基于MEA-WNN的短波通信最佳频率预测

针对短波通信中通信频率选择不恰当导致信号衰落严重、传输不可靠等问题,提出一种基于思维进化的小波神经网络(mind evolutionary algorithm-wavelet neural network,MEA-WNN)和混沌理论相结合的短波通信频率预测方法.采用具有良好的非线性拟合特性的小波神经网络作为预测模型,利用混沌理论重构相空间,确定神经网络各层节点个数,并用思维进化算法优化网络的初始权值与网络中小波函数的伸缩因子和平移因子.实验表明,MEAWNN算法能提高短波通信f0 F2的预测精度.

Prediction of Best Frequency Parameters of HF Communication Based on MEA-WNN

Considering the problem caused by inappropriate selection of communication for HF communication frequency, such as, severe signal fading, unreliable transport, a method for HF communication frequency prediction was presented in this paper based on mind evolutionary algorithm-wavelet neural network combining with chaos theory. Wavelet neural network which has good characteristics of nonlinear fitting as predictive model was used. Chaos theory of phase space reconstruction was used to determine the number of layers of the neural network nodes. Also, mind evolutionary algorithm was used to optimize the initial weights of the network, the network stretch factor and the translation factor of wavelet function. Experiments show that MEA-WNN algorithm can improve the prediction accuracy of HF communication frequency.