Intelligent Fault Diagnosis of Air Brake System in Heavy Commercial Road Vehicles

Fault detection and isolation in brake system is crucial for the safe operation of autonomous vehicles. Failure to detect impending faults may result in component degradation, which leads to vehicle breakdown. In recent years, Machine Learning techniques have been widely used for fault diagnosis in vehicles and hence provide intelligence in fault prediction. This paper proposes a general fault detection and isolation method for air brake system in Heavy Commercial Road Vehicle (HCRV) using machine learning techniques. Decision tree and random forest methods have been used to learn fault patterns that are reflected in the wheel speed sensor data. The training and testing data for the diagnostic scheme were collected from the Hardware-in-Loop (HiL) set up of air brake system. To classify the fault/no-fault conditions of air brake system, a random forest approach gave good prediction accuracy of 94.47 %.