文献类型：期刊文献

中文题名：优选优生进化算法及4-CBA软测量模型参数估计

英文题名：An Evolution Algorithm with Select-Best and Prepotency Operator and Parameter Estimation of 4-CBA Model

作者：颜学峰[1];余娟[1];钱锋[1]

机构：[1]华东理工大学自动化研究所,上海200237

年份：2005

卷号：19

期号：2

起止页码：238

中文期刊名：高校化学工程学报

外文期刊名：Journal of Chemical Engineering of Chinese Universities

收录：CSTPCD;;Scopus;北大核心:【北大核心2004】；CSCD:【CSCD2011_2012】；

基金：国家973计划(2002CB312200);国家自然科学基金(60074027);国家863计划(2001AA413130);国家863计划(2002AA412110);十五国家科技攻关项目(2001BA201A04)

语种：中文

中文关键词：进化算法;遗传算法;对苯二甲酸;对羧基苯甲醛;参数估计

外文关键词：evolution algorithm; genetic algorithm; terephthalic acid; 4-carboxybenzaldehyde; estimate parameter

摘要：提出一种优选优生进化算法(select-best and prepotency evolution algorithm,SPEA),该算法中优选优生操作首先使群体中的所有个体都以相同的概率在自身所处环境的某个领域内选择最优个体进行交叉操作,不但保持后代个体的多样性,避免早熟,而且使群体中较优个体多次参加交叉操作,保留优良信息,同时也避免了遗传算法(genetic algorithm,GA)中占用大量计算时间的种群挑选操作;然后挑选交叉操作产生的较优个体参与变异形成下一代个体,从而增强算法的局部搜索能力,使求得的全局最优解有较高的精度。与GA相比较,SPEA计算复杂性低,离线性能和在线性能都有较大的改进,局部搜索能力和全局寻优能力都有较大的提高。举例将SPEA应用于对苯二甲酸(terephthalic acid,TA)中对羧基苯甲醛(4-carboxybenzaldehyde,4-CBA)含量软测量模型的参数估计,获得了满意的结果。
A novel evolution algorithm with select-best and prepotency operator, which was named as select-best and prepotency evolution algorithm (SPEA), was proposed. The select-best and prepotency genetic operator was defined as follows. Every individual of the population has the same chance to select the best individual within the some range around itself and crossover with the selected individual to produce the new individuals, then the best one of the new individuals is selected as the individual of next generation. Using the genetic operator, a diversity of individuals of subgeneration are produced to avoid premature, the excellent individuals will be selected time after time, and the reproduction operator of genetic algorithm, which consumes much calculating time of CPU, is avoided. What’s more, SPEA possesses an advantage of gradient descent optimization to enhance the accuracy of obtained optimum solution. In order to compare the performances of SPEA with those of the traditional GA (TGA), SPEA and TGA were applied to search the global optimum solution of a same analytical function. The results demonstrated that SPEA used less CPU time than TGA, the SPEA’s on-line and off-line performances and local search ability are all superior to those of TGA, and that the probability of finding the global optimal solution is larger than that of TGA. Further, SPEA was applied to estimate the model parameters of 4-carboxybenzaldehyde (4-CBA) content in the product of terephthalic acid. Satisfactory results are obtained.