换热网络优化中粒子群算法早熟现象的机理分析及改进策略  

Analysis and improvement strategy of the premature phenomenon in particle swarm optimization for heat exchanger network

作　　者：李帅龙 崔国民[1] 肖媛[1] LI Shuai-long CUI Guo-min XIAO Yuan(Institute of New Energy Science and Technology, University of Shanghai for Science and Technology, Shanghai 200093, Chin)

机构地区：[1]上海理工大学新能源科学与工程研究所,上海200093

出　　处：《能源工程》2017年第1期9-16,23,共9页Energy Engineering

基　　金：国家自然科学基金资助项目(51176125);沪江基金研究基地专项(D14001)

摘　　要：粒子群算法应用于换热网络综合主要存在的问题是在优化后期经常出现早熟收敛现象,由于算法全局搜索能力的迅速退化导致换热网络优化进程陷入停滞。通过考察种群多样性的变化,并采用灰度图跟踪每个粒子的差异性演化进程,揭示了算法早熟的本质,在此基础上提出了一种随机扰动策略,在粒子群搜索后期选择一部分粒子随机产生新的速度,改善这一阶段粒子群的种群多样性,增强算法的全局搜索能力,通过换热网络优化算例说明该策略的有效性。Particle swarm optimization （PSO） algorithm as a kind of heuristic method for heat exchanger networks syn- thesis （HENS） , has a strong ability to explore the global optimal region. However, the particles may trap into the local optimum and converge prematurely in the late evolution. To investigate the influence of the population diversity on the performance and the premature convergence of PSO, the differences of particles in the process of optimization were tracked using the gray scale map. Based on the investigation, a novel random disturbance particle swarm optimization （RDPSO） was proposed to enrich the population diversity consistently and strengthen the global search ability of PSO, giving a part of particles new and random velocities. It had been applied to several cases taken from the literature and the results were very encouraging and better than those in other improvements of PSO.

关 键 词：换热网络优化 粒子群算法 种群多样性 随机扰动 

分 类 号：TK124[动力工程及工程热物理—工程热物理]