论文部分内容阅读
为适应时变乃至模型未知的复杂环境,S.Haykin等提出了一种基于神经网络模式识别的雷达目标检测方法.作者在深入研究该检测系统中主要模块:用作分类器的MLP网络和用作特征提取的PCA网络的局限性后,提出分别运用模糊神经网络与Kohonen神经网络对其检测系统进行改进重构.本文主要介绍基于Kohonen网络特征提取的检测方法,并分别运用模拟与真实数据,与原PCA方法进行了性能比较实验与分析.
In order to adapt to the time-varying or even unknown model complex environment, S. Haykin et al. Proposed a radar target detection method based on neural network pattern recognition. The author deeply studies the main modules of the detection system: MLP network used as classifier and After the limitations of the PCA network for feature extraction, this paper proposes to improve the detection system respectively by using the fuzzy neural network and the Kohonen neural network.This paper mainly introduces the detection methods based on the Kohonen network feature extraction, and separately uses the simulation and the real data, Compared with the original PCA method performance comparison experiments and analysis.