On Optimality Criteria in Identification Problems
The identification problem is largely formulated as a problem of finding a plant operator estimate optimal in terms of a specified criterion. The resultant solution permits estimation of the model adequacy. The criterion and the model structure are selected individually for each model. For complex systems the large number of plant parameter interrelations and of external disturbances make a completely adequate model non-feasible or undesirable. True, a more accurate model and a better system description is supposed to give a better forecast and control of the system but studying each input and its responses takes more time, funds and material resources. Hence it is clear that identification of large systems requires quantitative assessment of the effect of each system input on its output variable and of the accuracy with which a model where these inputs are integrated simulates the actual processes; a decision should be made, which system variables should be represented in the model. Identification is also impossible without systems approach, studying the hierarchical structure and knowledge of the entire system functioning.
Unable to display preview. Download preview PDF.