Fault diagnosis through integrated domain adversarial training and maximum average difference
Currently,the industrial Internet of Things has been successfully applied in the intelligent manufacturing industry.The large amount of data in the Internet of Things has promoted the development of industrial equipment health monitoring based on deep learning.Due to domain mismatch in the mechanical fault diagnosis monitoring data collected under different working con-ditions or equipment,models trained with training data may not be effective in practical applications.Therefore,it is crucial to study fault diagnosis methods with domain adaptive capabilities.An intelligent fault diagnosis method based on an improved do-main adaptive method was proposed in this paper.Specifically,two feature extractors for feature space distance and domain mis-match are trained using maximum average difference and domain adversarial training to enhance feature representation.Due to the separate classifier training feature extractor,further ensemble learning is utilized to obtain the final result.The experimental results show that this method is effective and has practical value in domain mismatch fault diagnosis.
domain adaptationdomain adversarial training(DAT)ensemble learningfault diagnosismaximum mean differ-ence(MMD)
万方数据
