鉴于强制进化随机游走算法(random walk algorithm with compulsive evolution,RWCE)在优化换热网络时可能出现有潜力结构被差解代替,提出了一种结构保护策略,增加一个与原种群平行进化的新种群,新种群执行结构保护,个体与原种群一一对应并接收其当前最优解,同时采用一种降维邻域搜索的进化方式,设置个体各维变量更新概率确定搜索维数,充分挖掘个体结构进化潜力,提升了算法的局部搜索能力;原种群则沿用RWCE的主要操作,保留了较强的全局搜索能力.将采用结构保护策略的RWCE算法用于有分流换热网络优化,取得了优于现有文献的结果.%A structure-protection strategy was proposed to eliminate the deficiency since structures with great evolutionary potentials could be substituted with bad solutions when random walk algorithm with compulsive evolution (RWCE) was applied to heat exchanger network optimization. A new population evolving in parallel with the original population was set to execute structure protection. Individuals in the new population are corresponded with ones in the original population and received their current optimal solutions to form a protective effect. Meanwhile, a novel evolution technique named as "dimensionality-reduction local search" was proposed for the new population. The technique determined the search dimension by controlling the update probability of each dimension, which could tap the evolutionary potentials of the protected solutions fully and thus improving the local search ability. Moreover, major evolution operations of RWCE were adopted by the original population and thus the strong global search ability was maintained. The RWCE with the structure-protection strategy was applied to optimize heat exchanger network with stream splits, some results superior to the reported ones were obtained, demonstrating that the proposed method possessed strong abilities in both the global and local search and improved the search quality.
展开▼
