Enhanced Fault Detection of Vehicle Lateral Dynamics Using a Dynamically Adjustable Bayesian Network Structure and Extented Kalman Filter

Abstract

In order to improve reliability and safety in many technical systems; the importance of the fault diagnosis and detection is increasing day by day. This is particularly important in safety-related industries such as airplanes, trains, automobiles, power plants, and chemical plants. In this study, a model-based fault detection method that can detect faults affecting the lateral dynamic system of a vehicle is proposed. The developed fault detection algorithm includes both Bayesian Network and Extended Kalman Filter (EKF). EKF plays an effective role especially in detecting the faulty speed measurements. In the fault detection algorithm, six residual values are calculated. The threshold values of all the calculated residuals are determined using real test dataset. Depending on whether the residuals exceed the threshold value or not, the fault generation coefficients in the Bayesian Network are also dynamically updated to provide precise information regarding which sensor has a fault. The implementation of the fault detection algorithm is carried out using real test data and the numerical simulations are performed in MATLAB/Simulink environment. The results show that the proposed fault detection algorithm gives over 92% fault probability by using Bayesian and EKF structure together.