本文提出了一种采用多层前馈神经网络的雷达探测技术即脉冲压缩方法,并且采用巴克码进行说明。这个网络拥有13个输入单元、3个隐含单元和1个输出单元。应用逆向传播的学习方法来训练网络,运用这种方法可以很容易地达到40dB的峰值信噪比,而且其处理时间比那些采用相关器和失配滤波器的方法要更快。 In this paper, we propose a radar detection technique using multi-layer feedforward neural network, ie pulse compression method, and use Barker code to illustrate. This network has 13 input units, 3 hidden units and 1 output unit. Using a backward propagation learning method to train the network, a peak signal-to-noise ratio of 40 dB can easily be achieved using this method, and its processing time is faster than those using correlators and mismatched filters.