN

收录：EI(收录号：20230084643)

语种：英文

外文关键词：Fault detection - Feature Selection - Standardization - Uncertainty analysis - Wastewater treatment

摘要：In real-world fault diagnosis, it is never possible to enumerate all fault types beforehand, since there are always unseen situations that may arise unexpectedly. Therefore, fault diagnosis is essentially an open-world recognition task. A desirable open-world model must be able to perform (1) open-set recognition (OSR), i.e., classify the known and reject the unknown and (2) incremental learning (IL), i.e., incrementally learn knowledge of new classes and retain memory of old classes. In this work, we formulate an open-world fault diagnosis scheme based on uncertainty and prototypes to address this problem. Firstly, we provide a paradigm for data standardization applicable to the open world. Secondly, we low-overhead estimate the probability that a sample belongs to the unknown based on the uncertainty in the classifier's probability prediction for OSR. Then, valid new samples are manually screened and labeled. Finally, we leverage the dissimilarity-based sparse subset selection algorithm to pick more diverse and valuable exemplars from old classes as prototypes for IL. Furthermore, we extend the open-set F-Measure (OSFM) to the weighted OSFM to make it applicable even when the classes are imbalanced. Experimental results on multiphase flow facility and wastewater treatment plant demonstrate the feasibility and superiority of the proposed method. ? 2023, The Authors. All rights reserved.