基于适应小波收缩的浮点数编码遗传算法  

Floating point representation genetic algorithm based on adaptive wavelet shrinkage

作　　者：崔明义[1] 邵超[1] 

机构地区：[1]河南财经政法大学计算机与信息工程学院,郑州450002

出　　处：《计算机应用》2014年第7期2071-2073,2079,共4页journal of Computer Applications

基　　金：国家自然科学基金资助项目(61202285);河南省科技攻关项目(132102210138)

摘　　要：通过独立同分布分析浮点数编码(FPR)噪声,用适应小波收缩的方法消除噪声对遗传算法性能的影响,在算法运行中用变异操作实现消噪。针对阈值变化对小波系数的影响,以单基因证明小波消噪变异的正确性;提出适应小波收缩构建软阈值函数,将函数运算植入算法的动态运行中;给出了具体的实现算法,用实例验证了算法的可行性。仿真实验表明,所提算法显著提高了收敛速度,收敛点与理论值相一致。Noise of floating point representation was analyzed to be independent identically distributed. The noise was removed by adaptive wavelet shrinkage, to improve the performance of genetic algorithm. Denoising was implemented with mutation operation in algorithm running. Aiming at influence on wavelet coefficient with threshold variety, correctness of wavelet denoising mutation was proved with single gene. Soft threshold function was proposed with adaptive wavelet shrinkage, and the function was embedded into dynamic algorithm running. A concrete algorithm was given, and an example was used to verify the feasibility of the algorithm. The simulation experiment indicates the algorithm improves convergence rate, and the convergence point is the same as theoretical convergence point.

关 键 词：浮点数编码 噪声 适应小波收缩 遗传算法 消噪变异 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构] TP302.7[自动化与计算机技术—计算机科学与技术]