本文研究了基于模块化神经网络的谐振区雷达目标识别方法,该方法先用BP网络进行波形预测,再用最大后验概率准则或修正最小平方误差准则进行分类。通过计算机模拟,证实了该方法有较好的识别性能,且对信号留数变化,即目标的姿态变化不敏感。另外,该方法还具有实现简单,结构扩展方便等优点。 In this paper, the method of radar target recognition based on modularized neural network is studied. This method uses BP network to predict the waveform, and then classifies it by the maximum a posteriori probability criterion or the modified least square error criterion. Through computer simulation, this method is proved to have better recognition performance, and it is insensitive to changes in signal residuals, that is, target attitude changes. In addition, the method also has the advantages of simple implementation, convenient structure expansion and the like.