| 一种自适应多策略差分进化算法及其应用 |
| |
| 引用本文： | 徐斌,陶莉莉,程武山.一种自适应多策略差分进化算法及其应用[J].化工学报,2016,67(12):5190-5198. |
| |
| 作者姓名： | 徐斌 陶莉莉 程武山 |
| |
| 作者单位： | 1.上海工程技术大学机械工程学院, 上海 201620;2.上海第二工业大学工学部, 上海 201209 |
| |
| 基金项目： | 上海高校青年教师培养资助计划项目（ZZgcd14002）；上海市科委地方高校能力建设项目（14110501200）。 |
| |
| 摘 要： | 针对差分进化算法由于固定参数设置而易早熟或陷入局部最优的问题，提出了一种自适应多策略差分进化算法（SMDE）。该方法以基本差分进化为框架，首先引入一个变异策略候选集合，一个缩放因子候选集合和一个交叉参数候选集合，然后在搜索过程中，以过去的搜索信息为基础，自适应地为下一时刻进化群体中的每个个体从候选集合中选择一组合适的变异策略和控制参数，以便在不同的进化时刻设置合适的变异策略和控制参数。对10个常用的标准测试函数进行优化计算，并与其他算法的结果进行了比较，实验结果表明，SMDE具有较好的搜索精度和更快的收敛速度。将SMDE用于化工过程动态系统不确定参数估计问题，实验结果表明该算法能较好地处理实际工程优化问题。
|
| 关 键 词： | 差分进化算法 自适应 多策略 动态系统 参数估计 |
| 收稿时间： | 2016-09-09 |
| 修稿时间： | 2016-09-16 |
A self-adaptive differential evolution algorithm with multiple strategies and its application |
| |
| XU Bin,TAO Lili,CHENG Wushan.A self-adaptive differential evolution algorithm with multiple strategies and its application[J].Journal of Chemical Industry and Engineering(China),2016,67(12):5190-5198. |
| |
| Authors: | XU Bin TAO Lili CHENG Wushan |
| |
| Affiliation: | 1.School of Mechanical Engineering, Shanghai University of Engineering Science, Shanghai 201620, China;2.College of Engineering, Shanghai Second Polytechnic University, Shanghai 201209, China |
| |
| Abstract: | A self-adaptive differential evolution algorithm with multiple strategies(SMDE) was proposed to overcome premature or localized optimization of differential evolution(DE) as a result of fixed parameter settings. Based on basic framework of classical DE, the first step in SMDE was to create a candidate set of mutation strategy, scale factor(F) and crossover rate(CR). In the followed searching process, mutation strategy, F and CR for each individual variable in next evolutionary generation were determined self-adaptively from the corresponding candidate set according to knowledge learnt from previous searches, so that proper mutation strategies and control parameters could be set at various evolution stages. Compared to other famous DE variants on optimizing 10 routine standard testing problems, SMDE had better search precision and faster convergence rate. Moreover, study on estimation of uncertain parameters in dynamic process systems of chemical engineering showed that SMDE could effectively solve engineering optimization challenges. |
| |
| Keywords: | differential evolution algorithm self-adaptive multiple strategies dynamic system parameter estimation |
|
| 点击此处可从《化工学报》浏览原始摘要信息 |
|
点击此处可从《化工学报》下载全文 |
