Search

Random walk algorithm with compulsive evolution (RWCE) performs better for global optimization of heat exchanger network synthesis than other stochastic methods though it faces the inevitable decrease in optimization efficiency during the late period. This paper aims to analyze the relationship between population diversity and optimization performance. Firstly, population diversities are contrasted to explain the relatively satisfactory performance of RWCE. It is found that the optimal structures for individuals become significantly similar with evolution. Secondly, key evaluation indexes are established to identify similar individuals and clusters formed during the evolution. A noteworthy observation of the study is the clustering phenomenon in RWCE where most individuals cluster together with structural community while 5% individuals alone are excluded from the clusters. Finally, a population-diversity enhancement strategy is established to weaken the negative effect of the clustering phenomenon and enhance the global optimization efficiency. The new strategy is efficient in achieving more economical solutions compared to those in literatures. [ABSTRACT FROM AUTHOR]