Parameter Identification - The random initial estimation issue

Experimental offline identification algorithm is based on minimization of criteria function considering the unknown mathematical model variables. We often use software support of various computing programs to find the minimum of a scalar function of several variables. However, operator has to know initial estimation of the searched parameters to get proper results. If operator sets bad initial estimation, the identification process would find another local minimum or would diverge. This paper is focused on random approach to initial estimation of simple mathematical model using basic MATLAB software support.