| Heat exchange network generally exists in the energy recovery link of process industrial system,and its optimization research is of great practical significance to improve energy utilization efficiency and reduce emissions.Heat exchanger network synthesis is a complex mixed integer nonlinear programming problem,and its serious non-convex nonlinear characteristics lead to many local extremums in its solution domain,and the global optimization is facing huge challenges.Due to the greedy structure acceptance criterion,the traditional heuristic methods are very easy to fall into the trap of local optimum,which affects the ability of structure variation and evolution.Therefore,aiming at improving the global optimization ability,this paper focuses on the difficulties of structural optimization and the evolution mechanism of individual selection,and proposes a damping optimization algorithm and a variety of performance improvement strategies,in order to comprehensively improve the optimization quality and efficiency of the algorithm.The main research contents and innovations are as follows:1)In view of the fact that the unstructured model of the heat exchange network with shunt nodes cannot achieve the uniform distribution of individuals in the structure and continuous variables under the condition of limited population size,the construction strategy of uniform initial population is proposed,that is,the solution space of the model is expanded by adjusting the generation mode of the initial structure;on this basis,while improving the diversity of structural matching,the population uniformity accompanying optimization strategy is added to improve the evolutionary vitality of the population by periodically eliminating individuals with similar structure.The results show that the generation of the initial structure can effectively expand the space of structural optimization and delay the process of falling into local extremum.2)In order to reduce the greed of algorithm structure evolution,a new damping RWCE optimization algorithm is established.The approach to the local optimal solution is delayed by probabilistically not accepting the solution with reduced cost,so as to jump out of the current local optimal solution area and enhance the global search ability.At the same time,based on the difference of structural evolution ability in different optimization stages,a phased delay strategy is proposed to take a reasonable value of delay probability to promote structural evolution.Through the verification of the numerical example,it shows that the damping optimization method can effectively jump out of the local extreme value to obtain a better structure,thereby improving the global optimization ability of the algorithm.3)On the basis of the aforementioned research,a selective delay strategy is proposed by establishing the judgment index of structural variation.That is,by identifying the source of annual comprehensive cost decline,selectively accept the better solution obtained by continuous variable optimization with a certain probability and adjusting the value of the delay probability in different optimization stages to promote the fuller evolution of structural variables.The research shows that under the guidance of whether the structure changes or not,selecting the structure whose objective function decreases can effectively delay the process of falling into local optimum due to the rapid optimization of continuous variables,and improve the performance of structural evolution.4)Based on the judgment of delay demand at different stages in the optimization process,a dynamic selection function of delay probability based on the evolution process is established,and the delay probability is randomly adjusted according to the structural variation ability in the optimization process to avoid premature convergence of the algorithm.At the same time,the health index of the optimization state of the algorithm is established to evaluate the optimization performance of the global search of the current population.By setting parallel evolutionary population,periodic back generation and other operations,the loss of population evolutionary vitality is delayed,and the optimization efficiency and quality of the algorithm are comprehensively improved.In a word,this paper establishes a damped RWCE algorithm for heat exchange network by delaying the individual from falling into the local optimal solution prematurely and adjusts and controls the parameters of its evolution process.It not only promotes the full evolution of structural variables,but also takes into account the optimization of continuous variables,which provides an important research idea for the global optimization of mixed integer nonlinear programming problems. |