论文部分内容阅读
给出一种隐层由小波基组成的神经网络用于实现频率步进毫米波雷达目标一维距离像的识别 .利用小波变换所具有的良好的时频分析特性 ,实现了输入输出之间映射关系的多分辨学习 .介绍了小波神经网络的数学框架及其误差反向学习算法 .详细描述了用小波神经网络进行识别的步骤 .将所提出的小波神经网络用于频率步进毫米波雷达目标一维距离像的识别 .实验结果表明该方法对目标距离像的识别是有效的
A hidden layer neural network composed of wavelet bases is proposed to realize the one-dimensional distance image recognition of frequency step millimeter-wave radar. By using the good time-frequency analysis features of wavelet transform, the mapping between input and output is realized Relational multiresolution learning.The mathematic framework of wavelet neural network and its error back learning algorithm are introduced.The steps of identifying with wavelet neural network are described in detail.The proposed wavelet neural network is applied to the target of frequency step millimeter wave radar One-dimensional distance image recognition. Experimental results show that this method is effective for the recognition of the target distance image