Set-Membership-Based State Estimation and Fault Detection for Lane-Keeping System via Linear Parameter-Varying Model

This article presents a method for state interval estimation and fault detection in lane-keeping systems for intelligent connected vehicles, utilizing a linear parameter-varying model and set-membership estimation approach. First, the inherently nonlinear vehicle dynamics model is transformed into a linear parameter-varying model. By retaining local nonlinear terms, the number of subsystems is reduced, thus decreasing the complexity of the method. The design flexibility of the observer is improved through specific matrix transformations. Using the $\cal {L}_{\infty } $ technique, the conservatism of nonlinear term constraints is reduced while maintaining performance. The observer gain matrix solution is reformulated as a linear matrix inequality (LMI) problem without increasing computational complexity. Based on the interval estimation results, a fault detection logic is proposed that eliminates the need for manually set thresholds. Simulation results confirm the effectiveness and feasibility of the proposed method.