多样性分布参数的粒子群算法及其在过程动态优化中的应用

Particle Swarm Optimization for Diversity Distribution Parameters and Its Application in Dynamic Optimization Process

提出一种多样性分布参数的粒子群算法(DDPPSO)。在DDPPSO算法中,每个粒子在初始化时拥有各自的惯性权重和加速因子。在迭代时由每个粒子的寻优性能决定其参数的权重,进而计算参数群体的加权平均值。根据加权平均值与自适应方差,通过正态分布产生下一代参数个体,从而实现参数群体的多样性分布,为算法的寻优提供实时最佳的控制参数。标准测试函数实验表明,在寻优性能上DDPPSO算法较新改进的PSO算法有较大提高。最后,将DDPPSO算法应用于Park-Ramirez生物反应器的动态优化,获得满意的结果。

The particle swarm optimization for diversity distribution parameters （DDPPSO） was proposed. In DDPPSO, each particle has its own inertia weight and acceleration factor during initialization. According to the optimization performance of each particle, the weight value of its parameters can be determined to calculate the weighted average values of parameters. Basing on weighted average values and adaptive variance and the nor- mal distribution which generating the parameter of next generation, the diversity distribution of parameters can be achieved to provide real-time and optimal control parameters for the algorithm optimization. Experiment on normative test function shows that the DDPPSO outperforms other improved PSO algorithms in performance. Applying DDPPSO to dynamic optimization of Park-Ramirez bioreactor can bring about satisfactory results.