Self-validating gas sensor fault diagnosis method based on EEMD sample entropy and SRC

Aiming at the fault diagnosis problem of self-validating gas sensor under the condition of non-linear and non-stationary, a sensor on-line fault diagnosis method is proposed to conduct the feature extraction and intelligent identification for the sensor signals in different fault modes. Firstly, the sensor output signal is adaptively decomposed to a series of intrinsic mode functions (IMFs) by ensemble empirical mode decomposition (EEMD) according to sensor signal change, and each of IMFs and residue is conducted by the sample entropy analysis to extract the complete features of the sensor output signal. Afterwards, the over complete dictionary is comprised of the feature vectors of training samples in different fault conditions by using sparse representation-based classification (SRC). The SRC classifier is updated subsequently to improve the adaptivity of it for fault diagnosis. The minimum l1-norm constraint problem is applied to obtain the sparse represent coefficient of testing sample and sensor fault type identification is determined by reconstruction error minimum between test sample and its reconstructed signal in different fault conditions. The experimental results show that the proposed method can significantly extract more features of the sensor fault signal compared with the other fault diagnosis methods and the fault diagnosis recognition rate increases over 4% and reaches 97.14%. © 2016, Chinese Institute of Electronics. All right reserved.