Identification and State Estimation for Linear Parameter Varying Systems with Application to Battery Management System Design

In this dissertation, the identification and state estimation for linear parameter varying (LPV) systems as well as their applications to battery management system design are investigated. First, two complete LPV system identification procedures are described. One procedure uses a layered optimization process while the other uses a subspace based method. Both methods provide theoretically sound and practical processes under which realistic LPV system identification problems can be solved. Secondly, controller and observer design techniques for LPV systems are examined. In particular, the stability conditions in the form of parameter dependent linear matrix inequalities (LMI) that result from the applications of the standard Lyapunov stability theory and other advanced techniques such as L2 and Hinfinity to LPV systems are discussed in detail. Also discussed are the some of the techniques for solving these LMIs. Lastly, because real systems often contain parametric uncertainty, the use of input to state stability to characterize closed loop performance of controllers or state estimator under such condition is also reviewed.;The tools developed for LPV systems are then applied to solve the problems of model identification and state of charge (SoC) estimation for battery cells. The model identification problem is tackled using both identification schemes so that differences in performance and effectiveness between the methods can be compared and contrasted. A SoC estimator based on LPV system state estimation techniques is then designed using the model identified. Because parametric uncertainty is inherent in the estimator designed, the stability and performance of the estimator is analyzed using the notion of input to state stability. Experimental data is then used to illustrate the efficacy of this method. The goal of these applications is to show the relevance of the LPV structure and techniques to problems in battery management system design, so that research will be done to solve other problems in this area under the same framework.