Improvements in bounded error parameter estimation using distribution theory
Abstract
A bounded error estimation procedure based on integro-differential polynomials linking the inputs, the outputs and the parameters of the model is presented in this paper. These polynomials are obtained from differential algebra tools given input-output polynomials. The use of the distribution theory permits to obtain new relations in which the order of derivatives of the model outputs are smaller. This method is applied on the water tank example and the results are compared with the classical method based on the simple use of input-output polynomials. As it will be seen, this method significantly improves the parameter estimation results.
Origin : Files produced by the author(s)
Loading...