摘要
状态变量带约束的过程动态优化问题是化工系统工程的重要课题,有一定的难度。通过将其转换为等价的非线性规划后,可采用元启发式方法求解。人工免疫系统的克隆选择算法(CSA)简练易用,全局搜索性能良好,但局部寻优能力较弱,且无处理约束的机制。为此,拟引入免疫网络自学习算子,均匀设计方法,以及目标与约束分离的处理机制,构建改进的克隆选择算法(ICSA),并将其用于状态变量带约束的间歇反应器和乙醇生物反应器的动态优化等实例,效果良好。试验结果表明三种策略有效地改进了CSA的性能,使ICSA能以较少的计算代价搜索到较优的控制策略。
Process dynamic optimization with state-variable constraints is an important subject in process systems engineering,and is difficult to be solved. Through transforming dynamic optimization problem into equivalent nonlinear problem,the meta-heuristic methods were adopted to solve it. Clonal selection algorithm （CSA） originated from artificial immune systems is easy to be implemented and has well global exploration ability. However,its local exploitation ability is relatively weak,and it lacks of constraints handling mechanism. In this paper,the CSA was improved in following sides：（1） self-learning operator of immune network was introduced to enhance local exploitation ability; （2） uniform design method was adopted to generate the initial antibodies; （3） separation method of objective and constraints was introduced to handle state-variable constraints. Finally,an improved clonal selection algorithm （ICSA） was proposed,and it was used to solve the constrained dynamic optimization problems of both reactor and bioreactor of producing ethanol. The satisfactory results illustrate that these three adopted strategies can improve the performance of CSA effectively,and the ICSA can find the optimal control strategy with less computation cost.
出处
《高校化学工程学报》
EI
CAS
CSCD
北大核心
2009年第5期858-863,共6页
Journal of Chemical Engineering of Chinese Universities
基金
国家科学基金资助项目(20276063)
关键词
克隆选择算法
自学习算子
均匀设计
状态变量
约束
动态优化
clonal selection algorithm
self-learning operator
uniform design
state variable
constraints
dynamic optimization
