关键词: 粒子群算法, 启发式算法, 换热网络综合, 局部搜索策略

Abstract: The optimization of heat exchanger networks synthesis (HENS) still remains an open problem because of the complexity nature of stream matches. The complexity of mixed integer nonlinear program (MINLP) keeps the opening for the heuristic algorithms including particle swarm optimization (PSO), and the HENS play an important role in the chemical process optimization. Two strategies that dedicated to improve the performance of PSO for optimal design of heat exchanger network are proposed. Although the standard PSO algorithm is capable of detecting the promising region, it exhibits the deficiency of that they cannot perform a refined local search to compute the optimum with high accuracy once there. The matter that the best position ever recorded is not the true minimum fitness will lead to the excessive roaming the search spaces and frequent updating. Local search technique to overcome above difficulties was proposed. In local search component, two local search strategies that perform more refined search around potential solutions of particle at hand were proposed. In cases take consideration of the fixed equipment cost, a formula alteration strategy was proposed to overcome the matter that, due to existence of the non-zero fixed equipment cost a, the evolution will likely to be trapped into a local optimum. Since the area cost is relatively small compared with the fixed equipment cost a in the initial iteration phase of evolution. The proposed strategies has been applied to several four streams cases taken from the literature and the results are very encouraging. The presented case studies revealed special search ability in local optimization incorporated with the proposed strategies.

Key words: particle swarm optimization, heuristic algorithm, heat exchangers network synthesis, local search strategies