Research on the application of neural network in diaphragm icing sensor fault diagnosis

Abstract

As the core component of the Icing Detection System of aircrafts, the reliability of Diaphragm Icing Sensor is a key factor for the ice detection system to work normally. This paper makes use of Neural Network and Autoregressive Exogeneous Model (ARX) to set up the output prediction model of the diaphragm icing sensor. Compare the predicted output of the model with the actual output to diagnose sensor faults of the sensor. According to the data acquiring from our experiment platform of Diaphragm Icing Sensor, it has been proved that this method is effective for fault diagnosis of the Diaphragm Icing Sensor. © 2009 Springer Berlin Heidelberg.