【摘要】 针对GA遗传算法种群多样性差、局部寻优能力差等问题,提出了多种群遗传算法(MGA)。该算法利用间断平衡理论,构建多种群、多交叉算子操作方式并结合局部搜索方法和种群动态调整策略,提高算法的局部寻优能力和寻优速度。通过与GA和ISGA算法相比,MGA运行时间短,搜索性能强。利用MGA优化MKLSSVM参数,建立基于MGA-MKLSSVM的水泥篦冷机二次风温预测模型。结果表明,此模型辨识精度高、泛化能力强。更多还原

【Abstract】 Aiming at the problem of poor diversity and poor local optimization ability of the genetic algorithm,a multipopulation genetic algorithm( MGA) was proposed. According to the theory of Punctuated Equilibria,the algorithm took the manipulation of multiple populations,and multiple crossover operators. Meanwhile,the MGA algorithm had also composed of the local search method and population dynamics adjustment strategy to improve the local search ability and speed. Compared with GA and ISGA algorithms,MGA running time is short and has a better optimization performance.Finally,the MGA algorithm was applied to optimize the multi-kernal least square vector mechaine( MKLSSVM)parameters. And then established the secondary air temperature model of the cement grate cooler based on MGAMKLSSVM. The results show that this model has high recognition precision and strong generalization ability.更多还原