神经网络在二相编码信号旁瓣抑制中应用

Application of neural network for bi nary-coded sidelobe suppression

研究了人工神经网络在二相编码信号距离旁瓣抑制问题中的应用. 在H.K.Kwan等提出的将多 层感知器网络用于二相编码信号距离旁瓣抑制问题的基础上,采用了一种改进的神经网络学 习算法对网络进行训练. 用13 bit巴克码对网络进行了测试. 实验结果表明,经改进的神经 网络学习算法训练后的网络抗噪性能、多目标分辨能力以及收敛速度都明显优于一般误差反 向传播学习算法(EBP)训练所得的结果.

Neural network approach to range sidelobe suppression is studied. Based on a mul ti-layer perception network approach to range sidelobe suppression presented by H.K. Kwan. et al., a modified learning algorithm is used to train the neura l network. Experiment results for 13-bit Barker code show that the anti-noise performance, resolution capacity of multi-object and convergent velocity of the neural network trained by the modified algorithm obviously outperform those of the network trained by normal error back propagation algorithm.