基于Aspen Plus和NSGA-Ⅱ的隔壁塔多目标优化研究

Multi-Objective Optimization of Dividing Wall Columns with Aspen Plus and NSGA-Ⅱ

以年总操作费用(TAC)和再沸器负荷为目标,提出了基于遗传算法NSGA-Ⅱ的优化方法,并将该方法应用于BTX分离隔壁塔的优化设计。首先应用Aspen Plus软件建立了BTX分离工艺的隔壁塔Radfrac两塔模型,并在Matlab平台上,通过MAP接口工具箱实现Matlab对Aspen Plus的操作与控制,同时Matlab调用NSGA-Ⅱ进行优化,完成种群大小为600、遗传代数为28的模拟过程,得到了10组Pareto解。研究表明,对于Pareto解分布,气相分配量βg、液相分配量βL、侧线抽出位置NS和液相分配位置NL基本不变,进料位置NF、预分馏塔板Nj和主塔板数Ni存在一定的线性关系。

An optimized method was used to evaluate the performance of dividing wall columns in BTX separation with multi-objective genetic algorithm to minimize the TAC and Heat duty. A two-column Radfrac model was established for the separation of BTX by Aspen Plus software. Matlab can operate and control Aspen Plus using MAP and it can use NSGA-II for optimization. Ten Pareto fronts were obtained with population of six hundred and generation of twenty-eight. The results show that interconnection vapor flow ？g, interconnection liquid flow ？L, stage of side product NS and stage of interconnection liquid NL are relatively stable on the base of distribution of Pareto front. However, there is a linear relationship between the number of the stages of prefractionator and that of the main column. The feed stage has a linear relationship with the prefractionator stage numbers.