Development and Testing of Methods of Parameter Estimation of Dynamic Systems in Consideration of Restrictions

Abstract

For the identification of complicated systems on the basis of measured data, models will often be set up by means of the known estimation methods which do not describe the causal relation in the right way. This is due, for example, to a strong correlation of the input variables. Parameter estimation methods with restrictions representing one possibility for overcoming these difficulties are being developed and tested for weighting function and difference equation models. The restrictions for the parameters have been obtained from theoretical identification and are available as a priori information. The projects realized of the prediction of the investment costs of automation installations, of the development of a water quantity model of a river and the setting up of a temperature model of a greenhouse serve to prove the efficiency of the methods examined.