研究分析了模拟退火算法(SA)的缺陷,采用分等级搜索机制,提出了可持续模拟退火算法(HFCSA)后,针对HFCSA算法编码受限问题,引入遗传编程的思想和编码方式,提出一种基于遗传编程的可持续模拟退火算法-GP-HFCSA算法。该算法可以在开放式搜索空间中搜索,实现结构和参数的协同进化。在算法对比测试实验(8特征值问题)中,GP-HFCSA算法的设计结果优于参照算法,证明了GP-HFCSA算法具有良好的效率及自动设计能力,具有广泛的学术价值及应用前景。
In view of the sustainable insufficiencies of simulated annealing algorithm, adopting hierarchical fair competition model and genetic programming coding, a new sustainable simulated annealing algorithm (GP-HFCSA) for open-ended search design was investigated. This algorithm can search in an open-ended space and evolve structure and parameter in the same time. Combining with bond graph, GP-HFCSA can automatically design the multi-field dynamic system which is from an embryo system to an aimed one. So this supplies a new way of auto design in multi-field dynamic system. In the experiment-8-eigenvalue placement problem, GP-HFCSA algorithm shows strong automatic design ability and outperforms the compared algorithm in design result. So, it's proved that GP-HFCSA algorithm has good ability of auto design and will be applied in broad fields in future.