论文部分内容阅读
惯性权重是微粒群算法(PSO)的重要参数,它可以平衡算法的全局和局部搜索能力的关系,改善算法的性能.对此,提出一种基于强化学习的适应性微粒群算法(RPSO).首先将不同惯性权重调整策略视为粒子的行动集合;然后通过计算Q函数值,考察粒子多步进化的效果;进而选择粒子最优进化策略,动态调整惯性权重,以增强算法寻找全局最优的能力.对几种经典函数的测试结果表明,RPSO能够获得良好的性能,特别是对多峰函数效果更加明显.
Inertial weight is an important parameter of Particle Swarm Optimization (PSO), which can balance the relationship between the global and local search ability of the algorithm and improve the performance of the algorithm.In this paper, an Adaptive Particle Swarm Optimization (RPSO) based on reinforcement learning is proposed. Firstly, the different inertia weight adjustment strategy is considered as the particle’s action set. Then, the Q function value is used to investigate the effect of multi-step evolution of particle. Then the particle optimal evolution strategy is selected to dynamically adjust the inertia weight to enhance the algorithm to find the global optimum The test results of several classical functions show that RPSO can obtain good performance, especially for multi-peak function.