顾及噪声的CORS站高程非线性建模方法

Nonlinear modeling method of CORS height considering noises

针对当前连续运行参考站(CORS)高程坐标的数据预测,仅建立了包含周期项、趋势项的模型,未考虑到噪声项影响的问题,提出1种顾及噪声的CORS站高程非线性建模方法:对高程时间序列,建立非线性最小二乘周期拟合模型;对得到的模型残差项(噪声项),建立自回归(AR)预测模型;最后将非线性最小二乘周期拟合模型与AR模型进行组合,对CORS站高程分量进行预测。实验结果表明,非线性最小二乘周期拟合和AR的组合模型,较传统的非线性最小二乘周期拟合模型预测精度高。

Aiming at the problem that the current height coordinate prediction of Continuously Operating Reference Stations(CORS)is only eatablished the models with period term and trend term,without considering the impact of noise term,the paper proposed a nonlinear modeling method for the height of CORS regarding for noises:a nonlinear least squares periodic fitting model for the height time series was established;an Auto-Regressive(AR)prediction model was given for the residual term(noise term)of the obtained model;finally the nonlinear least squares periodic fitting model was combined with the AR model to predict the height component of the CORS stations.Experimental result showed that the prediction accuracy of the proposed combined model would be higher than that of the traditional nonlinear least squares period fitting model.