Multi-model identification system properties near extremum: simulations and analysis

This article concerns the question about properties and behavior of the multi-model adaptive-searching identification methods near extremum point. The model of the relation between criterion and parameters is proposed. Numerous simulations of identification process was held under different parameters of this model and identification system. Advantages and drawbacks of the two extremum point estimation are studied. Recommendations and rough calculations to determine correct parameters of identification system itself is given.