邻域搜索算法的统一结构和混合优化策略

邻域搜索算法的多样性和研究成果的分散性阻碍了算法的系统化研究与发展 ,因此建立统一的结构框架很有必要。论文通过对邻域搜索算法的流程和主要环节的分析与归纳 ,基于空间分解和进程分解的思路提出了邻域搜索算法的一种统一结构 ,为优化算法的系统化研究和新型混合优化算法的开发提供了可行途径。进而 ,基于统一结构 ,利用遗传算法和模拟退火算法在优化结构和优化行为上的互补性 ,将它们的优化机制和优化操作合理结合 ,提出了具有并行搜索和概率突跳特性的全局高效 GASA混合策略。

Unified framework for neighbor search algorithms and hybrid optimization strategies

The multiformity of neighbor search algorithms (NSAs) and the diversity of previous results greatly hinder the systematical research and development of NSAs, so a unified framework to organize progresses in the field. This paper analyzes the unified procedure and the main components of NSAs. Then, space division and process division are used to develop a unified framework for NSAs, which provides a feasible method for systematical research and for developing hybrid optimization strategies. Furthermore, the complementary aspects of the structure and behavior of genetic algorithm and simulated annealing are used to combine the mechanisms and operators to develop a class of global hybrid strategy with parallel searching and probabilistic jumping abilities.