论文部分内容阅读
传统的自适应方法一般采用梯度下降法搜索最佳权，但梯度法的性能对步长敏感。本文介绍一种全新的、无需步长选择的稳定收敛的ＦＩＲ 方法。该方法采用几何中心法进行递推。仿真结果表明由该方法导出的算法性能很好。
The traditional adaptive method generally uses the gradient descent method to search the optimal weight, but the performance of the gradient method is sensitive to the step size. This article describes a new, stable FIR method that does not require step size selection. The method uses the geometric center method for recursion. Simulation results show that the algorithm derived by this method performs well.