基于滤波与BP神经网络的水下探测信号检测算法

针对水下探测系统探测船舶磁场信号时信噪比较低的问题,首先根据磁异常信号的频域特征,设计了约束最小二乘FIR滤波器,通过对含噪信号进行带通滤波,滤除高频噪声;再采用BP神经网络对低频分量进行学习,提取船舶目标特征信号。将该算法应用于船模实测实验,结果表明:该算法可以显著提高信噪比,增强对船舶磁场信号的检测能力。

Signal detection algorithm of underwater detection based on band-pass filter and BP neural network

Aimed at the low SNR in detecting magnetic signal of a ship,a detection algorithm was proposed based on band-pass filter and BP neural network.Firstly,according to the frequency characte-ristic of magnetic anomaly signal,the constrained least square FIR filter was designed by which to filter the raw signal and eliminate the high frequency noise.Then,BP neural network was used for processing the low frequency components so as to pick up the characteristic signal of ship target.The results of ship model experiments show that the algorithm can remarkably improve SNR and enhance the capability of detecting signals of ship′s magnetic field.