摘要
通过分析传统SA算法原理和存在的不足,提出三种改进:增加记忆功能,避免遗失当前最优解;设置稳定抽样判定条件,保证全局搜索能力。提供7种扰动机制,提高结果改进效果。设计对比实验验证各种改进,分析出较好参数配置,构造较理想的改进SA算法。经过国际公认的TSPLIB提供的实验数据的验证,改进算法在性能上比GA和传统的SA算法均有较大提高。
By analyzing the principles and the shortcoming of traditional simulated annealing algorithm, three kinds of improvements were proposed： increased memory function to remember the current best solution; set up basic condition for steady sample, ensuring the global search ability. Provided seven kinds of disturb methods im- prove the result. Designed contrast experiment to verify these improvements, and then analyzed the combination of parameters, based on the well combination an acceptable improved simulated annealing algorithm was constructed. After the experimental data verification provided by the internationally recognized TSPLIB, the conclusion be can drew that the improved algorithm was much better than Genetic Algorithm and traditional simulated annealing algo- rithm by performance.
出处
《科学技术与工程》
北大核心
2013年第25期7552-7556,7572,共6页
Science Technology and Engineering
关键词
模拟退火算法TSP
记忆功能
扰动机制
simulated annealing algorithm TSP remember function disturb method
