差分进化算法的性能分析及其动力学模型参数估计

The performance analysis of differential evolution algorithm and its application in parameter estimation of a dynamic model

差分进化算法以遗传算法为基础,在算法中引入扰动矢量,利用个体间的距离和方向等信息进行搜索,克服遗传算法容易“早熟”的缺陷。此外,差分进化算法收敛性及鲁棒性好,控制变量需要得不多,简单易用。分析差分进化算法的寻优性能,并将它应用于铯一铷一钒系低温硫酸催化剂中,SO2氧化反应的参数估计,证明利用差分进化算法得到的模型更加精确。

Differential evolution algorithm which is based on genetic algorithm is proved to be an efficient algorithm with strong ability in global optimizing. It introduced disturbing vector and does a good job in global searching through using the information of distance and directions between different individuals. In addition, this method converges faster and is robust. It requires few control variables and is easy to use. In the former part of this article, DE＇s global optimizing ability has been analyzed. Then, in the last part, DE has been applied to nonlinear parameter estimation of the model of low temperature SO2 Oxidation with Cs-Rb-V sulfuric asid catalyst to deciding the best parameters of this model. Results indicate that the model decided by DE is more accurate.