The problem of the identification of models defined by complex systems
is frequently encountered in chemical engineering. Due to convergence
difficulties and/or an excessive computational burden, simplified mod
els are often employed. Similarly approximate models are used in a pre
liminary phase for the estimation of the most significant parameters.
In these cases the residuals (i.e. the difference of the experimental
data to the values predicted by the approximate model) do not belong t
o a well-defined distribution function. Thus the usual regression meth
ods, such as those based on maximum likelihood, can sometimes lead to
seriously biased estimates. A new regression technique for avoiding ei
ther under- or overestimation due to compensation or cumulation of exp
erimental errors and model deviations is presented in this paper. Copy
right (C) 1996 Elsevier Science Ltd