循环神经网络自适应有源噪声对消研究

介绍了一种新颖的非线性逢适应滤波器－－循环神经网络自适应滤波器。由于这种循环神经等效于非线性ＩＩＲ滤波器，具有学习非线性函数到任意的精度及自适应能力，这种滤波器优于线性滤波器，能够适应各种噪声环境。本文将该滤波器用于有源噪声对消，仿真结果表明了这种循环神经网络自适应有源噪声对消系统具有良好的抗噪声性能。

Study of Recurrent Neural Network Adaptive Active Noise Cancellation

A novel class of nonlinear adaptive active noise cancellation filte r-recurrent neural networks adaptive active noise cancellation filter is introdu ced. The recurrent neural networks, which is equivalent to IIR filter, is able o f learning any nonlinear function to arbitrary accuracy, this filter is better t han the linear filter and adapt to different kinds of noisy environments. It is applied for adaptive active noise cancellation. Experimental results of adaptive active noise cancellation show that the proposed recurrent neural networks adap tive active noise cancellation filter is not also of good anti-noise performance .