Research On GNSS Real-time Precise Satellite Clock Estimation Algorithm

High-precision and real-time satellite orbit and clock products are the prerequisite for realizing real-time precise point positioning(RT-PPP)technology,and the accuracy of the products directly determines the positioning performance of RT-PPP.The current real-time orbit prediction product(2 hours)has an accuracy of about 5 cm,which meets the demand for centimeter-level positioning.However,due to the complex environment of space,the accuracy of real time predicted clock products(2 hours)is only about 5 dm,and the low accuracy of predicted clock limits the RT-PPP application.Therefore,the high-precision GNSS real-time satellite clock estimation technology based on real-time data stream has become a research hotspot for RT-PPP.The current GNSS real-time satellite clock estimation algorithm has limitations such as being greatly affected by orbital errors,not taking into account the influence of unstable stations,and imperfect quality control,which make it difficult to obtain robust and high-precision real-time satellite clock products.To address the above problems,this paper systematically investigates the mathematical model and quality control of GNSS real-time precise satellite clock estimation,constructs a stochastic model that takes into account the orbit error,proposes a functional model that takes into account the station coordinate error,and designs a fast comprehensive quality control method,and verifies the effectiveness and accuracy of the proposed algorithm by comparing with precise clock products and analyzing the real-time dynamic PPP results.The main research results achieved are as follows:(1)To address the problem of the impact of real-time satellite orbit error on clock estimation,an improved stochastic model that takes into account the orbit error and satellite altitude angle is proposed by using the real-time orbit priori accuracy information of the IGS ultra-fast orbit product.The results show that compared with the traditional altitude angle stochastic model,the average GPS satellite clock accuracy estimated by the improved stochastic model is improved by 8.1%,and the estimation accuracy of all satellite clock is better than 0.15 ns.(2)To address the problem that the unknown coordinate errors of unstable monitoring stations directly affect the clock estimation accuracy,an extended GNSS clock estimation function model(the extended model)is proposed to identify the unstable stations by the station coordinate variations and perform real-time synchronous estimation for their coordinates.The experimental results show that when 5 cm error is added to some station coordinates,the average accuracy of satellite clock estimated by the extended model is improved by 24.7%compared with the traditional clock model,and the average accuracy is basically comparable compared with the reference clock product without station coordinate error,and the average difference between them is less than 0.005 ns.(3)A fast synthetical quality control method is proposed to address the problem of inefficient quality control in multi-GNSS real time clock estimation.The standard point positioning technique is used to remove the anomalous observations from the prior residuals,and then the post residuals of parameter estimation are used to remove the remaining gross observations.The experimental results show that the calculation time of the proposed method meets the real-time requirements,and the average GNSS clock estimation accuracy is 0.029 ns,which is 23.7% and 69.1% higher than the accuracy of the single prior and single post quality control methods,respectively.The real-time dynamic PPP 3D positioning accuracy of the proposed method improves by 4.8% and 31.8%,and the positioning convergence time decreases by 15.5% and 16.0%,respectively,compared with that of the single prior and single post quality control methods.