JOINT ESTIMATION OF THE STATE AND PARAMETERS OF DYNAMIC CONTROL OBJECTS BASED ON THE MAINE ESTIMATOR

The issues of constructing an adaptive joint estimation of the state and parameters of dynamic control objects using the Maine estimator are considered. There are various variants of the extended filter, and a variant based on iterations between parameter and state estimates was used in the work. In this version of the extended Kalman filter, the problem of joint parameter and state estimation is solved in such a way that parameter estimation is performed before state estimation. Then, the parameter values are used to assess the state. In this case, further iterations between the state vector estimation and the parameter vector estimation are possible. Estimation algorithms are given when using system and measurement models with constant coefficients and non-stationary noises, the correlation matrices of which are unknown but are functions of time. A part of the measurement sample is used to solve the filtering problem. Based on the system and measurement models, recurrence relations are determined for extrapolation and filtering algorithms. Expressions for identifying the covariance matrices of the object noise and measurement interference are given. The given recurrence relations can be the basis for obtaining algorithms for joint estimation of the state and parameters of dynamic control objects.