A hybrid genetic algorithm for the fitting of models to electrochemical impedance data

Nonlinear regression algorithms have been extensively applied in the parametric analyses of electrochemical impedance spectroscopy (EIS). However, the problem of how to provide good initial estimates of parameters automatically has not been thoroughly solved yet. For this reason, this paper offers a hybrid algorithm, which consists of the genetic algorithm (GA) based on probability theory and the deterministic computation based on analytic theory. The paper improves the usual GAs and puts forward the concept of weighting mutation. The hybrid algorithm overcomes the shortcoming that the usual GAs are liable to produce premature convergence. This paper has preliminarily solved the problem of how to produce good initial estimates of parameters automatically by a program design in the data processing of EIS. In principle, the design idea of the hybrid algorithm is of universal validity and can be used for solving the problem of how to provide good initial estimates of parameters automatically, which appears in many kinds of nonlinear regressions.