连续多蚁群算法的构建及其在过程动态优化中的应用  

Continuous Multi Ant-Colony Optimization and its Application in Dynamic Optimization Problems of Chemical Processes

作　　者：蒲黎明[1] 俞欢军[1] 陈德钊[1] 

机构地区：[1]浙江大学化学工程与生物工程学系,浙江杭州310027

出　　处：《高校化学工程学报》2008年第5期871-876,共6页Journal of Chemical Engineering of Chinese Universities

基　　金：国家自然科学基金(20276063)

摘　　要：动态优化为过程系统工程的重要课题,现有解法存在较多不足,为此构建了连续多蚁群算法(CMACO),可直接用于由动态优化转换成的非线性规划问题。该算法克服了经典蚁群算法只适用于离散问题的局限性,以最优食物源为目标,有多个子群同时搜索。各子群的信息素呈正态分布,独立引导蚂蚁寻优。子群间又相互交流,协同搜索,并逐轮调整子群规模、分布中心和宽度。在可行区域内既全面探索,又加强挖掘,提高了全局优化的性能和速率。将其用于Park-Ramirez和Lee-Ramirez生物反应器的补料流率优化,在优化结果和计算代价上都有一定的优势。In order to solve dynamic optimization （DO） problems efficiently, a new method called continuous multi ant-colony optimization （CMACO） was proposed. This method overcomes the limit of standard ant colony algorithm for solving continuous optimization problems. Control vector parameterization approach was introduced to transform original DO problems to non-linear programming problems, which was directly solved by CMACO. A normal distribution was adopted to simulate pheromone distribution, which was updated according to iterative results. The mean and deviation of pheromone distribution were dynamically tuned for performing local exploitive optimization. Moreover, in order to enhance global exploratory ability, multi ant-colony strategy with specific cooperation mechanism was proposed. The efficiency of the method was illustrated with two challenging DO problems of fed-batch bioreactors, and the results show CMACO has well global optimization ability and fast convergence speed.

关 键 词：多蚁群算法 化工过程 动态优化 信息素 正态分布 流加式生物反应器 

分 类 号：TP021.8[自动化与计算机技术]