摘要
针对现有随机算法对高维能量集成网络(换热网络)求解困难与易陷入局部最优的问题,提出了一种基于参考点的非支配排序遗传算法(NSGA-Ⅲ)的换热网络优化方法。该方法使用特殊的支配关系引入了自适应离散化Pareto前沿与参考点机制,对那些非支配且接近参考点的种群个体进行保留。本文基于非等温混合分流分级超结构模型,以用户最关心的环境影响指数、投资费用、操作费用与改造工程量为改造优化目标,从多个方面、多个层次考虑换热网络改造问题,为用户提供多种可供选择的改造方案。案例研究表明,改造后的换热网络以每年57304USD的投资费用节省25%的年度操作费用,同时还具有较小的环境影响指数与改造工程量;与文献相比,可以提供多元化的节能改造方案,满足用户的不同改造需求,且获得优于文献的综合改造结果,表明该方法具有较强高维求解能力。
Aiming at the problem that the existing random algorithm solves the problem of highdimensional energy integrated network(heat exchanger network) and is easy to fall into local optimum, a non-dominated sorting genetic algorithm(NSGA-Ⅲ) based on reference points was proposed to optimize heat exchanger networks. In this method, the adaptive discretization Pareto frontier and reference point mechanism were introduced to reserve the population individuals that are non-dominated and close to the reference point by using a special dominant relation. Based on the non-isothermal mixed-flow hierarchical superstructure model, it was considered the retrofit of heat exchanger network from many aspects and levels, by taking the environmental impact index, investment cost, operation cost and retrofit engineering quantity as the optimization objectives. The purpose was to provide users with a variety of alternative retrofit programs. The case study showed that the investment cost of the modified heat exchanger network is 57304 USD per year, which saves 25% of the operating cost annually. Compared with the literature, it can provide diversified energy-saving retrofit schemes to meet different retrofit requirements of users, and obtain better comprehensive retrofit results than the literature, which demonstrates that the NSGA-Ⅲ has the superiority of solving high-dimensional problem.
作者
谢小东
范伟
蒋宁
郭风元
李恩腾
徐英杰
XIE Xiaodong;FAN Wei;JIANG Ning;GUO Fengyuan;LI Enteng;XU Yingjie(College of Mechanical Engineering,Zhejiang University of Technology,Hangzhou 310023,Zhejiang,China)
出处
《化工进展》
EI
CAS
CSCD
北大核心
2020年第3期872-881,共10页
Chemical Industry and Engineering Progress
基金
浙江省自然科学基金(LY18E060010)
国家自然科学基金(51206147)。
