Sensor Fault Diagnosis Using Ensemble Empirical Mode Decomposition and Extreme Learning Machine

An algorithm using Ensemble Empirical Mode Decomposition (EEMD) and Extreme Learning Machine (ELM) for the detection and classification of sensor fault is presented in this paper. Under this method, the standardized sensor signal is decomposed through EEMD into the original signal, several Intrinsic Mode Functions (IMFs), and residual signal. Then, the variance, reduction ratio and normalized total energy of each IMF and residual are calculated as the sensor fault features. Subsequently, the feature vectors are input into the Extreme Learning Machine (ELM), which is utilized as the classifier for the detection and identification of sensor faults. The fault diagnosis simulation result of the carbon dioxide sensor indicates that this method can not only be effectively applied to the fault diagnosis of carbon dioxide sensors but also provide reference for the fault diagnosis of other sensors.