基于边界变异的一种新的粒子群优化算法

New Particle Swarm Optimization Base On Boundary Mutation

针对传统粒子群优化算法易早熟,收敛精度低,特别是在解决大维数问题时,效果很不理想等缺点。针对这类问题,首先提出一个判别机制,判定算法什么时候达到早熟,若达到早熟则提出一种基于边界与随机变异的方法使部分粒子进行变异,从而使粒子重新分散后,再进行搜索。通过对四个经典测试函数的数值仿真实验证明,该方法能极大地提高算法的寻优能力,特别是在高维函数寻优时获得了较好的优化效果。

The traditional particle swarm optimization algorithm was premature convergence, low accuracy.Especially in solving large dimension problems, the effect is not ideal.In order to solve these problems,first it presents adecision mechanism when the algorithm reaches puberty.Ifthe algorithm reaches puberty,present a method based on the boundary and random variation.It makes part of particle variation and then these particles are variationso that the particles bounce early regional and search again 遥 Based on the 4 classical test functions of numerical simulation experiment, the method can greatly improve the searching capability especially in highdimensional function optimization it obtained better optimization effect.