小波分解及神经网络在海浪预报中的应用

提出采用小波分解与递推平方根法神经网络模型相结合的方法进行海浪预报。采用小波分析方法既能把握海浪的发展变化趋势，又能简化预报模型，同时基于递推平方根法的神经网络模型预报方法不仅收敛速度快，又能很好地提高精度，减少计算量。即先将不规则海浪信号用小波分析方法进行多尺度一维分解，得到相对简单、规则的准周期分量信号，然后用一种基于递推平方根法的神经网络模型对各重构信号进行预报，最后对预报结果进行集成。最终的仿真结果验证了该方法的有效性。

Application of Wavelet Decomposition and Neural Network in Sea Wave Prediction

A new method of the sea wave prediction is presented in this paper, which uses both the method of wavelet decomposition and the neural network model based on recursive square root. Because the trend of the motion of the sea wave could be obtained easily and the model for the sea wave prediction could be simplified when the method of wavelet is employed. Further more, the method using neural network model based on recursiv~ square root methods can not only easily be carried out with rapid constringency, but also can improve the precision of the prediction with much less computing step. First in this paper the irregular sea waves are decomposed by multilevel 1 - D wavelet to gain relative simple and regular period signals. Then the neural network model based on recursive square - root methods is used to predict the reconstructed signals. At last, the results of the prediction are composed. Finally, the results of the simulation experiments show that this method is effective.