摘要
选择ADPCM的步长修正因子M涉及复杂目标函数的多变量优化,是一个海量计算问题。克隆—思维进化算法(CMEA)继承了进化算法(MEA)中趋同操作的择强汰弱思想,同时引入克隆算子将相对较弱个体的强势因素保留到下一代。提出一种新的基于N进制编码的克隆重组方法用于优化具有复杂目标函数的ADPCM的8个步长修正因子,实验结果表明从第5次迭代后,CMEA的信噪比(SNR)一直高于MEA算法;在前5次迭代中,CMEA平均每次迭代改善1.03dB,高出MEA算法0.4dB。此外MEA的计算量约为CMEA的1.67倍,CMEA比MEA算法抗早熟。
Choice of ADPCM＇s step-size updating factors M had a sea capacity of computing which optimizes multi-variables with complex cost function. The MEA＇s similartaxis operator was come into by the clone mind evolution algorithm （CMEA）, which introduced the clone operator to leave the strong component of the weak individuals to next iterativeness. A novel clone crossover scheme was provided based on N-coding which was used to optimize ADPCM＇s 8 step-size updating factors with complex＇cost function. The experiment result showed that the CMEA＇s SNR was averagely reformed about 1.03dB every generation in previous 5 iterativeness, which was exceed MEA＇s by 0.4dB and overran the MEA＇s from fifth generation. Furthermore, the MEA＇s quantity of computing is equal to CMEA＇s by 1.67 and the latter is of anti-prematurity.
出处
《通信学报》
EI
CSCD
北大核心
2006年第3期28-31,共4页
Journal on Communications
基金
国家自然科学基金资助项目(60372058)
山西省自然科学基金资助项目(20041046)~~
