查看文章

The recent drives to increase efficiency in vehicle systems has led to an increased interest in developing vehicle health management systems. The use of artificial intelligence and machine learning algorithms would be vital for these applications to identify trends in vehicle performance and make inferences of the current and future state of health of safety-critical subsystems. This paper presents a study done using the outputs from diagnostic and prognostic models based on data gathered by Health and Usage Monitoring System sensors on-board Armoured Personnel Carriers. Based on the requirements and the data being processed for insights, the outputs from these models are subject to different reasoning techniques, inference tools and algorithms. This includes a sensor signal validation and anomaly detection tool in which a trained probabilistic neural network is used to identify off-nominal behaviour in sensor data, thus aiding in the health assessments and integrity checks of sensors. Additionally, Kalman filtering is employed to utilize the dynamic equations that govern the operation of the powertrain. An Extended Kalman Filter (EKF) algorithm is developed to determine instances where there are large discrepancies between the measured and estimated value, indicating a possible fault.