多层前向神经网络在ECG信号滤波中的应用

THE APPLICATION OF BACK PROPAGATION NEURAL NETWORK IN THE FILTERING OF ECG SIGNALS

针对心电信号检测中主要存在的几类噪声,利用BP神经网络的非线性映射能力设计了基于BP神经网络的非线性滤波器,实现了对心电信号中工频干扰、基线漂移、电极移动等噪声的滤除。经仿真与实验验证表明,该方案对于工频干扰、基线漂移、电极移动等噪声均有较好的滤波效果,且算法具有较好的收敛特性,还可以方便地实现非线性滤波。

To deal with the noises existing in ECG signals, we design nonlinear filter based on multilayer back propagation neural network which has the capacity of nonlinear transformation, and achieve the cancellation of powerline interference, baseline wander and electrode motion. In virtue of simulation and experiment validation, the noises of powerline interference, baseline wander and electrode motion can be filtered effectively. The algorithm has very good convergence, and is easy to carry out nonlinear filters.