船载雷达测波海浪谱预报方法研究  

Research on Forecasting of Wave Spectrum Based on X-band Radar

作　　者：张新宇[1] 石博文[2] 石爱国[1] ZHANG Xin - yu SHI Bo - wen SHI Ai - guo(Dept. of Navigation, Dalian Naval Academy, Dalian Liaoning 116018, China Navigation College, Dalian Maritime University, Dalian Liaoning 116026, China)

机构地区：[1]大连舰艇学院航海系,辽宁大连116018 [2]大连海事大学航海学院,辽宁大连116026

出　　处：《计算机仿真》2017年第4期31-35,共5页Computer Simulation

摘　　要：以船载雷达测波系统WaMoS II船载实时测量的海浪谱信息为基础,应用于波分解思路和时间序列分析方法,进行海浪谱预报研究。因海浪具有非平稳性和非线性,采用具有训练算法简捷、泛化能力强、非线性处理能力强等优点的回声状态网络作为时间序列预报方法,对已知海浪谱时间序列样本数据进行训练,建立非线性模型,对未知海浪谱信息进行预报。通过应用上述方法,实现了海浪频谱预报和方向谱预报,其中方向谱的预报为首次实现。仿真结果表明,上述算法在中短期范围内(预报长度6~8小时)可以达到较高的预报精度。The paper focused on forecasting of wave spectrum based on data obtained by shipboard X - band Radar system WaMoS II, with the basic idea of wavelet decomposition and time series analysis. As ocean wave is non - stationary and nonlinear, the paper choose echo state networks （ESNs） with merits of celerity, generalization ability and non - linearity as forecasting method. Without prior information of sea waves, only using the real measured wave spectrum data,the ESNs method is applied to solve the problem of time series forecasting. Using this method, wave frequency spectrum and directional spectrum forecasting is carried out, where directional spectrum forecasting is the first time. Results show that the method satisfies the need of online forecasting within 15 seconds.

关 键 词：回声状态网络 频谱 方向谱 时间序列预报 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]