System Identification: Theory for the User, 2nd Edition

7. Parameter Estimation Methods

Suppose a set of candidate models has been selected, and it is parametrized as a model structure (see Sections 4.5 and 5.7), using a parameter vector θ. The search for the best model within the set then becomes a problem of determining or estimating θ. There are many different ways of organizing such a search and also different views on what one should search for. In the present chapter we shall concentrate on the latter aspect: what should be meant by a “good model”? Computational issues (i.e., how to organize the actual search) will be dealt with in Chapters 10 and 11. The evaluation of the properties of the models that result under various conditions and using different methods is carried out in Chapters 8 ...