混沌和神经网络相结合预测短波通信频率参数

• 作 者: 简相超 郑君里
• 机 构: 清华大学电子工程系!北京100084,清华大学电子工程系!北京100084
• 出 处: 清华大学学报(自然科学版)
• 发表日期: 2001年01期
• 关键词: 混沌 神经网络 预测 短波通信 奇异值分解 噪声

为提高短波通信的可靠性 ,提出了一种将混沌和神经网络相结合的方法预测短波通信频率参数。利用混沌方法重构相空间系统吸引子 ,用前向多层神经网络拟合吸引子上的全局整体映射 ,构成混合预测模型。实验结果表明 ,将此混合模型用于预测短波通信频率参数如 F2 层临界频率 ffo F 2 ,能达到较好的预测效果 ,可以应用到实际预测系统中。还将基于奇异值分解的噪声消减滤波算法应用到数据预处理中 ,预测结果表明了这种办法能提高预测精度

In order to improve the reliability of shortwave communication, a method of combining chaos with neural network is proposed to predict the frequency parameters of shortwave communication. The chaos method is used to reconstruct the phase space system attractor, and the forward multi-layer neural network is used to fit the global global map on the attractor to form a hybrid forecasting model. The experimental results show that this hybrid model can be used to predict the short-wave communication frequency parameters such as ffo F 2 F2, which can achieve good prediction results and can be applied to the actual prediction system. The noise reduction filtering algorithm based on singular value decomposition is also applied to data preprocessing. The prediction results show that this method can improve the prediction accuracy