Guaranteed parameter estimation for discrete-time chaotic systems

This paper considers the problem of parameter estimation from noisy measurements of discrete-time chaotic systems. The guaranteed approach assumes that the uncertainty is represented by intervals of possible values of the unknown variables (state, model parameter and measurement errors). The developed algorithm is based on interval analysis and can be used in the forward and backward time direction. The result of the guaranteed estimation is interval estimates that contain the true values of the unknown variables. The proposed algorithm can be usefully associated with common estimation methods developed in the field of optimization approach and estimation in real time. If the estimation problem is solved by the least squares method or its modifications, the guaranteed algorithm can be used to specify the set of possible values of the unknown variables. It decreases the number of local minima of the cost function. Computed interval estimates may also be used to verify the results obtained using the modifications of the Kalman filter for nonlinear systems. In the practical section, the dependence of the results on the number of available measurements and noise level is examined.