粒子群算法在多峰值函数优化中的应用

Brief analysis on PSO and its operation in matlab

粒子群算法是一种基于群体的智能算法,具有较强的全局搜索能力,并能通过对一定数量粒子的迭代运算获得问题的全局最优解。将粒子群算法应用于多峰值函数优化中可以避免常规方法难以同时搜索出多个极值而陷于局部极值的问题。基于matlab平台的仿真实验中,引入粒子群初始化位置拥挤距离检测,并在peaks函数上进行测试,可以有效实现全局和局部搜索,并能较好地保持粒子的多样性,从而获得多峰值函数的最优解。

PSO （Particle Swarm Optimization） is an intelligence algorithm based on swam. It is characteristic of strong global searching ability and able to reach for globally optimal solution through iterative operations of a certain quantity of particles. The application of PSO in multiple hump function can avoid problems caused by conventional methods which are able to search no multiple extrema simultaneously but relative extrema. In simulation experiments based on matlab, it is appropriate to take in crowding distance detection of particle swam＂ initial position and peaks function testing, which are able to achieve global and local searching, maintain the diversity of particles, and there- fore to realize the optimal solution of multiple hump function.