基于双种群交叉学习的粒子群优化算法

粒子群优化算法因其支配参数少、收敛速度快、易于实现等特点被广泛应用，但是粒子群优化算法存在精度低、容易陷入局部优化的问题。为此提出一种基于双种群交叉学习的粒子群优化算法。在该算法中，整个种群被分为普通子种群和精英子种群。普通子种群采用综合变异机制，该机制通过设置概率参数使普通子种群随机选择朝着优秀粒子的方向或者保持自身方向进行变异，以侧重寻找可能解区域。精英子种群则采用交叉学习机制，将粒子的历史最优和全局最优个体进行交叉生成范例，从而引导粒子对可能解区域进行局部搜索，还提出了一种非线性惯性权重来平衡粒子的全局勘探和局部开发能力。为了验证算法的有效性，在十六个基准问题上进行测试并与其他七种粒子群优化算法变体比较，实验结果表明该算法在求解精度和收敛速度总体排名第一，验证了该算法求解性能优于其他粒子群优化算法变体。

Particle swarm optimization algorithm with dual population cross-learning

Particle swarm optimization(PSO) algorithm is widely used because of its few dominant parameters, fast convergence speed and easy implementation. However, the algorithm has low precision and is easy to fall into the problem of local optimization. This paper proposed a particle swarm optimization algorithm with dual population cross-learning(DPCPSO). In this algorithm, it divided the whole population into ordinary sub-population and elite sub-population. The ordinary sub-population adopted a comprehensive mutation mechanism. This mechanism made the ordinary sub-population randomly choose the direction of the excellent particles or maintain its own direction to mutate by setting the probability parameter, so as to focus on finding the possible solution area. The elite sub-population adopted the cross-learning mechanism to cross-generate the historical optimal and global optimal individuals of the particles, so as to guide the particles to locally search the possible solution area. To balance the global exploration and local exploitation capabilities of particles, this paper proposed a nonlinear inertia weight. To verify the effectiveness of the algorithm, the proposed algorithm was tested on 16 benchmark problems and compared with other seven variants of particle swarm optimization algorithm. The experimental results show that the proposed algorithm ranks the first in solving accuracy and convergence speed, and it verifies that the algorithm performance is better than other variants of particle swarm optimization algorithm in solving.