基于模拟退火算法的换热网络双层优化方法

Bilevel Optimization Method for Heat Exchanger Network Synthesis Based on Simulated Annealing Algorithm

针对同时存在整型变量及连续变量两种变量形式的换热网络优化问题,提出一种基于模拟退火算法的双层优化方法,分别处理代表结构的整型变量及代表费用的连续变量优化。外层算法以经济费用目标作为判断换热网络结构优劣的标准,通过随机扰动产生试探结构,借助模拟退火机制逐步进化外层结构;内层算法以流股间匹配的换热量作为优化变量,采用模拟退火算法实现费用连续变量的优化,再反馈到外层的结构优化过程。两个具体算例证明:模拟退火算法应用于换热网络结构的整型变量优化可以找到较优换热网络结构,而双层优化方法可有效处理具有整型变量和连续变量两种变量形式的换热网络优化问题,取得较好的换热网络设计。

A bilevel optimization method based on simulated annealing（SA） algorithm is proposed for heat exchanger network synthesis. The upper level randomly generates the topologies of heat exchanger networks combined with 0-1 integer variables and then these topologies are sent to the lower level in which the heat load of units are optimized by the SA algorithm designed for continuous variables. The upper algorithm converges to the optimal solution by evaluating the minimum annual cost of each topology with the Metropolis rule. The results show that the SA algorithm can optimize the topologies of the heat exchanger networks effectively, and this bilevel optimization method is suitable for the optimization problems with integer variables and continuous variables simultaneously and able to find more economical heat exchanger networks.