Distributed GNSS (Global Navigation Satellite System) collaborative pseudo-range differential positioning method suitable for vehicle-mounted terminal

Abstract

The invention discloses a distributed GNSS (Global Navigation Satellite System) collaborative pseudo-range differential positioning method suitable for a vehicle-mounted end, which comprises the following steps of: firstly, calculating a differential correction number error of a collaborative vehicle through a collaborative vehicle GNSS observation value, and constructing a differential correction number error prediction model and a training data set; pre-training the difference correction number error prediction model by using a random forest method; then, a pre-trained difference correction number error prediction model is utilized to predict and compensate difference correction number errors of different cooperative vehicles in the cooperative positioning system, and all difference correction numbers after error compensation are provided for a target vehicle; and finally, the target vehicle selects an optimal difference correction number, pseudo-range difference positioning is carried out according to the optimal difference correction number, and a final positioning result of the target vehicle is obtained. According to the method, the difference correction number error is predicted based on a random forest method, and the performance of the DDGNSS in a complex urban environment is further enhanced through a multi-dynamic-base-station cooperative positioning framework.

Description

Technical Field

[0001] The present invention relates to the field of satellite navigation technology, and in particular to a distributed GNSS collaborative pseudo-range differential positioning method suitable for a vehicle-mounted terminal. Background Art

[0002] The Differential Global Navigation Satellite System (DGNSS) is a widely used technology to improve GNSS positioning performance. Its fundamental principle is to use differential corrections broadcast by base stations to correct for common errors in pseudorange measurements. However, DGNSS still faces limitations in practical applications due to the number and distribution of base stations, communication range and reliability, and the increased incidence of non-common errors in urban environments. Therefore, with the rapid development of the Internet of Vehicles (IoV) and the Internet of Things (IoT), dynamic base station differential GNSS (DDGNSS) technology based on a distributed architecture has gained increasing attention. Unlike traditional DGNSS, which relies on fixed base stations, DDGNSS uses nearby vehicles or mobile devices as dynamic base stations to generate differential corrections and share them with nearby receivers. This differential information is used to correct pseudorange measurements, improving the positioning accuracy of the target receiver. This approach leverages widely distributed, low-cost receivers to broadcast differential information, reducing the need for fixed base station deployment and lowering system costs. Furthermore, the close-range communication between the dynamic base stations and receivers enhances communication stability. However, DDGNSS also faces new challenges. In particular, when the positioning accuracy of the dynamic base station itself is not high, the differential corrections it generates may introduce significant errors, affecting the positioning performance of the receiver.

[0003] Currently, methods for addressing positioning uncertainty inherent in dynamic base stations primarily include weighted elimination based on pseudorange residuals, multi-base station differential correction fusion, and model-driven error prediction. Weighted elimination based on pseudorange residuals quantifies the positioning uncertainty of dynamic base stations by statistically analyzing the residuals of pseudorange measurements, thereby eliminating pseudorange data with large errors. However, this method only addresses the overall error magnitude and lacks the ability to discern variations in satellite pseudorange errors in different directions, making it difficult to address the impact of nonlinear errors in complex environments. Multi-base station differential correction fusion utilizes measurements from multiple dynamic base stations and integrates the differential correction information through weighted averaging or fusion algorithms, thereby reducing the impact of differential correction errors from a single base station. However, since base station measurements may be correlated, ignoring this correlation can lead to over-convergence of the fusion results, further reducing correction accuracy. Model-driven error prediction uses statistical or physical models to analyze factors such as the distance between the base station and the receiver and the signal propagation path to predict and correct positioning errors. However, these methods often rely on assumptions about environmental conditions, limiting their performance in dealing with multipath effects and obstructions in complex urban environments.

[0004] In recent years, machine learning methods, due to their powerful nonlinear modeling capabilities, have become a powerful tool for addressing the uncertainty of dynamic base station differential positioning. Unlike traditional methods, machine learning directly learns the complex relationship between pseudorange errors and environmental characteristics from data, without relying on explicit models. It excels in handling nonlinear relationships, feature interactions, and model interpretability. It can more effectively handle complex environmental errors and, through model training, predict and compensate for differential correction errors. Comparing various machine learning methods, models such as random forest (RF) have shown advantages in real-time performance, robustness, and feature interpretability, making them well-suited for DDGNSS applications in urban environments. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a distributed GNSS collaborative pseudo-range differential positioning method suitable for vehicle-mounted terminals, predict the differential correction number error based on the random forest method, and further enhance the performance of DDGNSS in complex urban environments through a multi-dynamic base station collaborative positioning framework.

[0006] The technical solution of the present invention is: The distributed GNSS collaborative pseudo-range differential positioning method applicable to the vehicle terminal specifically includes the following steps: (1) First, the estimated value of the differential correction number is calculated by the pseudo-range measurement value in the cooperative vehicle GNSS observation value. At the same time, the reference value of the differential correction number is calculated based on the reference position of the cooperative vehicle. The estimated value of the differential correction number and the reference value are then differenced to obtain the differential correction number error of the cooperative vehicle. Then, a differential correction number error prediction model is constructed. A training data set is constructed according to the cooperative vehicle GNSS observation value, and the training data set is annotated by the calculated differential correction number error of the cooperative vehicle. The training data set is input into the differential correction number error prediction model, and the differential correction number error prediction model is pre-trained using the random forest method to generate a pre-trained differential correction number error prediction model. (2) Use the pre-trained differential correction error prediction model to predict and compensate the differential correction errors of different cooperative vehicles in the collaborative positioning system, and provide the differential correction numbers after all errors are compensated to the target vehicle; (3) The target vehicle selects the optimal differential correction number and performs pseudo-range differential positioning based on the optimal differential correction number to obtain the final positioning result of the target vehicle.

[0007] The GNSS observation values include real historical data of pseudorange measurement values, satellite altitude angles, satellite azimuth angles, signal-to-noise ratios and pseudorange residuals of cooperative vehicles in urban road environments. The estimated value of the differential correction number is calculated based on the pseudorange measurement values in the GNSS observation values, and the satellite altitude angles, satellite azimuth angles, signal-to-noise ratios and pseudorange residuals are used as feature values in the training data set.

[0008] The steps of using the pre-trained differential correction number error prediction model to predict and compensate for the differential correction number errors of different cooperative vehicles in the cooperative positioning system are as follows: S21, use the single point pseudo-range positioning method to locate the cooperative vehicle, first calculate the GNSS receiver on the cooperative vehicle With observation satellites Pseudorange measurements between , see the following formula (1) for details: (1); In formula (1), represents the speed of light, represents the clock offset of the observing satellite, Represents the clock offset of the GNSS receiver, represents the ionospheric delay, represents the tropospheric delay, represents the error caused by multipath or cross-link interference and receiver noise effects, Representative observation satellite With GNSS receiver The distance between The calculation of is shown in the following formula (2): (2); In formula (2), 、 、 are the coordinates of the observing satellite, 、 、 are the coordinates of the GNSS receiver; Then calculate the differential correction number of the cooperative vehicle , the calculation process is shown in the following formula (3): (3); In formula (3), Representative measurement satellite The distance between the positioning result of the collaborative vehicle and The calculation of is shown in the following formula (4): (4); In formula (4), 、 、 is the coordinate of the collaborative vehicle positioning result; S22, extract the characteristic values from the GNSS observation values of the cooperative vehicle and construct a test data set, input the test data set into the pre-trained differential correction number error prediction model for prediction, and obtain the prediction error of the differential correction number , the differential correction number after compensation The calculation process is shown in the following formula (5): (5); In formula (5), represents the differential correction number of the cooperative vehicle calculated by equation (3).

[0009] The target vehicle selects the optimal differential correction number, and performs pseudo-range differential positioning according to the optimal differential correction number to obtain the final positioning result of the target vehicle in the following specific steps: S31, the different cooperative vehicles in the cooperative positioning system will compensate the difference correction number after the error All are sent to the target vehicle in real time, and the target vehicle calculates the current epoch through the sum of squares of pseudo-range residuals The characteristic value of the positioning accuracy of each cooperative vehicle , eigenvalue The calculation of is shown in the following formula (6): (6); In formula (6), Represents the current epoch of the cooperative vehicle The sum of squares of pseudorange residuals under Represents the last epoch of the cooperative vehicle The sum of squares of pseudorange residuals under ; S32, compare the eigenvalues of different cooperative vehicles in the cooperative positioning system, select the cooperative vehicle corresponding to the minimum eigenvalue, and use the error-compensated differential correction number sent by this cooperative vehicle as the As the optimal differential correction number, pseudo-range differential positioning is performed according to the optimal differential correction number to obtain the final positioning result of the target vehicle.

[0010] Advantages of the present invention: (1) The present invention uses a dynamic base station (cooperative vehicle) to replace the traditional fixed reference station, which is suitable for a wide-area pseudo-range differential positioning system for vehicles or mobile devices, effectively expanding the scope of application of differential positioning while reducing system costs.

[0011] (2) The present invention predicts the differential correction number error based on the random forest method, and predicts and compensates the differential correction number errors of different cooperative vehicles in the cooperative positioning system through the GNSS observation values of the cooperative vehicles, providing more accurate pseudo-range correction information for the target vehicle, thereby enhancing the reliability and accuracy of the positioning system.

[0012] (3) Since the availability of dynamic base stations is easily affected by complex environments, the availability of pseudo-range differential positioning of the target vehicle is generally improved by increasing the number of dynamic base stations, that is, using multiple cooperative vehicles; for the fusion of differential information from multiple dynamic base stations, existing methods usually rely on weighted averaging or interpolation to merge differential corrections from different dynamic base stations; however, for dynamic base stations with poor performance and high positioning uncertainty, their clock deviations are difficult to predict; when the clock deviations of different dynamic base stations cannot be effectively alleviated, their differential corrections cannot be fused, so the present invention calculates the characteristic value of the positioning accuracy of each dynamic base station (cooperative vehicle) through the pseudo-range residual sum of squares (PRSS), selects the optimal differential correction number based on the characteristic value, and performs pseudo-range differential positioning based on the optimal differential correction number, thereby avoiding the influence of the differential correction number with large error on the positioning accuracy of the target vehicle, and improving the positioning accuracy and availability of the target vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 This is a schematic diagram of the impact of cooperative vehicle positioning uncertainty on the pseudo-range differential correction number it generates.

[0014] Figure 2 It is a flow chart of the present invention.

[0015] Figure 3 This is a diagram of the architecture of the differential correction number error pre-training and test data set error compensation of the present invention.

[0016] Figure 4 This is a flow chart of selecting the optimal differential correction number of the present invention. DETAILED DESCRIPTION

[0017] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0018] See Figures 1-4 , a distributed GNSS collaborative pseudo-range differential positioning method applicable to vehicle-mounted terminals, specifically including the following steps: (1) First, the estimated value of the differential correction number is calculated by the pseudo-range measurement value in the cooperative vehicle GNSS observation value. At the same time, the reference value of the differential correction number is calculated based on the reference position of the cooperative vehicle. The estimated value of the differential correction number and the reference value are then differenced to obtain the differential correction number error of the cooperative vehicle. Then, a differential correction number error prediction model is constructed. The satellite altitude angle, satellite azimuth angle, signal-to-noise ratio and pseudo-range residual of the cooperative vehicle GNSS observation value are used as the feature values in the training data set to construct a training data set. The training data set is annotated by the calculated differential correction number error of the cooperative vehicle. The training data set is input into the differential correction number error prediction model. The differential correction number error prediction model is pre-trained using the random forest method (RF) to generate a pre-trained differential correction number error prediction model. The random forest method is used to pre-train the differential correction error prediction model. The training data set constitutes the input space, and its area is recursively divided into two sub-regions 、 , the output value of each sub-region constructs a binary decision tree, and the optimal feature The corresponding optimal threshold The following formula (7) is used to obtain: (7); In formula (7), Represents the true regression target value of the sample; Pair Features Iterate and scan features The threshold value is selected and Minimized pair , using the selected feature-threshold pair to divide the region into and , and determine the corresponding output value, the two sub-regions repeat the feature-threshold selection steps until the stopping condition is met, so that the input space is divided into regions , thus generating a decision tree; Random forest is composed of randomly selected Decision trees form a set. For the A decision tree, querying data points The predicted value of Indicates that Is an independent decision tree The value of from Random selection Features, The prediction value of a decision tree is in the form of the following formula (8): (8); In formula (8), is the set of data points selected before building the decision tree, is a data point The collection of is included in The number of data points in ; Random forest is a combination of a large number of decision trees and the average value of the decision trees that make up the random forest. The rule form of random forest is as follows (9): (9); In formula (9), represents the number of decision trees in the random forest, The larger the value, the better the performance of random forest, but the computational cost will also increase. Therefore, when using A comprehensive balance should be made when the value is obtained. Once the rules of random forest are obtained , we can use the new GNSS measurements of cooperative vehicles in urban road environments to obtain the prediction error of the differential correction number ; (2) Use the pre-trained differential correction error prediction model to predict and compensate the differential correction errors of different cooperative vehicles in the collaborative positioning system, and provide the differential correction numbers after all errors are compensated to the target vehicle. The specific steps are as follows: S21, use the single point pseudo-range positioning method to locate the cooperative vehicle, first calculate the GNSS receiver on the cooperative vehicle With observation satellites Pseudorange measurements between , see the following formula (1) for details: (1); In formula (1), represents the speed of light, represents the clock offset of the observing satellite, Represents the clock offset of the GNSS receiver, represents the ionospheric delay, represents the tropospheric delay, represents the error caused by multipath or cross-link interference and receiver noise effects, Representative observation satellite With GNSS receiver The distance between The calculation of is shown in the following formula (2): (2); In formula (2), 、 、 are the coordinates of the observing satellite, 、 、 are the coordinates of the GNSS receiver; Then calculate the differential correction number of the cooperative vehicle , the calculation process is shown in the following formula (3): (3); In formula (3), Representative measurement satellite The distance between the positioning result of the collaborative vehicle and The calculation of is shown in the following formula (4): (4); In formula (4), 、 、 is the coordinate of the collaborative vehicle positioning result; S22, extract the characteristic values from the GNSS observation values of the cooperative vehicle and construct a test data set, input the test data set into the pre-trained differential correction number error prediction model for prediction, and obtain the prediction error of the differential correction number , the differential correction number after compensation The calculation process is shown in the following formula (5): (5); In formula (5), represents the differential correction number of the cooperative vehicle calculated by equation (3); (3) The target vehicle selects the optimal differential correction number and performs pseudo-range differential positioning based on the optimal differential correction number to obtain the final positioning result of the target vehicle. The specific steps are as follows: S31, the different cooperative vehicles in the cooperative positioning system will compensate the difference correction number after the error All are sent to the target vehicle in real time, and the target vehicle calculates the current epoch through the sum of squares of pseudo-range residuals The characteristic value of the positioning accuracy of each cooperative vehicle , eigenvalue The calculation of is shown in the following formula (6): (6); In formula (6), Represents the current epoch of the cooperative vehicle The sum of squares of pseudorange residuals under Represents the last epoch of the cooperative vehicle The sum of squares of pseudorange residuals under ; S32, compare the eigenvalues of different cooperative vehicles in the cooperative positioning system, select the cooperative vehicle corresponding to the minimum eigenvalue, and use the error-compensated differential correction number sent by this cooperative vehicle as the As the optimal differential correction number, pseudo-range differential positioning is performed according to the optimal differential correction number to obtain the final positioning result of the target vehicle.

[0019] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A distributed GNSS collaborative pseudo-range differential positioning method suitable for vehicle-mounted terminals, characterized by: The specific steps include: (1) First, the estimated value of the differential correction number is calculated by the pseudo-range measurement value in the cooperative vehicle GNSS observation value. At the same time, the reference value of the differential correction number is calculated based on the reference position of the cooperative vehicle. The estimated value of the differential correction number and the reference value are then differenced to obtain the differential correction number error of the cooperative vehicle. Then, a differential correction number error prediction model is constructed. A training data set is constructed according to the cooperative vehicle GNSS observation value, and the training data set is annotated by the calculated differential correction number error of the cooperative vehicle. The training data set is input into the differential correction number error prediction model, and the differential correction number error prediction model is pre-trained using the random forest method to generate a pre-trained differential correction number error prediction model. (2) Use the pre-trained differential correction error prediction model to predict and compensate the differential correction errors of different cooperative vehicles in the collaborative positioning system, and provide the differential correction numbers after all errors are compensated to the target vehicle; (3) The target vehicle selects the optimal differential correction number and performs pseudo-range differential positioning based on the optimal differential correction number to obtain the final positioning result of the target vehicle.

2. The distributed GNSS collaborative pseudo-range differential positioning method applicable to a vehicle-mounted terminal according to claim 1, characterized in that: The GNSS observation values include real historical data of pseudorange measurement values, satellite altitude angles, satellite azimuth angles, signal-to-noise ratios and pseudorange residuals of cooperative vehicles in urban road environments. The estimated value of the differential correction number is calculated based on the pseudorange measurement values in the GNSS observation values, and the satellite altitude angles, satellite azimuth angles, signal-to-noise ratios and pseudorange residuals are used as feature values in the training data set.

3. The distributed GNSS collaborative pseudo-range differential positioning method applicable to a vehicle-mounted terminal according to claim 2, characterized in that: The steps of using the pre-trained differential correction number error prediction model to predict and compensate for the differential correction number errors of different cooperative vehicles in the cooperative positioning system are as follows: S21, use the single point pseudo-range positioning method to locate the cooperative vehicle, first calculate the GNSS receiver on the cooperative vehicle With observation satellites Pseudorange measurements between , see the following formula (1) for details: （1）； In formula (1), represents the speed of light, represents the clock offset of the observing satellite, Represents the clock offset of the GNSS receiver, represents the ionospheric delay, represents the tropospheric delay, represents the error caused by multipath or cross-link interference and receiver noise effects, Representative observation satellite With GNSS receiver The distance between The calculation of is shown in the following formula (2): （2）； In formula (2), 、 、 are the coordinates of the observing satellite, 、 、 are the coordinates of the GNSS receiver; Then calculate the differential correction number of the cooperative vehicle , the calculation process is shown in the following formula (3): （3）； In formula (3), Representative measurement satellite The distance between the positioning result of the collaborative vehicle and The calculation of is shown in the following formula (4): （4）； In formula (4), 、 、 is the coordinate of the collaborative vehicle positioning result; S22, extract the characteristic values from the GNSS observation values of the cooperative vehicle and construct a test data set, input the test data set into the pre-trained differential correction number error prediction model for prediction, and obtain the prediction error of the differential correction number , the differential correction number after compensation The calculation process is shown in the following formula (5): （5）； In formula (5), represents the differential correction number of the cooperative vehicle calculated by equation (3).

4. The distributed GNSS collaborative pseudo-range differential positioning method applicable to a vehicle-mounted terminal according to claim 3, characterized in that: The target vehicle selects the optimal differential correction number, and performs pseudo-range differential positioning according to the optimal differential correction number to obtain the final positioning result of the target vehicle in the following specific steps: S31, the different cooperative vehicles in the cooperative positioning system will compensate the difference correction number after the error All are sent to the target vehicle in real time, and the target vehicle calculates the current epoch through the sum of squares of pseudo-range residuals The characteristic value of the positioning accuracy of each cooperative vehicle , eigenvalue The calculation of is shown in the following formula (6): （6）； In formula (6), Represents the current epoch of the cooperative vehicle The sum of squares of pseudorange residuals under Represents the last epoch of the cooperative vehicle The sum of squares of pseudorange residuals under ; S32, compare the eigenvalues of different cooperative vehicles in the cooperative positioning system, select the cooperative vehicle corresponding to the minimum eigenvalue, and use the error-compensated differential correction number sent by this cooperative vehicle as the As the optimal differential correction number, pseudo-range differential positioning is performed according to the optimal differential correction number to obtain the final positioning result of the target vehicle.

Images