Improved Electromagnetism-like Mechanism Algorithms For Single-objective Global Optimization Problem

Global optimization is an important independent branch of modern optimization designs. It has been widely used in many fields such as science, engineering, real life and so on. In recent years, heuristic optimization algorithm has been studied and developed greatly due to its intelligence and wide applications.Unconstrained and constrained optimization problems of single objective global optimization are studied and two heuristic methods based on attraction repulsion mechanism in electromagnetism—improved Electromagnetism like Mechanism algorithms are proposed in this paper. The main contributions of this thesis can be summarized as follows:1. To solve unconstrained optimization problems according to Electromagnetism like Mechanism (EM), an improved EM algorithm is proposed. First, the lower bound of objective function is introduced to the formula of the particle charge in order to avoid its parameter sensitivity, then the formula of the total force vector is also improved in order to reduce the amount of computation and computer overflow, finally the step size is added to the mutation in order to avoid trapping into local solution. Based on these, an unconstrained EM optimization algorithm(UEM) is designed.2.An improved EM algorithm(CEM) is proposed for constrained optimization problems. Firstly, the constrained optimization problem is transformed into an unconstrained optimization problem by external point method. Secondly, the orthogonal design is adopted to produce uniformly distributed initial population. Then, the formula of the particle charge is re defined to decrease the amount of computation calculation and improve the efficiency. Finally, the CEM is proposed.3. The numerical simulations are made on two improved EM algorithms. UEM is executed on 10 standard test functions, and the results are compared with those of both EM algorithm and genetic algorithm by using the same parameters. The results indicate that UEM improved the precision of the obtained solution and the UEM algorithm is effective. CEM is executed on 6 standard test functions, and the results are compared with Simulated Annealing Algorithm. The results demonstrate that the proposed algorithm is competitive, efficient, robust and fast convergence.