赵 健^1,2, 樊彦国^1,2, 丁 宁¹
1.中国石油大学(华东) 地球科学与技术学院;2.青岛海洋科学与技术国家实验室海洋矿产资源评价与探测技术功能实验室

摘要:
在对海平面变化规律进行深入分析的基础上, 应用最小二乘神经网络组合模型对海平面变化趋势进行预测; 对卫星测高海平面异常序列中的周期项及线性趋势项利用最小二乘模型进行拟合, 残差部分则采用径向基函数神经网络模型进行预测。对中国近海海域卫星测高海平面异常序列的预测表明,连续1个月的预测精度为0.52 cm, 3个月的预测精度为0.65 cm, 证明了该组合模型在海平面变化短期预测方面的可靠性, 其在海平面变化预测领域具有较高的应用价值。
关键词:海平面异常最小二乘拟合径向基函数神经网络预测精度
DOI：10.11759/hykx20161012001
分类号:
基金项目:中央高校基本科研业务费专项资金资助(18CX02066A); 山东省自然科学基金项目(ZR2014DQ008); 中国石油科技创新基金项目(2015D-5006-0302)

Sea level anomaly forecasting using least square and the radial basis function neural network
ZHAO Jian1,FAN Yan-guo,DING Ning
Abstract:
Sea level change is characterized by nonlinear, time-varying, and highly uncertain characteristics and it is difficult to obtain satisfactory forecasts using conventional linear models. Based on a comprehensive analysis of sea level changes, we applied a least square-neutral network combined method to the short-term forecasting of sea level change using sea level anomaly (SLA) data. Periodic terms and linear trends in sea level change were fitted and extrapolated using the least square model, while the forecast of the stochastic residual terms was performed using the radial basis function (RBF) neural network model. A test of the combined model with different RBF network structures was carried out in China’s offshore waters using satellite altimetry SLA data Accuracies of 1 month and 3 months’ forecasts were within 0.52 cm and 0.65 cm, respectively. The results prove the reliability of the least square-neutral network combined model in short-term forecasting of sea level variability; the model has significant applicability in the field of sea level change forecasting.
Key words:sea level anomalyleast square fittingradial basis function networkforecast accuracy





PDF全文下载地址:

点我下载PDF