Conditioned Inverse Impact of the Initial Data Uncertainty in Parametric Identification Problems
Olga Kantor · 2021 3rd International Conference on Control Systems, Mathematical Modeling, Automation and Energy Efficiency (SUMMA) · 2021
The paper deals with the analysis and assessment of the influence of the initial data uncertainty on the accuracy in parametric identification problems. An approach to assessing the influence of the permissible variation of absolute accuracy in the parametric identification problem on the expected changes in the initial data is presented, based on the study of the resulting inverse effect of the initial data uncertainty. With regard to the problem of constructing an extended Cobb-Douglas production function, a practical testing of the presented method for solving parametric identification problems under conditions of the initial data uncertainty is given. The method develops the ideas of L.V. Kantorovich on the definition of exact bilateral boundaries for the desired parameters and allows you to analyze the adequacy of the resulting models.