An Automated Approach to Filter Design for Online State- and Parameter Estimation on Unknown, Non-analytic Models

Abstract

Abstract One frequently performed task based on topologically created models in the domain of control engineering is the design of a filter for state and parameter estimation. However, designing such a filter is an error-prone and difficult process even in the case of a well-known model. In fact, the control engineer using such a model may not know all relevant details (e.g. nonlinearities) that are obligatory in order to get appropriate estimations. Thus, this paper presents a solution to automatically design a filter for state and parameter estimation for unknown models based on sigma-point filters [2] , achieving a higher accuracy especially for models with major nonlinearities. This task is carried out by means of a graphical user interface (GUI) in Matlab. The GUI is designed as a step-by-step tool that guides the user through all necessary steps in order to receive appropriate parameters for the underlying filter algorithms. The algorithms are based on an Unscented Kalman Filter (UKF), which will be described and illuminated. Eventually, the whole process of designing a filter by the help of the GUI will be evaluated on an exemplary model.