Enhanced Method for Fault Detection and Diagnosis of Vehicle Sensors using Parity Equations

For driver assistant systems and drive-by-wire architectures fault detection and diagnosis are essential parts. Fault detection using parity equations is a well known approach which can be implemented in a straightforward way. Especially for fault diagnosis of vehicle sensors good isolating patterns for the interpretation of the residuals are available. However, in critical driving situations false alarms can occur, which may compromise the efficiency of safety relevant stability systems. In this paper a method is presented which reliably detects critical driving situations utilizing the estimated nominal cornering stiffness. The instantaneous cornering stiffness is estimated using the sideslip angle obtained by an observer. Using this quantity the nominal cornering stiffness can be estimated in order to discern the linear and nonlinear region of the tire model. In the nonlinear region false alarms are likely to occur and simple fault detection using parity equations cannot be used. Utilizing this approach, false alarms of the fault detection for the sensors of lateral acceleration, yaw rate, and steering angle can be avoided. The proposed fault detection and diagnosis concept clearly indicates the validity of the detection, and the performance is demonstrated with measurement data.