Joint state-parameter estimation for active vehicle suspensions: A Takagi-Sugeno Kalman filtering approach

In the paper, we present a novel nonlinear approach of combined on-line state and parameter estimation for controlled vehicle suspensions. With respect to vehicle dynamics, the vehicle body mass is a parameter that is crucial for the performance of state observers. Simultaneously, its value can significantly vary during operation, e. g. due to additional load. Hence, a joint estimation approach is adopted by augmenting the state vector with the unknown body mass. Based on a Takagi-Sugeno (TS) representation of the augmented nonlinear suspension model, the overall nonlinear observer is constructed by employing the Kalman filter theory for each linear subsystem. Stability of the error dynamics of the global observer is then enforced by means of linear matrix inequalities (LMI). In simulations and experiments on a hybrid quarter-vehicle test rig using stochastic disturbance inputs, the joint estimation approach is shown to maintain high estimation accuracy, despite the uncertain body mass parameter.