Performance of Inductive Method with Unaccounted Interference (case study: polynomial, harmonic, noise). Mikhail Alexandrov, Natalia Ponomareva

Abstract. The noise immunity of the inductive method of model self-organization (IMMSO) has been studied in detail in many publications. The method demonstrates amazing possibilities in case of the correct choice of model class and presence of 'white' noise. But the IMMSO developers give a pessimistic prognosis in the case of using any non-random unaccounted interference. Nevertheless just this case is interested for new IMMSO applications where a priori information about model class is unknown (or almost unknown). In the paper we experimentally elaborate the performance of IMMSO on a simple example of polynomial model class in the presence of unaccounted harmonic component and white noise. Our aim is to obtain qualitative results and explain them using frequency presentations.

Keywords. Inductive modeling, noise immunity, IWIM 2009.

References.

1. Ivakhnenko, A.: Inductive Method of Model Self-Organization of Complex Systems. Kiev, Naukova Dumka,.1982 (rus.)

2. Ivakhnenko, A., Stepashko, V.: Noise Immunity of Modeling. Kiev, Naukova Dumka, 1985 (rus.)

3. Ivakhnenko, A., Savchenko, E.: Investigation of Efficiency of Additional Determination Method of the Model Choice in the Modelling Problems with Application of GMDH Algorithms. Journ. of Inform. and Automat. Sciences, Begell House Inc, 2008, Vol. 40, N_3, pp. 4-22

4. Stepashko, V.: Method of Critical Variances as an Analytical Tool of the Inductive Modeling Theory. Journ. of Inform. and Automat. Sciences, Begell House Inc, 2008, Vol. 40, N_3, pp. 47-58