[关键词]
[摘要]
文献[7]提出的估计方法直接以误差矩阵的2范数建立适应度函数,文中利用该矩阵是Toeplitz矩阵的特性,构造了一个新的适应度函数,提出了一种基于基因分量独立变异的实数遗传算法。该方法不需进行矩阵的特征值分解,充分利用了遗传算法的全局寻优特性,获得了良好的参数估计性能。仿真结果表明,相对文献[7],达到同样要求所需要的运算量减小了一半。
[Key word]
[Abstract]
The method of estimation in paper [ 7 ] is to build fitness function by the 2 - norm of error matrix directly. For the error matrix is a Toeplitz matrix, a novel fitness function was presented. Then a new method, the real -valued genetic algorithm which is based on genes weights variation independently, is developed in this paper. The method need not to calculate eigen- value decomposition and can take good advantage of the overall ability of seeking the optima result of the genetic algorithm. So a good parameter estimation performance can be acquired. Computer simulation proves that for the same result, the operation quantity reduced half compared to the paper[7].
[中图分类号]
TN957
[基金项目]