Bond graph model for prognosis and health management of mechatronic systems based on energy activity

Timely and accurate, detection and prediction, of fault is beneficial for every safety. However, it is a necessity for modern autonomous systems like autonomous vehicles and robots due to the absence of continuous human supervision. This paper attempts to establish energy as a viable parameter for fault identification. For this Element Activity Index is used as a metric for identification, calculated in this paper using only the sensor data. The proposed technique is then implemented on a bicycle vehicle dynamic model. Abrupt faults of different intensities are introduces in different elements of the model and their proper detection and isolation is checked using the proposed technique. The visibly different residual trends allows us for accurately detect the presence and also the location of the fault.