换热网络中大步长与接受差解的关联分析及改进策略

Correlation Analysis and Improvement Strategy of Large Step Length and Accepting Imperfect Solution in Heat Exchanger Network

RWCE算法优化换热网络时,较大的优化步长能够在接受差解的作用下促进结构变异,从而有效地跳出局部最优解。但持续的大步长无法保留结构变异后新生成的、换热量较小的换热单元,影响结构进化效率。基于大步长与接受差解的关联性,提出了一种大步长与接受差解联动策略,将接受差解和采用大步长同频,仅当接受差解时才使用大步长,并必定接受大步长作用下的新结构,保证大步长对结构变异的促进作用,解决大步长作用下新生成单元难以保留的问题,全面提升结构进化效率。将大步长与接受差解联动策略应用于20sp和15sp算例,分别得到结果为1 395 389和1 511 551$/a的最优换热网络结构,性能较基础RWCE算法提升7.06%和1.69%,验证了该策略对提升算法结构进化效率的有效性。

When the RWCE algorithm is used to optimize the heat exchanger network, the larger optimization step length can promote the structural variation under the effect of accepting the imperfect solution, thus effectively jumping out of the local optimal solution.However, the continuous large step size cannot retain the newly generated heat exchange unit with small heat transfer capacity after structural variation, which affects the structural evolution efficiency.Therefore, based on the correlation analysis between large step length and accepting imperfect solution, a linkage strategy of large step length and accepting imperfect solution is proposed in this paper.The large step length is used only when the imperfect solution is accepted, and the new structure under the action of large step length must be accepted, which can not only ensure the promotion of large step length on structural variation, but also solve the problem that the new generation units is difficult to retain under the effect of the large step length, so as to comprehensively improve the structural evolution efficiency.In this paper, the strategy of accepting imperfect solution and large step length linkage is applied to 20 sp and 15 sp cases, and the optimal heat exchanger network structures with the results of 1 395 389 $/a and 1 511 551 $/a are obtained respectively.Compared with the basic RWCE algorithm, the performance is improved by 7.06% and 1.69%,which verifies the effectiveness of the strategy to improve the efficiency of structural evolution.