噪声有源控制的递归神经网络方法

Active Noise Control Using a Diagonal Recurrent Neural Network

使用 Filter- X算法研究有源噪声控制问题 ,存在需要较高阶次的滤波器和当噪声出现非线性时控制效果不佳的缺陷。为此提出一种基于对角递归神经网络的非线性噪声有源自适应控制方法 ,并给出一种基于误差梯度下降的在线学习算法 ,同时证明了闭环控制系统在 L yapunov意义下的稳定性。数值仿真表明 。

The active noise control (ANC) problem is studied. An adaptive active nonlinear noise control approach using a diagonal recurrent neural network is derived, which can overcome the disadvantages of the Filter X method. An on line learning algorithm based on the error gradient descent method is proposed. the stability of closed loop system is proved in Lyapunov′s sense. Simulation results show that this method is very efficient to the noise control.