Pattern search particle swarm optimizer for global optimization of multimodal functions

In order to solve the defects in the heuristic algorithm, the paper proposes an improved intelligent algorithm based on pattern search and particle swarm optimizer. The algorithm adopts the improved exploratory move and pattern move to strengthen the optimization direction and speed up the local convergence. Meanwhile, the particle rebirth strategy is applied to replace the too close particles, which can strengthen the optimization variables diversity and effectively avoid falling into the local optimum. In addition, this algorithm repeats initializing searching space to protect the optimization algorithm from the premature convergence and optimization stagnancy. At the experiments of solving complex multi-modal composition benchmark functions, contrast with the heuristic algorithms, the stability and accuracy of the new algorithm are verified.