Comparison of identification algorithms based on the combining maximum principle and the regularization

Abstract

The variational methods of parametric identification which use the physical features of the studied systems in the form of the Hamilton — Ostrogradskii variational principle are studied. On this basis the identification algorithms resistant to measurement noise and having high convergence rate of their estimates to the actual values are produced. This is confirmed by comparing the results of mathematical simulation of the developed algorithms with the Kalman filter.