A GPS receiver information fusion sharing method

Abstract

This application discloses a GPS receiver information fusion and sharing method, relating to the field of satellite navigation and positioning technology. The method includes: pre-setting a master receiver and a slave receiver to collect raw observation data respectively, forming a joint observation dataset; identifying a common-view satellite set; calculating the pseudorange difference measured by the two receivers, pre-setting a pseudorange threshold, and removing observation data with a pseudorange difference greater than the pseudorange threshold to generate a high-quality observation dataset; calculating the relative clock bias and relative clock bias change rate in the common-view satellite set to obtain a fused observation dataset; constructing a joint observation dataset through time alignment and cross-validation to achieve deep fusion at the raw data level; effectively smoothing single-machine observation noise and solving the heterogeneous clock fusion problem; accurately estimating the relative clock bias and its change rate, dynamically selecting the optimal working mode based on the real-time environmental state to prevent filter divergence, and maintaining usable positioning output through information complementarity or historical data.

Description

Technical Field

[0001] This invention relates to the field of satellite navigation and positioning technology, and in particular to a method for information fusion and sharing of GPS receivers. Background Technology

[0002] In complex environments such as urban canyons, tree-lined roads, and tunnel entrances and exits, the positioning accuracy of Global Navigation Satellite Systems (GNSS) often decreases or even becomes interrupted due to signal blockage and multipath effects.

[0003] Existing technical solutions mostly remain at a shallow fusion level, exhibiting significant shortcomings. A common approach is result-level optimization, which involves running two independent positioning and calculation modules in parallel and simply comparing and selecting the final outputs such as position and velocity, for example, choosing the result with a better positioning accuracy factor or more satellites. This approach fails to utilize the complementary and constraining relationships between the original observation data, resulting in low information utilization. When both results are affected by varying degrees of interference, it cannot achieve a result superior to either single-machine fusion through deep fusion. Another approach is to perform a simple weighted average of observations, such as averaging pseudorange observations of the same satellite from two receivers. While this method utilizes the original data, it ignores a core challenge of dual-receiver systems: the two receivers typically use independent crystal oscillators, whose clocks differ and drift. Directly mixing observations from different time bases introduces systematic bias, potentially reducing accuracy.

[0004] Furthermore, existing solutions generally lack sophisticated evaluation and robustness mechanisms for observation data quality. In complex environments, individual satellite observations may deviate significantly from the true value due to severe multipath or transient interference. Simple averaging or selection strategies cannot effectively identify and remove these abnormal data, leading to skewed overall solution results and poor system robustness. Existing dual-receiver solutions fail to achieve deep fusion at the observation data level, are insufficient in handling independent clock differences, and have limited robustness, thus restricting the substantial improvement of positioning accuracy and reliability in harsh scenarios. Summary of the Invention

[0005] This application provides a GPS receiver information fusion and sharing method, which solves the problems of shallow information fusion level and ineffective handling of independent clock differences in existing dual-receiver systems. It realizes deep fusion at the level of raw observation data and adaptive fusion positioning with high-precision relative clock difference compensation, and achieves the technical effect of further improving positioning accuracy in signal obstruction and multipath environments.

[0006] This application provides a GPS receiver information fusion and sharing method, including:

[0007] S1: Pre-set the main receiver and the secondary receiver to collect raw observation data respectively, align all raw observation data in time, calculate the quality factor of the raw observation data collected by each receiver at the same time, compare them, and filter out the raw observation data with the quality factor greater than the rejection threshold to form a joint observation dataset;

[0008] S2: Based on the joint observation dataset, select satellites that are simultaneously monitored by the main receiver and the secondary receiver to form a common-view satellite set; for each satellite in the set, calculate the pseudorange difference measured by the two receivers, pre-set the pseudorange threshold, and remove observation data with pseudorange difference greater than the pseudorange threshold to generate a high-quality observation dataset.

[0009] S3: Calculate the relative clock bias and relative clock bias change rate in the common-view satellite set, correct the observation data of all secondary receivers in the high-quality observation dataset, and obtain the fused observation dataset;

[0010] S4: Select the working mode based on the current environmental status and pattern rules, perform localization by combining the fused observation dataset, and output the localization results and confidence index.

[0011] Furthermore, the mode rules are set as follows: based on the total number of effective satellites and the horizontal accuracy factor value, if the total number of effective satellites is less than four, the observation complementarity mode is adopted; if the total number of effective satellites is greater than or equal to four and the horizontal accuracy factor is less than a preset threshold, the tightly coupled fusion mode is adopted; if the horizontal accuracy factor is greater than or equal to a preset threshold, the weighted least squares fusion mode is adopted; if neither receiver can locate, the cooperative acquisition mode is enabled.

[0012] Furthermore, the raw observation data includes satellite number, pseudorange, carrier phase, Doppler shift, signal-to-noise ratio, and the receiver's own local timestamp; each row in the joint observation dataset contains receiver identifier, time identifier, satellite identifier, and raw observation data.

[0013] Further, calculate the quality factor: ,in, For quality factor; Map the signal-to-noise ratio to the [0,1] interval. It is the ratio of signal carrier power to noise power spectral density, i.e., signal-to-noise ratio, and the unit is dB-Hz. If it is below 30dB-Hz, it is 0, and if it is above 50dB-Hz, it is 1. The satellite's elevation angle; The number of epochs for continuous tracking refers to the total number of valid observation epochs successfully acquired by the receiver during continuous tracking of a satellite. An epoch refers to a specific moment corresponding to the observation data, which serves as a time reference point. , , These are the weighting coefficients, with initial values ​​set to 0.5, 0.3, and 0.2 respectively. Satellite observation data with a quality factor less than 0.4 are removed, and the observation signals are filtered based on the quality factor to generate the effective observation set of the main receiver and the effective observation set of the secondary receiver, which together constitute the joint observation dataset.

[0014] Furthermore, the pseudorange difference between the two receiver observations is calculated, including: for the same satellite i, the pseudorange observation equations for the two receivers R1 and R2 are:

[0015]

[0016]

[0017] in, These are the actual geometric distances between receivers R1 and R2 and satellite i, respectively. The distance error is caused by the receiver clock difference; It is ionospheric delay; It is a tropospheric delay; It is a multipath effect; It is receiver thermal noise; calculate pseudorange difference. .

[0018] Furthermore, the relative clock bias and relative clock bias rate of change in the common-view satellite ensemble are calculated, including: differentiating the clock biases of consecutive epochs to obtain the clock bias rate of change. The rate of change of relative clock bias is obtained by differentiating the relative clock bias over consecutive epochs. The relative clock bias estimate is obtained by averaging across all common-view satellites.

[0019]

[0020] Where C is the set of satellites viewed together, and i represents the i-th satellite.

[0021] Furthermore, the fused observation dataset is obtained, including: defining the state vector. It includes three-dimensional position, three-dimensional velocity, master receiver clock bias and clock bias rate of change, relative clock bias and relative clock bias rate of change; the observation vector Z is composed of pseudoranges from all effective satellites, and the observation equation is: ,in, It is a nonlinear observation function. To reduce observation noise, a fusion observation equation is constructed, and the observation data from the secondary receiver is corrected to the time reference of the primary receiver using clock bias. The corrected pseudorange is: Where t is the current epoch time, This is used to estimate the reference time for clock bias.

[0022] Furthermore, for satellite i of the main receiver, the nonlinear observation function is: For satellite j of the secondary receiver, the nonlinear observation function is: .

[0023] Furthermore, the method also includes: for each co-view satellite, performing carrier phase smoothing processing on the raw pseudorange of the primary receiver and the secondary receiver respectively to obtain smoothed pseudorange. :

[0024]

[0025] in, The smoothing time constant, For carrier wavelength, The carrier phase change is used to calculate the smoothed pseudorange difference. Combined with the original pseudorange difference As an auxiliary criterion, high-quality observation datasets are selected.

[0026] Furthermore, the confidence level P is calculated using the following formula:

[0027]

[0028] in, The number of valid satellites; The maximum number of available satellites is set to 12. The threshold is used; age is the delay in the positioning result. It is the attenuation factor; The weights are 0.4, 0.3, and 0.3 respectively, initially; the confidence level P∈[0,1], P>0.7 indicates high confidence and can be output; when P≤0.7, it prompts the user that the current positioning accuracy is low.

[0029] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0030] By performing time alignment, quality screening, and cross-validation on the raw pseudorange, carrier phase, and other observation data from the two receivers, a unified joint observation dataset is constructed, achieving deep fusion from the result level to the raw data level. Through tight-coupled filtering and solution using a state-space model, the observation information of the two receivers on all visible satellites can be mutually constrained and jointly contribute to the final state estimation. It can effectively smooth the noise of single-machine observations and supplement and enhance the observations of the other receiver when the number of satellites observed by a single machine is insufficient or the geometric configuration is poor.

[0031] The problem of heterogeneous clock fusion is solved by estimating and compensating for the relative clock difference between the two receivers. The clock difference between the two independent receivers is clearly identified and modeled. By calculating the pseudorange difference of their common-view satellites, the relative clock difference and its rate of change can be accurately estimated. All observation data from the secondary receiver are uniformly corrected to the time reference of the primary receiver, overcoming the shortcomings of existing technologies that ignore clock differences or process them coarsely.

[0032] A quality factor is constructed by integrating multiple factors such as signal-to-noise ratio, satellite elevation angle, and continuous tracking time to initially screen raw observations. Further cross-validation of observations from the same satellite using dual receivers effectively identifies and eliminates anomalous data caused by localized interference such as severe multipath interference. Chi-square tests using the innovation sequence can detect and eliminate gross errors in real time. The optimal operating mode is dynamically selected based on real-time environmental conditions to prevent filter divergence when satellite geometry deteriorates, maintaining usable positioning output through information complementarity or historical data. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of a GPS receiver information fusion and sharing method according to an embodiment of the present invention. Detailed Implementation

[0034] To facilitate understanding of the present invention, a more complete description of this application will be given below with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to enable a more thorough and complete understanding of the disclosure of the present invention.

[0035] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains; the terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to limit the invention; the term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0036] This application utilizes a dual-receiver system consisting of a main receiver, a secondary receiver, and a serial communication module. Both the main and secondary receivers employ multi-frequency receiving chips supporting both BeiDou and GPS systems. The main receiver has an embedded ARM processor responsible for data fusion. Each receiver is connected to an independent antenna and exchanges raw observations and positioning results via a custom binary protocol through the serial port. The core function of the receiver (GPS receiver / GNSS receiver) is to receive electromagnetic wave signals transmitted by satellites through the antenna, process and calculate these signals, and ultimately output information such as position, velocity, and time.

[0037] Example 1: As Figure 1 As shown, a GPS receiver information fusion and sharing method includes:

[0038] S1: The master receiver and the slave receiver are pre-set to collect raw observation data respectively. All raw observation data are time-aligned. The quality factor of the raw observation data collected by each receiver at the same time is calculated and compared. The raw observation data with a quality factor greater than the rejection threshold are selected to form a joint observation dataset.

[0039] In this embodiment, after receiving the satellite signal, the receiver tracks, locks onto, and decodes the signal, outputting the raw observation data as digital data. The main receiver and the secondary receiver each collect the raw observation data at a fixed frequency, including: satellite number, pseudorange, carrier phase, Doppler shift, signal-to-noise ratio (SNR), and the receiver's own local timestamp. The satellite number identifies which satellite the signal originates from. Pseudorange measures the time it takes for the satellite signal to travel from the satellite to the receiver, multiplied by the speed of light, obtained by comparing the phase difference between the ranging code emitted by the satellite and the ranging code copied locally by the receiver. Carrier phase measures the difference between the phase of the satellite carrier signal received by the receiver and the phase of the carrier signal generated locally by the receiver, with an accuracy 2-3 orders of magnitude higher than pseudorange. Doppler shift measures the change in carrier frequency caused by the relative motion between the receiver and the satellite. The signal-to-noise ratio (SNR) measures the ratio of signal power to noise power spectral density, reflecting signal quality.

[0040] Both parties send raw observation data packets (including timestamps) to each other via serial port. The data packet format uses a custom binary protocol and includes a frame header, data type, timestamp, data payload, and checksum. After the master receiver collects data from both receivers, time alignment is required because the two receivers have independent clocks and timestamps that deviate from each other. Using the master receiver's time as a reference, the slave receiver's timestamp and serial port transmission delay are used to align the slave receiver's data to the master receiver's timeline. All raw observation data are time-aligned. The quality factor is calculated for the raw observation data collected by each receiver at the same time, and these factors are compared. Raw observation data with a quality factor greater than the rejection threshold are selected to form a joint observation dataset.

[0041] In satellite navigation, the smallest unit of data output by the receiver is the raw observation data for each satellite. The quality factor is determined based on several factors: signal strength (assessing the satellite's signal strength; if the signal is too weak (a baseline threshold is set based on historical observation data; if it falls below this threshold, it's considered weak), the receiver may not be able to hear it clearly, leading to errors in measurement data); geometric position (assessing whether the satellite's position in the sky is optimal for observation; if the satellite is too low (near the horizon), the signal has a long path through the atmosphere, resulting in larger errors and susceptibility to ground reflection interference); and lock-on status (assessing the duration of the receiver's monitoring of the satellite; if the satellite has just been acquired, data instability increases). A high Q value indicates high reliability of the receiver's measurements of that satellite and is given a greater contribution to the final positioning result.

[0042] The joint observation dataset is a structured collection of data, where each smallest data unit represents an observation of a satellite by a specific receiver at a specific time. Each row (or element) in the joint observation dataset should contain the following fields: Receiver identifier: used to distinguish whether the data comes from the primary or secondary receiver; Time identifier: a unified timestamp after time alignment; Satellite identifier: which satellite was observed; Selected raw observation data: the core measurement data output by the receiver, with one raw observation data corresponding to the same satellite at the same time.

[0043] Preferably, the entire dataset is a time series. The outermost structure is arranged in ascending order according to a unified timestamp. At the same point in time, the data is arranged according to the following logic: First, group by receiver. First, list all observations from the primary receiver at this time, then list all observations from the secondary receivers at this time. Second, sort by satellite number. Within each receiver group, sort by satellite number in ascending order for easy searching and comparison.

[0044] In some embodiments, the quality factor Q is calculated: ,in, For quality factor; Map the signal-to-noise ratio to the [0,1] interval. It is the ratio of signal carrier power to noise power spectral density, i.e., signal-to-noise ratio, and the unit is dB-Hz. If it is below 30dB-Hz, it is 0, and if it is above 50dB-Hz, it is 1. The satellite's elevation angle; The number of epochs for continuous tracking refers to the total number of valid observation epochs successfully acquired by the receiver during the continuous tracking of a satellite. An epoch refers to a specific moment corresponding to the observation data. As a time reference point, the receiver samples the satellite signal once at fixed time intervals, and one sampling point is one epoch. , , These are weighting coefficients, with initial values ​​set to 0.5, 0.3, and 0.2 respectively. A pre-set rejection threshold is used, set to 0.4 in this embodiment, meaning satellite observation data with a quality factor less than 0.4. The observation signals are then filtered based on the quality factor to generate a valid observation set. (Main receiver) and (Secondary receivers) together constitute the joint observation dataset.

[0045] In this application, the signal-to-noise ratio (SNR) threshold of 30 dB-Hz is a recognized tracking threshold in the GNSS receiver field. Below 30 dB-Hz, it is difficult for the receiver's carrier tracking loop and code tracking loop to maintain lock, and lock loss is likely. At this point, the measurement noise is extremely high, and even cycle slips or gross errors may occur. 50 dB-Hz is a typical high SNR level in open environments. When 50 dB-Hz is reached, the signal quality is excellent, and the measurement error caused by thermal noise is negligible. Linear interpolation is performed between 30 and 50 dB-Hz, assuming that the SNR is linearly positively correlated with the quality within this range. The weighting coefficients are assigned based on the analytic hierarchy process (AHP) or expert experience. Since the SNR is a key factor determining the random error of the current epoch, the initial weight is set to 0.5. The elevation angle determines the lower bound of systematic errors (atmospheric delay, multipath), and the initial weight is set to 0.3. Tracking continuity reflects historical reliability, and its impact can be partially compensated for by state transitions in subsequent filtering; therefore, it has the lowest weight. In practical engineering, the weight allocation can be obtained through fitting and optimization using a large amount of measured data. The quality rejection threshold of 0.4 is an empirical threshold, meaning that the observation data must simultaneously possess a moderate to high signal-to-noise ratio, a not too low elevation angle, and a certain tracking history to be considered reliable. Below the quality rejection threshold, the observation data has excessive noise or too high error uncertainty, making it unsuitable for high-precision fusion filtering.

[0046] Under ideal conditions with no multipath and atmospheric delay, the measurement error of civilian single-frequency pseudorange is typically around 3-5 meters (1σ). In environments with slight multipath, the pseudorange error may increase to 10-15 meters. The verification threshold for pseudorange difference is set at 10 meters. If the difference is no greater than 10 meters, the measurements from both receivers are considered to be within the normal error range and without severe multipath or abnormal interference, and can be retained. If the difference is greater than 10 meters, it exceeds the normal pseudorange error range, indicating that at least one receiver has experienced severe interference or an uncorrected cycle slip, and all observation data from both receivers for that satellite are discarded.

[0047] S2: Based on the joint observation dataset, select satellites that are simultaneously monitored by the primary and secondary receivers to form a common-view satellite set C. For each satellite in the set, calculate the pseudorange difference between the two receivers, pre-set a pseudorange threshold, and remove observation data with pseudorange differences greater than the pseudorange threshold to generate a high-quality observation dataset.

[0048] At a given moment, the same satellite is observed simultaneously by both the primary and secondary receivers. Although the antennas of the two receivers are positioned differently, they are relatively close (like the front and rear of a car). Therefore, their observation geometry of the same satellite in space is very similar. Calculating the pseudorange difference for the co-observed satellite can eliminate common errors and isolate difference errors. Cross-validation is performed using the redundancy of the dual receivers. For co-observed satellites, the pseudorange difference is calculated. Based on historical data analysis, the pseudorange threshold was set to 10 meters. If the measurement is less than 1 meter, then at least one observational anomaly is considered, and that satellite's data from both sides is removed; if The rice is retained and used for subsequent fusion.

[0049] In some embodiments, preferably, for each common-view satellite, the original pseudorange of the primary receiver and the secondary receiver are subjected to carrier phase smoothing processing to obtain smoothed pseudorange. The smoothing algorithm can employ classic Hatch filtering:

[0050]

[0051] in, The smoothing time constant (e.g., 100 epochs). For carrier wavelength, This represents the carrier phase change. The smoothed pseudorange difference is calculated. Since smoothing reduces noise and improves the sensitivity of anomaly detection, the pseudorange threshold can be appropriately tightened, for example, set to 5 meters.

[0052] At the same time, the original pseudorange difference is preserved. As a supplementary criterion. If Miqie If the error exceeds the limit, the satellite observation is deemed abnormal and the satellite is removed. If only the smoothed difference exceeds the limit while the original difference does not, it may be due to cycle slips during the smoothing process, and further judgment is needed based on the cycle slip detection results.

[0053] In some embodiments, for the same satellite i, the pseudorange observation equations for two receivers R1 and R2 are:

[0054]

[0055]

[0056] in, These are the actual geometric distances between receivers R1 and R2 and satellite i, respectively. The distance error is caused by the receiver clock difference; It is ionospheric delay; It is a tropospheric delay; It is a multipath effect; This refers to receiver thermal noise. Subtracting the two formulas yields the geometric distance difference. Since the two antennas are very close (short baseline, e.g., 2 meters), this difference is very small (<2 meters) and can be accurately calculated using the known baseline vector. The ionospheric / tropospheric delay difference is negligible because the two antennas are almost in the same location, and the signal's path through the atmosphere is almost identical. Multipath difference and noise difference are also relevant. These are local errors unique to each of the two receivers. If one receiver suffers severe multipath interference, the multipath difference term will be large, leading to anomalies in the difference value. Calculating the pseudorange difference involves two steps: First, estimating the relative clock error by averaging the differences across all common-view satellites. This smooths out noise and minor geometrical differences, thus accurately estimating the relative clock error between the two receivers. Second, if a particular difference is significantly larger than the average level of other satellites, it indicates that the satellite has suffered severe local interference (mainly multipath) at one of the receivers. By calculating the difference, localized anomalous observations can be quickly identified and eliminated, preventing erroneous data from contaminating the fusion results.

[0057] S3: Calculate the relative clock bias and relative clock bias rate of change in the common-view satellite ensemble, correct the observation data from all secondary receivers in the high-quality observation dataset, and obtain the fused observation dataset. The clock bias rate of change is obtained by differentiating the clock biases of consecutive epochs. The rate of change of relative clock bias is obtained by differentiating the relative clock bias over consecutive epochs. The relative clock bias estimate is obtained by averaging across all common-view satellites:

[0058]

[0059] Where C is the set of satellites viewed together, and i represents the i-th satellite.

[0060] Define state vector It includes three-dimensional position, three-dimensional velocity, primary receiver clock bias and clock bias rate of change, relative clock bias and relative clock bias rate of change. The observation vector Z is composed of the pseudoranges of all effective satellites (including those unique to the primary receiver, those unique to the secondary receiver, and common-view satellites), and the observation equation is: ,in, It is a nonlinear observation function. For observation noise, when the primary receiver is on satellite i, the nonlinear observation function is: For satellite j of the secondary receiver, the nonlinear observation function is: A fusion observation equation is constructed, and the observation data from the secondary receiver is corrected to the time reference of the primary receiver using clock bias. The corrected pseudorange is: Where t is the current epoch time, This is used to estimate the reference time for clock bias.

[0061] By using the estimated relative clock bias and relative clock bias rate of change, the distance corresponding to the time deviation caused by the relative clock bias and relative clock bias rate of change is subtracted from the original pseudorange of the secondary receiver, thereby unifying the observations of the secondary receiver to the time reference of the primary receiver.

[0062] Extended Kalman filtering (EKF) is used for state estimation. The state transition matrix F and process noise covariance Q are dynamically set based on the actual motion. The observation noise covariance matrix R is a diagonal matrix, and the noise variance of each observation data point is... Determined by the following adaptive factor:

[0063]

[0064] in, The baseline variance is 1m². This is the receiver hardware quality factor (initially set to 1). It aims to give observations at low elevation angles and low signal-to-noise ratios a larger variance. In the Kalman filter update step, the innovation sequence is calculated. ,in, This is the actual observation vector at time k. In this invention, it is derived from the pseudorange observation data of all valid satellites (including the primary and secondary receivers) at the current epoch. Composition. This is the data we actually measured using hardware. It is a one-step prediction of the state, indicating that based on The optimal state estimate at time k is the system state (position, velocity, clock difference, etc.) predicted by the system's motion model (such as a constant velocity model). It is a nonlinear observation function that predicts the state. (For example, the predicted position and clock difference) are mapped to the observation space to obtain the predicted observation data. That is, the initial pseudorange is inferred from the predicted position. It's new information, the actual measured value. The difference between the observed and predicted values ​​represents the degree of inconsistency between the actual and predicted observations. If the system model is accurate and the observation data is normal, the innovation should be a noise sequence with a mean of 0. The fused observation dataset is obtained by correcting the observation data from all secondary receivers in the high-quality observation dataset using the fused observation equation and extended Kalman filtering.

[0065] The chi-square test for innovation is a technique for real-time monitoring of whether observed data is anomaly. If an observed data point is normal, its innovation should fall within a certain statistical confidence interval; if it falls outside the interval, it is considered an outlier (gross error) and is removed. The innovation corresponding to each observed data point is calculated according to the formula. .

[0066] In Kalman filtering, the variance of the innovation can be calculated. For the i-th observation, its innovation variance is extracted from the covariance matrix of the filter. To simplify the calculation, the variance of the observation noise is directly used as the measurement criterion. Compare the absolute value of the innovation with 3 times its standard deviation (the square root of the variance). If , it is considered that the observation is abnormal, and after removing it, the update is performed again.

[0067] After removing the outliers, the remaining observation data forms a new observation vector . The Kalman filter uses the new observation vector to re-execute the update step and calculate a more accurate state estimate. Automatically identify and remove severely contaminated observation data to prevent the positioning result from being biased, thus greatly improving the robustness of the positioning system in complex environments.

[0068] The filtered output obtains the state estimate value , that is, the position, velocity, and clock error parameters at the current moment. At the same time, the horizontal dilution of precision HDOP and the vertical dilution of precision VDOP are calculated to evaluate the geometric strength of the fusion result. HDOP and VDOP are indicators that measure the impact of satellite geometric distribution on positioning accuracy and reflect the quality of satellite distribution in space. When the satellites are evenly distributed in the sky, HDOP / VDOP is small and the positioning accuracy is high; otherwise, the positioning accuracy is low. They are both extracted from the state covariance matrix of the Kalman filter. Specifically: obtain the state estimate covariance matrix at the current moment k, which contains the variances and covariances of the estimation errors of all state variables (position, velocity, clock error, etc.); and perform coordinate system transformation to obtain the error variances in the (such as east, north, vertical directions) to form the covariance matrix of the position part, and calculate the horizontal dilution of precision and the vertical dilution of precision. According to the general satellite distribution regulations, when HDOP < 2: excellent satellite geometric distribution and high positioning accuracy; 2 < HDOP < 3: good geometric distribution, which can meet most navigation requirements; 3 < HDOP < 5: general geometric distribution and the positioning accuracy begins to decline. HDOP > 5: poor geometric distribution and the positioning result may be unreliable.

[0069] S4: Based on the current environmental conditions and dynamic pattern rules, the working mode is dynamically selected, and positioning is performed using the fused observation dataset. The final output includes the positioning result and confidence index, which includes longitude, latitude, altitude, velocity, and direction. This is to avoid fusion failure under extreme conditions. A threshold for the number of observed satellites is pre-set. Basic GNSS single-point positioning is achieved by solving a system of equations containing four unknowns: three-dimensional position coordinates (X, Y, Z) and receiver clock error equations. Requirement: An equation can be derived from the pseudorange observation data of each satellite. Therefore, to solve for the four unknowns, theoretically at least four independent equations are needed, meaning at least four satellites must be observed. Thus, a mode rule is defined to select the operating mode based on the current environmental state and the mode rule. Specifically, this includes: judging based on the total number of effective satellites and the horizontal accuracy factor (HDOP) value; if the total number of effective satellites is less than four, an observation complementarity mode is used; if the total number of effective satellites is greater than or equal to four and the HDOP is less than a preset threshold, a tightly coupled fusion mode is used; if the HDOP is greater than or equal to a preset threshold, a weighted least squares fusion mode is used; if neither receiver can locate, a cooperative acquisition mode is activated. Specifically, if the total number of effective satellites is less than four, the observation complementarity mode is entered, directly using the single-receiver positioning result after weighted average signal-to-noise ratio (prioritizing the primary receiver if its satellite count is ≥4, otherwise using the secondary receiver). If the total number of effective satellites is ≥4 and HDOP is less than 3, the tightly coupled fusion mode is maintained, i.e., fused positioning is used. If HDOP is ≥3, the weighted least squares fusion mode is entered, ignoring the dynamic model and only using the current epoch observations for weighted least squares calculation to avoid filter divergence. HDOP is a magnification factor that indicates how many times the pseudorange measurement error is amplified to become the final positioning error. Assuming a complex environment like an urban canyon, the pseudorange measurement error after robust processing is approximately 3-5 meters. If the satellite geometry is poor, for example, HDOP=5, the final horizontal positioning error could be as high as 5×3=15 meters. Such an error could even lead to incorrect lane identification. If the geometry is better, for example, HDOP<3, the maximum error can be controlled within 3×3=9 meters, which usually meets the needs of lane-level navigation (such as determining whether the vehicle is on a main road or a side road).

[0070] If neither receiver can locate (the number of valid satellites is less than 4), then the cooperative acquisition mode is activated. The ephemeris and Doppler information of the most recent successful location by the master receiver are sent to the slave receiver via serial port to assist it in quickly reacquiring. At the same time, the last valid position is output and marked as speculative navigation.

[0071] In some embodiments, the confidence level P is calculated using the following formula:

[0072]

[0073] in, For the number of effective satellites, The maximum number of available satellites (taken as 12). The threshold is 5, and age is the delay in the positioning result (in seconds). The attenuation factor is set to 0.5. These are the weights, initially 0.4, 0.3, and 0.3 respectively. The confidence level P∈[0,1], P>0.7 indicates high confidence and can be output; P≤0.7 prompts the user that the current positioning accuracy is low.

[0074] Compared with existing technologies, this invention has the following advantages: It directly utilizes the raw pseudorange observations from the dual receivers for tightly coupled Kalman filtering, which, compared to result-level selection or simple weighted averaging, fully leverages redundant information and significantly improves accuracy. Through relative clock bias estimation, it eliminates the influence of clock asynchrony between the two independent receivers, unifying the observation equations and resulting in more accurate fusion. By combining adaptive variance of signal-to-noise ratio and elevation angle with innovation testing, it effectively eliminates anomalous observations and improves robustness. It dynamically selects the optimal operating mode based on the environment, ensuring positioning continuity in complex environments.

[0075] In this embodiment, raw observation data containing pseudorange, carrier phase, and signal-to-noise ratio (SNR) information is synchronously acquired by the primary and secondary receivers. Time alignment is performed, and quality factors for integrated SNR, elevation angle, and tracking continuity are calculated for initial screening to construct a joint observation dataset. Next, satellites jointly observed by both receivers are selected, and cross-validation is performed by calculating and comparing their pseudorange differences to eliminate abnormal observations with excessive discrepancies, forming a high-quality observation dataset. Then, the relative clock bias and its rate of change between the two independent receivers are accurately estimated using the shared satellite observation data. This result is used to correct all secondary receiver observation data to the primary receiver's time reference, thereby achieving clock synchronization. Then, using position, velocity, clock bias, etc., as states and all synchronized observation data as input, a fusion observation equation is constructed. Extended Kalman filtering is used for state estimation. During the filtering process, different noise weights are assigned to observations of different qualities, and innovation checks are used to eliminate gross errors in real time, ultimately outputting a high-precision fusion positioning result and accuracy factor. Finally, based on the real-time effective number of satellites and the geometric accuracy factor, the system dynamically and intelligently switches between various working modes such as tightly coupled fusion, weighted least squares, observation complementarity, and collaborative acquisition, and outputs a confidence index calculated by comprehensively considering the number of satellites, geometric accuracy, and result freshness.

[0076] The technical means of this embodiment aim to systematically solve three core problems in dual-receiver systems: First, it addresses the problem of insufficient utilization of observation information and shallow fusion levels caused by traditional result selection or simple averaging, by mining the potential of redundant information through deep fusion at the raw data level. Second, it solves the problem of being unable to directly perform high-precision data fusion due to the independent and asynchronous clocks of the two receivers, by achieving the unification of heterogeneous data through relative clock difference estimation and compensation. Third, it solves the problem of abnormal observation data (such as multipath effects) easily contaminating the positioning results and causing poor system robustness in complex environments, by effectively isolating interference through multi-level quality assessment, cross-validation, and robust filtering mechanisms.

[0077] By implementing the above technical solutions, through deep tight-coupled filtering and clock synchronization processing, the positioning accuracy is significantly improved compared to single-receiver or simple dual-machine solutions, especially in scenarios where some satellites are blocked, where the advantages are even more pronounced. In terms of robustness, multiple data filtering and robustness mechanisms ensure that even when individual observations are severely disturbed, the overall system output remains stable and reliable, greatly reducing the risk of gross errors skewing the positioning results. Regarding continuity and availability, the adaptive multi-mode switching strategy guarantees seamless transition and continuous positioning output from open environments to scenarios with drastic signal changes, such as urban canyons and tunnel entrances, avoiding the problems of loss of lock or jumps that traditional solutions are prone to in harsh environments. It can output high-precision position and velocity information, as well as quantified confidence indicators.

[0078] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. For those skilled in the art, the present invention can have various modifications and variations. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A GPS receiver information fusion sharing method, characterized in that, include: S1: Pre-set the main receiver and the secondary receiver to collect raw observation data respectively, align all raw observation data in time, calculate the quality factor of the raw observation data collected by each receiver at the same time, compare them, and filter out the raw observation data with the quality factor greater than the rejection threshold to form a joint observation dataset; S2: Based on the joint observation dataset, select satellites that are simultaneously monitored by the main receiver and the secondary receiver to form a common-view satellite set; for each satellite in the set, calculate the pseudorange difference measured by the two receivers, pre-set the pseudorange threshold, and remove observation data with pseudorange difference greater than the pseudorange threshold to generate a high-quality observation dataset. S3: Calculate the relative clock bias and relative clock bias change rate in the common-view satellite set, correct the observation data of all secondary receivers in the high-quality observation dataset, and obtain the fused observation dataset; S4: Select the working mode based on the current environmental status and pattern rules, perform localization by combining the fused observation dataset, and output the localization results and confidence index.

2. The method as described in claim 1, characterized in that, Calculate the quality factor: ,in, For quality factor; Map the signal-to-noise ratio to the [0,1] interval. It is the ratio of signal carrier power to noise power spectral density, i.e., signal-to-noise ratio, and the unit is dB-Hz. If it is below 30dB-Hz, it is 0, and if it is above 50dB-Hz, it is 1. The satellite's elevation angle; The number of epochs for continuous tracking refers to the total number of valid observation epochs successfully acquired by the receiver during continuous tracking of a satellite. An epoch refers to a specific moment corresponding to the observation data, which serves as a time reference point. , , These are the weighting coefficients, with initial values ​​set to 0.5, 0.3, and 0.2 respectively. Satellite observation data with a quality factor less than 0.4 are removed, and the observation signals are filtered based on the quality factor to generate the effective observation set of the main receiver and the effective observation set of the secondary receiver, which together constitute the joint observation dataset.

3. The method as described in claim 1, characterized in that, Calculate the pseudorange difference between the two receiver observations, including: For the same satellite i, the pseudorange observation equations for the two receivers R1 and R2 are as follows: ； ； in, These are the actual geometric distances between receivers R1 and R2 and satellite i, respectively. The distance error is caused by the receiver clock difference; It is ionospheric delay; It is a tropospheric delay; It is a multipath effect; It is receiver thermal noise; calculate pseudorange difference. .

4. The method as described in claim 1, characterized in that, Calculating the relative clock bias and relative clock bias rate of change in a common-view satellite ensemble includes: differencing the clock bias over consecutive epochs to obtain the clock bias rate of change. The rate of change of relative clock bias is obtained by differentiating the relative clock bias over consecutive epochs. The relative clock bias estimate is obtained by averaging across all common-view satellites. ； Where C is the set of satellites viewed together, and i represents the i-th satellite.

5. The method as described in claim 1, characterized in that, The fused observation dataset is obtained, including: defining the state vector. It includes three-dimensional position, three-dimensional velocity, master receiver clock bias and clock bias rate of change, relative clock bias and relative clock bias rate of change; the observation vector Z is composed of pseudoranges from all effective satellites, and the observation equation is: ,in, It is a nonlinear observation function. To reduce observation noise, a fusion observation equation is constructed, and the observation data from the secondary receiver is corrected to the time reference of the primary receiver using clock bias. The corrected pseudorange is: Where t is the current epoch time, This is used to estimate the reference time for clock bias.

6. The method as described in claim 5, characterized in that, For satellite i of the master receiver, the nonlinear observation function is: For satellite j of the secondary receiver, the nonlinear observation function is: .

7. The method as described in claim 1, characterized in that, The method further includes: for each common-view satellite, performing carrier phase smoothing processing on the raw pseudorange of the main receiver and the secondary receiver respectively to obtain smoothed pseudorange. : ； in, The smoothing time constant, For carrier wavelength, The carrier phase change is used to calculate the smoothed pseudorange difference. Combined with the original pseudorange difference As an auxiliary criterion, high-quality observation datasets are selected.

8. The method as described in claim 1, characterized in that, The confidence level P is calculated using the following formula: ； in, The number of valid satellites; The maximum number of available satellites is set to 12. The threshold is used; age is the delay in the positioning result. It is the attenuation factor; The weights are 0.4, 0.3, and 0.3 respectively, initially; the confidence level P∈[0,1], P>0.7 indicates high confidence and can be output; when P≤0.7, it prompts the user that the current positioning accuracy is low.

9. The method as described in claim 1, characterized in that, The mode rules are set as follows: based on the total number of effective satellites and the horizontal accuracy factor value, if the total number of effective satellites is less than four, the observation complementarity mode is adopted; if the total number of effective satellites is greater than or equal to four and the horizontal accuracy factor is less than a preset threshold, the tightly coupled fusion mode is adopted; if the horizontal accuracy factor is greater than or equal to a preset threshold, the weighted least squares fusion mode is adopted; if neither receiver can locate, the cooperative acquisition mode is enabled.

10. The method as described in claim 1, characterized in that, The raw observation data includes satellite number, pseudorange, carrier phase, Doppler shift, signal-to-noise ratio, and the receiver's own local timestamp; each row in the joint observation dataset contains receiver identifier, time identifier, satellite identifier, and raw observation data.

Images