Application of fault diagnosis method based on HHT-WNN in HVAC system

摘要

A fault diagnosis method based on HHT-WNN is presented for the detection and diagnosis of sensors failure in HVAC system. The method in this paper is composed of four steps: (1) Fault Detection, by the means of threshold; (2) Fault Feature Extraction, decomposing the acquired signal into two IMF component and one residual component by EMD and then extract the instantaneous amplitude of them by HSA to get the feature of different failure; (3) Fault Type Identification, identifying the fault type by the trained WNN; (4) Fault Recovery, replacing the ith fault sensor data x_i with y_i to achieve fault recovery. The simulation of sensor fault (i.e., complete failure, drift failure, and decreased accuracy failure) shows that this method can not only detect and diagnose the sensor failure accurately but also resolve the sensor failure.