Nonlinear and filter based estimation for vehicle suspension control

The paper analyzes the performance of a new estimation method for vehicle suspensions, which incorporates three parallel Kalman filters and takes into account the nonlinear damper characteristic of the suspension. For the performance evaluation, an Extended Kalman filter (nonlinear estimator) is utilized as a benchmark. The estimator structures are tuned by means of a multiobjective genetic optimization algorithm in order to maximize their performance. The advantages of the parallel Kalman filter concept are its low computational effort and good estimation accuracy despite the presence of nonlinearities in the suspension setup. Both estimators are compared to a computationally simple concept that gains the estimates directly from measurement signals by conventional filtering techniques. The performance of the estimators is analyzed in simulations and experiments using a quarter-vehicle test rig and excitation signals gained from measurements of real road profiles.