Adaptive automotive speed control

Abstract

Modern automotive speed control systems are designed to provide smooth throttle movement, zero steady state speed error, good speed tracking over varying road slopes, and robustness to system variations and operating conditions. Additionally, there is a need to minimize the number of controller calibrations for different vehicle applications. All of the above objectives cannot be simultaneously met by conventional fixed gain controllers which need different calibrations for different vehicle lines. With such requirements, an adaptive controller offers benefits over a conventional controller provided its complexity does not significantly exceed that of a conventional controller.

To limit the controller complexity, the adaptive design in this study is based on sensitivity analysis and slow adaptation using gradient methods. This design method allows the use of our a priori knowledge about the plant model in order to determine a stability region for a reduced order adaptive controller, in this case a simple PI controller. The adaptive algorithm, driven by the vehicle response to road load torque disturbances, tunes a PI controller to continuously minimize a single performance based cost functional for each different vehicle over varying road terrain. This results in performance not possible with a fixed gain controller. The adaptive controller has been tested on a number of vehicles with excellent results, some of which are presented here.