Fault Diagnosis of Direct Electro-pneumatic Brake Based on Model and Data-driven

According to dynamic features of strong coupling, non-linearity and uncertainty for direct electro-pneumatic brake, a fault diagnosis architecture based on model and data-driven is proposed to detect and isolate dynamic faults of solenoid valves and pressure sensors. Firstly, a sequentially structured fault diagnosis scheme is designed. Secondly, the state of the solenoid valves can be characterized after analyzing the drive current of solenoid valves and the fault diagnosis of solenoid valves is realized by using wavelet packet decomposition to extract the feature and the back propagation (BP) neural network to classify the fault type. Thirdly, the charging and discharging process of direct electro-pneumatic brake is described with analysis model, which can generate system residual. So the fault diagnosis for the sensor can be achieved. Finally, a test platform is designed and a series of experiments are conducted. Experiment results show that the fault diagnosis architecture and technologies designed in this paper can make fast and reliable diagnosis.