Adaptive Parameter Estimation Of Vehicle System Based On Estimation Parameter Error

In the control theory and application, it is important to improve the control performance of practical systems by designing an accurate model. However, it is not trival to obtain the accurate system model based on the first principle modeling methods. In order to design a model with high performance, it is found that the influence of model uncertainties can be demilished by estimating the unknown model parameters online. Consequently, how to accurately estimate the unknown system parameters of intricate system is a critical problem in the control field, which is also an essential issue to improve the performance of complex systems.Automotive vehicles are prevailing transport in the modern society, where the basic demand of vehicle systems is safety, comfortable driving, environmental and economic quality. Nowedays, it is essential to develop novel methods to eatimate the unknown parameters of vehicles, e.g., road gradient, vehicle mass. This is not only an important way to improve the performance of vehicle control systems, but also a promosing derection of vehicle industry.To address above mentioned issues, this thesis studies the adaptive online estimation of constant and time-varying parameters for a class of linearly parameterized systems. Moreover, this thesis presents a novel adaptive parameter estimation framework and some new adaptive laws for such linearly parameterized systems, which can guarantee the steady-state and transient error convergence performance (e.g. overshoot, convergence rate). The main work can be summarized(1) Adaptive estimation of linearly parameterized system with constant parameters. The vast majority of available parameter estimation methods mainly depend on a predictor or observer design. By introducing appropriate filter operations, an explicit expression of parameter estimation error is obtained and used to drive the adaptive laws. The proposed novel adaptive laws can achieve the improved estimation performance.(2) Adaptive estimation of linearly parameterized system with time-varying parameters. By dividing the time into small intervals, the time-varying parameters are approximated in terms of polynomials with unknown coefficients. Then a novel adaptive law design methodology is developed to estimate those constant coefficients within each interval, for which the parameter estimation error information is explicitly derived and used to drive the adaptations. Finite-time estimation convergence and the robustness against disturbances are all proved.(3) Adaptive parameter estimation with guaranteed prescribed performance. A prescribed performance function (PPF) and the associate transform are proposed, such that the parameter estimation can be reduced as a regulation problem of the transformed system by designing an adaptive law. To this end, a novel adaptive law based on the obtained parameter estimation error is developed, such that the error convergence can be guaranteed to be within the prescribed bound.On the other hand, this thesis also investigates the online estimation of vehicl mass and road gradient for automotive vehicles by using the aforementioend adaptations, and extensive comparative simulation results are provided based on a realistic vehicular system model constructed in CarSim(?) (version 8.1) and Simulink. The simulation results illustrate that the proposed methods can provide faster transient and better steady-state performance compared to RLS and gradient methods. Overall, the obtained estimation methods are potentially useful for improving the pefoemance of practical vehicle sytems.