摘要
针对反应动力学参数估计的复杂性,设计了一种新颖的优化方法—迭代全息搜索策略(IHRS)。全息搜索策略(HRS)将多维空间中的寻优问题直观地通过二维全息图上的近邻搜索来实现,是一种全局搜优效率较高的确定性优化方法,但它只能用于离散系统的优化。IHRS通过对连续变量进行离散化处理,并运用迭代计算逐步缩小离散系统与原连续系统的偏差,将复杂的多维连续变量优化问题转化为多个串联的较为简单的离散变量组合优化问题,再运用HRS寻优。并讨论了等分区间数、搜索域收缩率、群规模等参数对IHRS搜优效率的影响。六维Alpine函数测试表明,IHRS的全局优化性能优于单纯形法和遗传算法(GA)。将IHRS应用于SO2催化氧化反应动力学模型参数的估计,取得了满意的结果。
Holographic research strategy （HRS） is a determinate optimization method with high efficiency in searching for the global optimum, However, HRS can＇t be applied to optimize continuous variables, it can only be used in optimizing discrete systems, Therefore, it is necessary to improve HRS and to ensure the optimization algorithm can be applied in multidimensional continuous systems, A novel optimization algorithm named as iterative holographic research strategy （IHRS） was designed for these purposes. IHRS changes continuous variables into discrete variables in the searching region firstly, and then finds the optimum in the discrete system, In order to reduce the deviation between the continuous system and the discrete system, IHRS adopts iterative algorithm to shrink the searching region gradually according to the location of the current optimal value. The influence of several important parameters of IHRS, including the number of the region being partitioned averagely, the shrinking factor of the searching region and the individual amount of a cluster, on the optimization efficiency of IHRS was discussed. Six-dimensional Alpine function was applied to testing IHRS, the results demonstrate that its global optimization performance is superior to those of genetic algorithm （GA） and simplex method. Further, IHRS was applied to estimate the kinetic model parameters of SO2 catalytic oxidation, and satisfactory results were obtained.
出处
《高校化学工程学报》
EI
CAS
CSCD
北大核心
2007年第3期494-499,共6页
Journal of Chemical Engineering of Chinese Universities
基金
浙江省工业催化重中之重学科开放基金(200602)
关键词
全息搜索策略
迭代
动力学参数估计
二维全息图
遗传算法
holographic research strategy
iterative
kinetic parameters estimation
two-dimensional hologram
genetic algorithm
