【摘要】 根据CORS(Continuously Operating Reference System)站点多年积累的连续观测数据,研究站点坐标高程分量的变化规律。对高程时间序列进行了谱分析,并对CORS站点高程时间序列进行了预报分析。采用BP神经网络方法对时间序列模型进行补偿,建立了一个融合模型。实例分析结果表明,与传统的多项式函数拟合模型和时间序列模型相比,提出的融合模型预报精度可提高10%以上。更多还原

【Abstract】 Based on the continuously observation data of CORS stations,it analyzed the change regulation of vertical coordinate time series. The height time series of CORS stations are investigated by using spectrum method. And this paper also studied the prediction analysis of vertical coordinate time series. By using BP neural network method,a fusion model was built. The result indicates that the fusion model achieve a higher precision. The accuracy of fusion model is about 10 percent higher than the polynomial fitting function model and time series model.更多还原