Review and enhancement of cautious parameter estimation for model based control: a specific realization of regularization

This paper reviews a method of cautious parameter estimation which was proposed initially for estimating models of industrial processes for the purpose of model based control. Essentially, the method is able to influence the resulting estimation algorithm by introducing 'fictitious' measurements which correspond to a priori knowledge of the values of specific model parameters. The paper shows that the method is a particular realization of regularizations. Enhancements to the original method are also proposed.