Deep Fault Diagnosis for Rotating Machinery with Scarce Labeled Samples
Deep Fault Diagnosis for Rotating Machinery with Scarce Labeled Samples作者机构：School of Automation Hangzhou Dianzi University School of Electronic and Information Engineering Beijing Jiaotong University School of Mathematics and Statistics Henan University Hofon Automation Co. Ltd
出 版 物：《Chinese Journal of Electronics》 (电子学报(英文))
年 卷 期：2020年第29卷第4期
页 面：693-704页
核心收录：
学科分类：0711[理学-系统科学] 0808[工学-电气工程] 0809[工学-电子科学与技术（可授工学、理学学位）] 07[理学] 08[工学] 080401[工学-精密仪器及机械] 080203[工学-机械设计及理论] 0804[工学-仪器科学与技术] 080402[工学-测试计量技术及仪器] 0802[工学-机械工程] 0701[理学-数学] 080201[工学-机械制造及其自动化]
基 金：supported by the National Key Research and Development Program(No.2016YFE0200900) the National Natural Science Foundation of China(No.61806064,No.61806062,No.61751304,No.61873077) Open Foundation of Key Laboratory of Advanced Public Transportation Science,Ministry of Transport,PRC
主 题：Fault diagnosis Deep CNN Spectrograms Knowledge-transferring
摘 要：Early and accurately detecting faults is crucial for the modern manufacturing *** proposed a novel Deep fault diagnosis(DFD) method for rotating machinery with scarce labeled samples.A spectrogram of the raw vibration signal is calculated by applying a Short-time Fourier transform(STFT).Several candidate Support vector machine(SVM) models are trained with different combinations of features in the feature pool with scarce labeled *** evaluating the pretrained SVM models on the validation set, the most discriminative features and best-performed SVM models can be selected, which are used to make predictions on the unlabeled *** predicted labels reserve the expert knowledge originally carried by the SVM *** are combined together with the scarce fine labeled samples to form an Augmented training set(ATS).Finally,a novel 2D deep Convolutional neural network(CNN)model is trained on the ATS to learn more discriminative features and a better *** results on two fault diagnosis datasets demonstrate the effectiveness of the proposed DFD.
同方期刊数据库