Hierarchical Differential Evolution for Parameter Estimation in Chemical Kinetics

Parameter estimation, a key step in establishing the kinetic models, can be considered as a numerical optimization problem. Many optimization techniques including evolutionary algorithms have been applied to it, yet their efficiency needs further improvement. This paper proposes a hierarchical differential evolution (HDE) in which individuals are organized in a hierarchy and mutation base is selected based on the hierarchical structure. Additionally, the scaling factor of HDE is adjusted according to both the hierarchy and the search process, elaborately balancing the exploration and exploitation. To demonstrate the performance of HDE, experiments are carried out on kinetic models of two chemical reactions: pyrolysis and dehydrogenation of benzene as well as supercritical water oxidation. The results show that the proposed algorithm is an efficient and robust technique for kinetic parameter estimation.