Abstract:
A new approach to model synthesis of dynamic process with non-stationary perturbations based on application of maximum condition of generalized power function is considered. A new stochastic filter is developed using this model and procedures of invariant embedding. Within the example it demonstrates the possibility of achieving the best precision characteristics by reducing the amount of computational effort in comparison with the Kalman filter in the traditional model of nonstationary process. This confirmed by the results of mathematical modeling.