Model-Based Estimation

4.0 credits · Graduate

Course covers a variety of ways in which models and experimental data can be used to estimate model quantities that are not directly measured. Covers methods for solving the class of inverse problems that take the following form: given partial information about a system, what is the behavior of the whole system? Main estimation methods presented are batch least-squares-type estimation for general problems and Kalman filtering for dynamic system problems. Course deals with the issue of observability, which amounts to a consideration of whether a given inverse problem has a unique solution, and briefly covers the concept of statistical hypothesis testing. Techniques for linear and nonlinear models are taught. Both theory and application are presented.