遗传算法进化中积木块的识别及利用研究.pdfVIP

40 2 西 安 交 通 大 学 学 报 V ol. 40 l 2 2006 2 JOU R NA L OF X IcA N JIA OT ON G U NIV ER SIT Y Feb . 2006 , , ( , 710049 , ) 提出了一种基于积木块识别的遗传算法, 该算法通过对进化过程中的 选积木块进行识别 与利用来加速搜索, 从而避免遗传算法随机搜索的盲目性. 利用经典的对称旅行商问题求解过程来 测试各种识别方法, 再利用积木块的识别结果改进原有遗传算法, 包括改进积木块的识别率以及基 于积木块的交叉变异算子. 与基本遗传算法的计算结果对比分析表明, 所提算法可显著提高遗传 算法的搜索效率, 减小遗传算法随机搜索的波动性. 遗传算法; 积木块; 旅行商问题 T P 18; U 116 A 025 3O9 87X( 2006) 02O0133O05 Study on the Recogni tion and Utilization of Building Block i n the Iterations of Genetic Algorithm L ôJ un , F eng Boqin , L i Bo ( S chool of Elect ronics and In format ion Engineering , Xican Jiaot ong U niv ersit y , Xic an 7 10049, China) Abstract A genet ic alg orit hm ( GA ) b ased on bu ilding blo ck reco gnition w as pro posed , in w hich bu ilding block candidat es w ere recog nized in evo lving pr ocess t o sp eed up t he sear ch so as t o avo id the blindnes s o f GA random searching. T he t y pical symm et ric T SP ( t raveling salesman pr oblem) solv ing proces s w as u sed to t est various reco gnition m et hods , and t hen t he test results w ere used to improv e t he tr aditional GA , including improv ing t he r ecognit io n rat e of building block s and mutat ion and cros sov er op er at or s based on bu ilding block . Co mp ared w it h t he co mput atio nal re- sult of t radit ional GA , it show s th at t he searching ef ficiency of GA can be improved remarkably and t he flu ct u ation of random searching can be redu ced by reco gniz ing building block. Keywords g enet ic alg orit hm ; bu ilding block; t raveling salesman problem ( GA ) , . ( H o lla