基于概率选择学习对象的粒子群优化算法

针对标准微粒群算法容易陷入局部极小的缺陷，对标准粒子群速度进化公式进行改进，提出一种基于概率选择学习对象的粒子群算法。找出比当前个体好的粒子，形成候选学习对象集，计算候选集中每个粒子被选中的概率，形成学习对象集，并加权利用学习对象集信息。该算法使得每个粒子可以充分利用整个种群的信息，有效地保证粒子群的多样性。对3个Benchmark测试函数进行了仿真，结果显示，该算法能有效地改善寻优性能，具有摆脱局部极值的能力。

Particle Swarm Optimization Based on Select Learning Object by Probability

A new particle swarm optimization based on select learning object probability is presented to improve the limited capability in escaping the local optima through modifying the velocity evolving formula. It uses particle’s fitness to choose better particle than the presented particle among swarm, and forms a candidate learning object set. Then, compute the selected probability of each particle among the candidate set, and form the learning object set, and use those information by weighting. In the presented algorithm, each particle can use the whole swarm information effectively and keep the diversity effectively. There Benchmark test function is selected. Experimental results demonstrate that the algorithm can improve optimizing performance effectively, and it can avoid getting struck at local optima effectively.