An improved AEA using local search and its application to parameter estimation for homogeneous mercury oxidation

The Alopex-based evolutionary algorithm (AEA) possesses the basic characteristics of evolutionary algorithms as well as the advantages of the simulated annealing algorithm and gradient descent methods. The AEA is good at exploration but poor at exploitation, which results to some extent in poor convergence. To improve the performance of the basic AEA, we propose a local search algorithm in this paper to generate the subsequent population when the results for the best individual have not improved after several consecutive iterations. Furthermore, we modify the annealing temperature in the AEA. Numerical simulation results of 20 benchmark functions show that this improved AEA algorithm (LAEA) can generate solutions of higher quality. Furthermore, we used LAEA to estimate reaction kinetic parameters for homogeneous mercury oxidation, and the satisfactory result shows its suitability for practical applications.