一种混合粒子群算法的贝叶斯网络优化模型

传统的粒子群优化算法通过群体中粒子间的合作和竞争进行群体智能指导优化搜索,算法收敛速度快,但较易陷入局部较优值,进入早熟状态。为了解决这个问题,提出了一种混合粒子群算法的贝叶斯网络优化模型,它可以通过当前所选择的较优解群构造一个贝叶斯网络和联合概率分布模型,利用这个模型进行采样得到更优解,用其可随机替换掉PSO中的一些粒子或个体最优解;同时利用粒子群算法对当前选择出的较优解群进行深度搜索,并将得到的最优解融入到较优解群中。分析可知,该方法可以提高算法有效性和可靠性。

A Bayesian network optimization model mixed particle swarm optimization algorithm

Traditional particle swarm optimization algorithms search through cooperation and competition among the particles in the swarm. It has a fast convergence rate, however, easy to fall into local optimal. In order to solve this problem, a Bayesian network optimization model based on mixed particle swarm optimization algorithm is proposed. It can use current optimal solutions that may come from PSO to construct a Bayesian network and joint probability distribution. It will use this distribution samples and get some better solutions, which will be integrated into Particle Swarm Optimization（PSO） algorithm to increase diversity.