Fault Detection Unknown Input Observer for Local Nonlinear Fuzzy Autonomous Ground Vehicles System Based on a Joint Peak-to-Peak Analysis and Zonotopic Analysis Threshold

This paper focuses on the fault detection problem of local nonlinear fuzzy autonomous ground vehicles systems with disturbances by developing set membership estimation. The intractable influence of disturbances and local nonlinear part on the design of fault detection thresholds is the principal difficulty in fault detection using the existing set membership estimation methods. To this end, a fault detection unknown input observer via a novel joint peak-to-peak analysis and zonotopic analysis threshold is proposed. Firstly, the disturbances are divided into decouplable and non-decouplable parts. An unknown input observer, designed with peak-to-peak performance criteria, is employed to achieve the decoupling of the decouplable disturbance part. Subsequently, based on the dynamic structure of the observer, the local non-linear part is represented as polyhedra through peak-to-peak analysis and zonotopic analysis, thereby providing an interval estimation for the system output. his method utilizes zonotopic analysis to mitigate the over-estimation issue in the initial stage inherent in traditional peak-to-peak performance-based interval estimation techniques. Furthermore, it retains the necessary degrees of freedom during the decoupling process of certain disturbances to optimize the observer's performance. Consequently, this approach significantly enhances the accuracy of the interval estimation as compared to existing methods. Moreover, the peak-to-peak performance of fault detection unknown input observer is analysed using the fuzzy basis-dependent Lyapunov function to decrease the conservatism. Lastly, simulations are analysed in order to validate the accuracy of the proposed thresholds.