The Improvement And Application Of Electromagnetism-like Mechanism Algorithm

The global optimization problem exists widely in many fields, such as engineering design, manufacturing, transportation and information processing, etc. It cannot be solved properly any more with the conventional optimization methods due to its characteristics of multiple local optimal solutions, non-easy acquisition of first-order or second-order information and so on. However, the heuristic optimization method has become the main way to solve the global optimization problem because of its obvious superiorities in versatility, intelligence, and so on.A new heuristic global optimization method which is the Electromagnetism-like Mechanism (EM) algorithm is studied in this paper. First, the principles and procedures of the EM algorithm are summarized in the paper. Second, in terms of the defects in the standard EM algorithm, such as large computation and low convergence rate, the paper improves the EM algorithm and proposes an improved Electromagnetism-like Mechanism algorithm which is an Normalization Electromagnetism-like Mechanism(NEM) algorithm. The new algorithm improves the initialization method of populations of particles, improves the fixed search neighborhood range of the local search to a variable one, improves the power formula by normalizing the objective function value, improves the resultant force formula by introducing a proper correction factor, and improves the moving step of particles. Finally, by introducing the expression methods of random keys and mutation operation on poor particles, the flow shop scheduling problem can be successfully solved with the NEM algorithm.Experimental results show that the NEM algorithm has great improvements in solution accuracy and velocity compared with the standard one, and it can have a better application to solving the problem of high dimensional function optimization. Besides, the application of the NEM algorithm to solving the flow shop scheduling problem is successful.The ability of the NEM algorithm proposed in the paper has been tested and verified, while it is still necessary to test the algorithm comprehensively. The further study on the algorithm should be enhanced in future work, and the emphasis should be focused on the solution of the constrained optimization problem and the discrete optimization problem.