BDS/GPS短期ISB建模和预报的Kalman滤波方法

ISB是多系统PPP数据处理中必须要考虑的一项误差,因此有必要对BDS/GPS短期ISB建模和预报进行研究。为了提高ISB预报精度,针对等权LS(least square)估计ISB模型参数时忽略了拟合数据权重不同的问题,提出了采用Kalman滤波对模型参数进行估计,并根据ISB拟合数据距预报时刻的远近调整Kalman滤波拟合数据的方差。本文采用7d的ISB数据进行建模,根据所建模型预报第8天的ISB值,并对预报精度和定位结果进行了验证。进行试验的4个测站Kalman拟合模型的ISB预报精度比LS拟合模型分别提高了29.7%、11.5%、43.5%和32.0%。采用Kalman拟合模型的ISB预报值作为先验约束,PPP平均定位精度在E和U方向上比采用LS拟合模型预报值分别多提高了2.7%和0.9%,比不加ISB先验约束在E、N、U方向分别提高了10.6%、26.3%和3.4%。

A Kalman Filter Method for BDS/GPS Short-term ISB Modelling and Prediction

Study of inter-system biases(ISB) is essential for the data processing of multi-constellation precise point positioning(PPP).When establishing BDS/GPS short-term ISB models, as equal-weight least square(LS) fails in taking the different weights of fitting ISB data into consideration, a method based on Kalman filter is proposed for the parameter estimation of ISB modelling, and the variance of fitting ISB data in Kalman filter is adjusted according to the distance between the data time and the forecast time.ISB models are established with ISB data over 7 days and the accuracy and applicability of ISB prediction for the 8th day is verified by static PPP experiments.The analysis result shows that the accuracy of ISB prediction generated by Kalman filter model is 29.7%, 11.5%, 43.5% and 32.0% higher than those generated by LS model at 4 stations, respectively.With the priori constraints of ISB prediction generated by Kalman filter model, the averaged RMS values of static PPP solutions are promoted by 2.7% and 0.9% higher in E and U components than those with priori constraints generated by LS model, and are promoted by 10.6%, 26.3% and 3.4% higher in E, N, U components than those without priori constraints, respectively.