Policy method for processing PPP-RTK time delay correction

Abstract

A strategy method for processing PPP-RTK time delay correction comprises the following steps: firstly obtaining a GNSS observation vector through a receiver to construct a network end observation model, combining the GNSS observation vector to construct a network end dynamic model, and then obtaining a network end filter according to the network end observation model and the network end dynamic model; the method comprises the following steps: dividing an area where a receiver is located into a plurality of grids, and screening correction information related to the grids according to corresponding grid numbers so as to construct a user side filter; when position correction is carried out, the network end filter firstly generates correction information, and then the network end filter transmits the generated correction information to the user end filter; the method comprises the following steps of: screening related correction information to obtain related correction information, then processing the screened related correction information to obtain vector data updated in real time, and uploading the vector data after measurement update to a server to obtain real-time position information. According to the invention, when the uncertainty of the correction product is neglected, the positioning information of the user is accurate.

Description

Technical Field

[0001] The present invention relates to an improvement in precise positioning technology, belonging to the field of global navigation positioning, and particularly relates to a strategy method for processing PPP-RTK time delay correction. Background Art

[0002] Precise Point Positioning (PPP-RTK) technology enables single receiver users to achieve rapid ambiguity resolution positioning. The successful implementation of PPP-RTK requires the support of a server. The server can generate positioning correction products, such as satellite orbits, satellite clock offsets, satellite instrument biases, and atmospheric delays, etc. Updating these correction products in the form of State Space Representation (SSR) can reduce the communication volume between the provider and the user's Kalman filter, thus allowing updates at longer time intervals; however, this also means that the user-side filter must perform time prediction on the corrections during the period when the correction products are unavailable, resulting in correction delays. In practical applications, the variance matrix of the uncertainty of the corrections in the user-side filter often cannot be provided for the user-side filter to use, because if detailed uncertainty information is provided, the advantage of SSR communication efficiency will be lost, which will cause the user filter to lose its minimum variance property and cannot accurately describe the quality of parameter estimation. When the uncertainty of the correction products is ignored, the solution result of the user filter may be inaccurate, leading to inaccurate positioning information for the user.

[0003] The Chinese patent application with the application number CN202211131093.2 and the application date of September 16, 2022, discloses a PPP-RTK positioning solution method and related device. The method includes: assigning non-combined floating-point filter information to a non-combined fixed solution filter, performing ionospheric constraint on the non-combined fixed solution filter of the current epoch through ionospheric correction numbers to output non-combined fixed solution filter information; performing position constraint on the ionosphere-free combined floating-point filter through the non-combined fixed solution position information in the non-combined fixed solution filter information to output the ionosphere-free combined floating-point solution after non-combined fixed solution position constraint; performing single-difference ambiguity fixing on the ionosphere-free combined floating-point solution, and obtaining single-difference ionospheric information through the obtained single-difference ambiguity information; determining the ionospheric activity level based on the single-difference ionospheric information and the ionospheric correction numbers to obtain a determination result; and selecting the single-difference ionospheric information or the non-combined fixed solution filter information for non-combined ambiguity fixing according to the determination result to output a non-combined fixed solution result. In the above solution, floating-point filtering processing is performed through non-combination and ionosphere-free combination respectively, and then the non-combined fixed solution position information obtained by the non-combined fixed solution filter is used to perform position constraint on the ionosphere-free combined floating-point filter to accelerate the convergence of the ionosphere-free combined floating-point solution. However, the above solution does not solve the problem that ignoring the uncertainty of the correction products will lead to inaccurate positioning information for the user.

[0004] Disclosing the information of this background art section is only intended to enhance the overall understanding of the background of this patent application, and should not be construed as an admission or any form of implication that this information constitutes prior art already known to those of ordinary skill in the art. Summary of the Invention

[0005] The object of the present invention is to overcome the problem in the prior art that ignoring the correction of product uncertainty can lead to inaccurate positioning information of users, and provides a strategy method for processing PPP-RTK time delay correction that ignores the correction of product uncertainty and has accurate positioning information of users.

[0006] To achieve the above object, the technical solution of the present invention is: a strategy method for processing PPP-RTK time delay correction, and the strategy method for processing PPP-RTK time delay correction includes the following steps:

[0007] First step: First, obtain the GNSS observation vector through the receiver, then define the augmented design matrix, combine the GNSS observation vector to construct the network-side observation model, then define the transition matrix, combine the GNSS observation vector to construct the network-side dynamic model, and then obtain the network-side filter according to the network-side observation model and the network-side dynamic model;

[0008] Second step: First, divide the area where the receiver is located into multiple grids, each grid corresponding to a grid number, then the receiver obtains its own position information through initial positioning to obtain the grid number corresponding to the receiver, and according to the corresponding grid number, screen out the correction information related to this grid to construct the user-side filter;

[0009] Third step: When performing position correction, the network-side filter first generates correction information, and then the network-side filter transmits the generated correction information to the user-side filter;

[0010] Fourth step: First, the user-side filter screens the transmitted correction information to obtain relevant correction information, then processes the screened relevant correction information to obtain real-time updated vector data, and after a preset time interval, the user-side filter uploads the measured and updated vector data to the server to obtain real-time position information.

[0011] In the first step, constructing the network-side observation model specifically is: Assume the network-side GNSS observation vector is constrained by the pseudo-observation value with zero mean ;

[0012] The enhanced observation vector ;

[0013] The pseudo-observation value , where is the corresponding noise vector, and the coefficient matrix D is used to construct the unknown parameters and to establish the connection between them. Subsequently, an augmented design matrix is defined, while , and at this time, the network-side observation model is:

[0014] .

[0015] The construction of the network-side dynamic model is specifically as follows:

[0016] Define the transition matrix ;

[0017] Augment the process noise ;

[0018] At this time, the network-side dynamic model is expressed as:

[0019] ;

[0020] The pseudo-observation vector imposes constraints on the parameter vector at epoch k. The dynamic model imposes constraints on the parameter vectors and between two consecutive epochs k - 1 and k.

[0021] The transition matrix also satisfies: ;

[0022] A set of pseudo-observation vectors and constitutes a part of the dynamic model. Multiply the dynamic model on the left by a one-to-one transformation matrix ;

[0023] where ;

[0024] Here E is a matrix with full column rank, satisfying , while is invertible; the variance matrix of the process noise is denoted by , that is, we get:

[0025] .

[0026] After avoiding duplicate constraints, only part of the dynamic model needs to be used in the network-side filter. This part is: ;

[0027] The noise vector with weighted constraints is the process noise vector function; According to the law of error propagation, the noise vector has a variance matrix of:

[0028] .

[0029] When repetitive constraints occur, an ionospheric weighting model is adopted to incorporate the spatial dependence of the inter-station ionospheric delay into the observation model of the network-side assisting the user-side.

[0030] In the second step, the user filter is constructed as follows: The recursive estimation of the user parameter vector relies on the correction vector provided by the network-side filter. Connect the corrections across epochs, that is:

[0031] ;

[0032] where represents the associated zero-sample pseudo-observation vector, which can predict the correction value from epoch to epoch . Therefore, the observation model of the user filter is expressed as:

[0033] ;

[0034] where the approximation symbol is because the uncertainty of the enhanced observation value is affected by the process noise and the estimation error .

[0035] By defining the transition matrix and the enhanced process noise vector , the dynamic model of the user filter is:

[0036] .

[0037] When , the variance matrix of the enhanced observation value is equal to 0. At this time:

[0038] .

[0039] In the time update step, for the parameter vector of the time prediction enhancement of the user filter, the user uses the least squares principle to solve , and the solution of the time update is used as an additional observation value to participate in the solution.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] 1. In a strategy method for processing PPP-RTK time delay correction in the present invention, through the high-precision correction information provided by the network side, the user side can fix the ambiguity within a short time, thereby achieving fast and real-time high-precision positioning, effectively reducing the influence of calibration prediction errors, that is, enabling the network side to update the calibration after a longer time interval, broadcast correction data to the user, minimizing the influence of calibration prediction errors and the uncertainty of time correlation, and at the same time solving the problem of how to avoid repetitive constraints in the network side filter to provide the user with a minimum variance correction solution. Therefore, when the present invention ignores the uncertainty of the correction product, the user's filtering solution result is accurate.

[0042] 2. In a strategy method for processing PPP-RTK time delay correction in the present invention, combined with the correction information provided by the network side, the user side can achieve centimeter-level high-precision positioning. Through the ionospheric weighting model and the time prediction of the correction information, this solution can effectively reduce the influence of correction delay on positioning accuracy. Even in an environment where the ionosphere is active or signals are blocked, through combination with a high-precision IMU, PPP-RTK can still maintain high-precision positioning. Therefore, the present invention is safe to use and stable in operation.

[0043] 3. In a strategy method for processing PPP-RTK time delay correction in the present invention, the user side obtains the GNSS observation vector through the receiver, combines the correction information provided by the network side filter, constructs the observation model and dynamic model of the user side filter. Through the time update and measurement update steps, the user side filter recursively estimates the unknown parameters of the user, and finally outputs a high-precision positioning result, ensuring that the user side can effectively handle the delay of the correction information, improving positioning accuracy and reliability. The present invention has accurate positioning and high reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a schematic diagram of the reference network of the present invention.

[0045] Figure 2 is a schematic diagram of the clock error prediction error of the inter-satellite single difference of the present invention.

[0046] Figure 3 is a schematic diagram of the ionospheric delay prediction error in the present invention.

[0047] Figure 4 is a diagram of the inter-satellite single difference combined correction prediction error in the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0048] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0049] See Figures 1 to 4, A strategy method for processing PPP-RTK time delay correction, the strategy method for processing PPP-RTK time delay correction includes the following steps:

[0050] In the first step, first obtain the GNSS observation vector through the receiver, then define the augmented design matrix, combine the GNSS observation vector to construct the network-side observation model, then define the transition matrix, combine the GNSS observation vector to construct the network-side dynamic model, and then obtain the network-side filter according to the network-side observation model and the network-side dynamic model;

[0051] In the second step, first divide the area where the receiver is located into multiple grids, each grid corresponds to a grid number, then the receiver obtains its own position information through initial positioning to obtain the grid number corresponding to the receiver, and according to the corresponding grid number, filter out the correction information related to the grid to construct the user-side filter;

[0052] In the third step, when performing position correction, the network-side filter first generates correction information, and then the network-side filter transmits the generated correction information to the user-side filter;

[0053] In the fourth step, first the user-side filter screens the transmitted correction information to obtain the relevant correction information, then processes the screened relevant correction information to obtain the real-time updated vector data. After a preset time interval, the user-side filter uploads the measured and updated vector data to the server to obtain the real-time position information.

[0054] In the first step, the construction of the network-side observation model is specifically as follows: Assume the network-side GNSS observation vector is constrained by the pseudo-observation value with zero mean ;

[0055] The enhanced observation vector ;

[0056] The pseudo-observation value , where is the corresponding noise vector, and the coefficient matrix D is used to construct the relationship between the unknown parameters and . Subsequently, an augmented design matrix is defined, and at the same time . At this time, the network-side observation model is:

[0057] .

[0058] The construction of the network-side dynamic model is specifically as follows:

[0059] Define the transition matrix ;

[0060] The enhanced process noise ;

[0061] At this time, the network-side dynamic model is expressed as:

[0062] ;

[0063] Pseudo-observation vector At epoch k, a constraint is imposed on the parameter vector The dynamic model imposes constraints on the parameter vectors and between two consecutive epochs k - 1 and k.

[0064] The transition matrix also satisfies: ;

[0065] A set of pseudo-observation vectors and forms part of the dynamic model. Left-multiplying the dynamic model by a one-to-one transformation matrix ;

[0066] where ;

[0067] Here, E is a matrix with full column rank, satisfying , and at the same time is invertible; the variance matrix of the process noise is denoted by , that is, we get:

[0068] .

[0069] After avoiding duplicate constraints, only part of the dynamic model needs to be used in the network-side filter. This part is: ;

[0070] The weighted constraint noise vector is a function of the process noise vector ; According to the error propagation law, the variance matrix of the noise vector is:

[0071] .

[0072] When there are repetitive constraints, an ionospheric weighting model is adopted to incorporate the spatial dependence of the inter-station ionospheric delay into the observation model for assisting the user terminal at the network side.

[0073] In the second step, the user filter is constructed as follows: The recursive estimation of the user parameter vector depends on the correction vector provided by the network-side filter. Connect the corrections across epochs, that is:

[0074] ;

[0075] Wherein, represents an associated zero-sample pseudo-observation vector, which can predict the correction value from epoch to epoch . Therefore, the observation model of the user filter is expressed as:

[0076] ;

[0077] Wherein, the approximate symbol is because the uncertainty of the enhanced observation value is affected by the process noise and the estimation error .

[0078] By defining the transition matrix and the enhanced process noise vector , the dynamic model of the user filter is:

[0079] .

[0080] When , the variance matrix of the enhanced observation value is equal to 0. At this time:

[0081] .

[0082] In the time update step, for the parameter vector of the user filter time prediction enhancement, the user uses the least squares principle to solve , and the solution of the time update is used as an additional observation value to participate in the solution.

[0083] The supplementary description of the present invention is as follows:

[0084] This technology extends the applicability of this strategy to multi-receivers, that is, small to regional GNSS reference networks. For this purpose, an ionospheric weighting model is adopted. Therefore, this extension is not trivial because when making spatial and temporal predictions of ionospheric corrections in the best linear unbiased prediction (BLUP) framework, care needs to be taken to avoid duplication of constraints.

[0085] Example 1:

[0086] A strategy method for processing PPP-RTK delay corrections, the strategy method for processing PPP-RTK delay corrections includes the following steps:

[0087] Step 1: First, obtain the GNSS observation vector through the receiver, then define the augmented design matrix, combine the GNSS observation vector to construct the network-side observation model, then define the transition matrix, combine the GNSS observation vector to construct the network-side dynamic model, and then obtain the network-side filter according to the network-side observation model and the network-side dynamic model;

[0088] Step 2: First, divide the area where the receiver is located into multiple grids, each grid corresponding to a grid number. Then the receiver obtains its own position information through initial positioning to get the grid number corresponding to the receiver, and according to the corresponding grid number, filter out the correction information related to this grid to construct the user-side filter;

[0089] Step 3: When performing position correction, the network-side filter first generates correction information, and then the network-side filter transmits the generated correction information to the user-side filter;

[0090] Step 4: First, the user-side filter filters the transmitted correction information to obtain the relevant correction information, and then processes the filtered relevant correction information to obtain the vector data updated in real time. After a preset time interval, the user-side filter uploads the measured and updated vector data to the server to obtain the real-time position information.

[0091] Embodiment 2:

[0092] Embodiment 2 is basically the same as Embodiment 1, and the difference is that:

[0093] Suppose the network-side GNSS observation vector:

[0094] is constrained by a pseudo-observation value with zero mean ; the enhanced observation vector , pseudo-observation value , where is the corresponding noise vector, and the coefficient matrix D is used to construct the relationship between the unknown parameters and ;

[0095] The enhanced observation vector ,

[0096] pseudo-observation value , where is the corresponding noise vector, and the coefficient matrix D is used to construct the relationship between the unknown parameters and ; subsequently, an augmented design matrix is defined, and at the same time , and at this time the network-side observation model is:

[0097] ;

[0098] In addition to the observation model, a dynamic model of the parameters involved also needs to be considered. The Kalman filter dynamic model is expressed as:

[0099] ;

[0100] where the zero-valued pseudo-observations relate the parameters between consecutive epochs k - 1 and k using their respective transition matrices respectively, where , so the matrix product connects the parameters from the i-th epoch to the j-th epoch (j > i). When j = i, (the identity matrix). The randomness of the pseudo-observations is characterized by their associated zero-mean process noise ;

[0101] The construction of the network-side dynamic model is specifically as follows:

[0102] Define the transition matrix ;

[0103] Enhance the process noise ;

[0104] At this time, the network-side dynamic model is expressed as:

[0105] ;

[0106] The pseudo-observation vector imposes constraints on the parameter vector at epoch k. The dynamic model imposes constraints on the parameter vectors and between two consecutive epochs k - 1 and k; if these two types of constraints do not repeat in a certain sense, the observation model and the dynamic model can be used for the network-side filter. However, care needs to be taken to avoid any potential problems of repetitive constraints;

[0107] The transition matrix also satisfies: ;

[0108] A combination of a set of pseudo-observation vectors and constitutes part of the dynamic model. Multiply the dynamic model on the left by a one-to-one transformation matrix ,

[0109] To identify the part of the dynamic model that cannot be derived from the pseudo-observation vectors and , multiply the dynamic model on the left by a one-to-one transformation matrix ,

[0110] Among them, ;

[0111] Here, E is a matrix with full column rank, satisfying , and at the same time is invertible. The variance matrix of the process noise is represented by , that is, we get:

[0112] .

[0113] After avoiding the repeated constraints, only part of the dynamic model needs to be used in the network-side filter, and this part is: ;

[0114] Two functions of the process noise vector and can be proven to be independent. Since the columns of matrix and the columns of matrix are linearly independent, the combination is independent of the pseudo-observation vector and , and the only remaining combination is the repeated constraint.

[0115] Embodiment 3:

[0116] Embodiment 3 is basically the same as Embodiment 1, and the difference is that:

[0117] The weighted constraint noise vector is a function of the process noise vector ; According to the error propagation law, the variance matrix of the noise vector is:

[0118] ;

[0119] When the repeated constraint appears, the ionospheric weighting model is adopted to incorporate the spatial dependence of the inter-station ionospheric delay into the observation model of the network-side assisting the user-side, and the observation equation is expressed as:

[0120] ;

[0121] Among them, the network-side and user-side observation equations are represented by and respectively, and include GNSS pseudorange and carrier phase observations; the subscript k refers to the time index, where it is assumed that the user has started collecting data after the network-side has collected data, so ; The unique parameter vectors and Respectively through the corresponding full-rank design matrices and are associated with the observations, which may include receiver positions, zenith tropospheric delays, receiver clock biases and drifts, network-end and user-end ionospheric delays, and carrier-phase integer ambiguities. The unknown correction vector contains parameters that are common to the network-end and user-end observation equations, such as satellite orbit biases, satellite clock biases, drifts, and ionospheric corrections at the user station (optional). The corresponding design matrices are and . The zero-mean observation noise vectors at the network end and user end are denoted by and respectively;

[0122] Here we distinguish between the following two unknown parameter vectors:

[0123] (1) The network-end ionospheric parameter vector belongs to the network-end specific parameters in the observation equations ;

[0124] (2) The grid ionospheric parameter vector belongs to the correction vector ;

[0125] As an illustrative example, Figure 1 shows four stations (blue triangles) at location A as the GNSS network end. These four stations experience the network-end ionospheric delay , while nine grid points (green stars) experience the grid ionospheric delay parameter ;

[0126] Assume that the vectors and are the last p and q variables of and respectively, i.e., . The inter-station spatial dependence of the slant ionospheric delay can be modeled by the weighted constraint , where the coefficient matrix D is specified as:

[0127] ;

[0128] where the full-column rank matrix is used to form the inter-station single-difference ionospheric delay at the network end. Therefore, represents the single-difference form of the undifferenced ionospheric delay . At the same time, the matrix is used to form by subtracting a part of the network ionospheric delay from as the undifferenced grid ionospheric delay In the single-difference form, to better understand the structure of matrix D, consider again Figure 1 the network terminals in . If m GNSS satellites are visible to all four network-terminal receivers, the sub-matrix of D can be expressed as:

[0129] ;

[0130] where . For this case, consists of m sub-vectors specific to the four network-terminal stations, and these m sub-vectors represent undifferenced ionospheric delays. Among them, the sub-vector of the first station is used as a reference and subtracted from the remaining m sub-vectors to form the form of inter-station single differences. Similarly, the undifferenced ionospheric delay of the first network-terminal receiver is subtracted from the m sub-vectors specific to the nine grids ;

[0131] To specify the transfer matrix and verify the equation , we need to make assumptions about the time behavior of the parameter vectors and . Due to the complex physical properties of ionospheric delays, its time modeling is challenging. Here, we assume that the time behavior of ionospheric delays follows a constant random process, and such an assumption is reasonable under sufficiently short time delays;

[0132] At this time, the corresponding transfer matrices of the ionospheric parameter vectors and can be simply specified by the identity matrix, that is, and ;

[0133] and represent the transfer matrices of the remaining parameter vectors and . Therefore, the transfer matrix of the entire parameter vector can be expressed as:

[0134] . Substituting into , we get:

[0135]

[0136] ; Regardless of how the transfer matrices and are selected;

[0137] The above equation can show that the equation holds. At the same time, this part of the ionospheric dynamic model characterized by should be discarded in the network-terminal filter;

[0138] At this time, we specify the coefficient matrix for the dynamic model of the network-side filter:

[0139] , for which we need 1) matrix E; 2) the variance matrix ;

[0140] First, specify matrix E. The corresponding complementary matrix E is:

[0141] ;

[0142] Among them, the matrix satisfies , such that is an invertible matrix. For the network-side in Figure 1 , is expressed as:

[0143] ;

[0144] Define the variance matrix of the process noise vector as:

[0145] ;

[0146] Among them and represent and , the process noise variance matrix of;

[0147] The ionospheric parameter as well as the process noise variance matrix of are given by and respectively, represents the covariance matrix, so:

[0148] ;

[0149] The purpose of the variance matrix S is to capture the random behavior of the oblique ionospheric delay, which is a function of the station separation and the satellite elevation angle. The dimension of the matrix S is equal to the total number of network-side and grid ionospheric slant delays in each epoch. For Figure 1 , the order of the matrix S is , and the matrix S is:

[0150] ;

[0151] Among them the correlation matrix captures the spatial dependence between stations, the diagonal matrix of ,

[0152] including an elevation assistance factor , the expressions for R and M are:

[0153] ;

[0154] where the positive scalar is expressed in square units of length, reflects the ionospheric activity, and is usually driven by factors such as solar activity and geographical latitude. The distance between stations and is represented by , the elevation angle of satellite relative to the network center is represented by , and the correlation coefficient is a decreasing function of the station spacing . As increases, rapidly approaches 0, indicating that when the distance between two stations is far enough, it is considered that their undifferenced ionospheric delays are completely different.

[0155] Example 4:

[0156] Example 4 is basically the same as Example 1, and the difference is that:

[0157] In the third step, constructing a user filter specifically means that the recursive estimation of the user parameter vector depends on the correction vector provided by the network-side filter, and the corrections across epochs are connected, that is:

[0158] ;

[0159] where represents the associated zero-sample pseudo-observation vector, which can predict the correction value from epoch to epoch . Therefore, the observation model of the user filter is expressed as:

[0160] ;

[0161] where the approximation symbol is because the uncertainty of the enhanced observation value is affected by the process noise and the estimation error .

[0162] By defining the transition matrix and the enhanced process noise vector , the dynamic model of the user filter is:

[0163] .

[0164] When the enhanced observation value has a variance matrix equal to 0, then:

[0165] .

[0166] In the time update step, the parameter vector for the user filter time prediction enhancement is solved by the user using the least squares principle and the solution

[0167] Example 5:

[0168] Example 5 is basically the same as Example 1, except that:

[0169] To explore the positioning performance of the proposed strategy, we selected three regional networks located at Place A, Place B, and Place C, representing small, medium, and large networks respectively.

[0170] We set the zenith accuracy of the pseudorange observation and the carrier phase observation to 0.3 m and 0.003 m respectively. The common sine function is used to describe the elevation angle-related accuracy. The LAMBDA method is used to implement partial ambiguity resolution (PAR) at the user side. The selected ambiguities are based on the decorrelated ambiguity variance matrix, and the fixed failure rate ratio test (FFRT) strategy is used to check whether the selected ambiguities have been fixed, with a minimum success rate of 99.9%. It should be noted that in order to extend the SPB update interval and reduce communication data, the wide-lane (WL) technique is applied. Due to its decorrelation characteristics, WL SPB shows significantly greater stability compared to the undifferenced form.

[0171] For comparison, in addition to the proposed strategy, we also considered the case of ignoring the correction uncertainty in the user filter. For this purpose, we used the data of the small-scale network located at Place A and processed the GPS dual-frequency (L1 / L2) data, adopting the following two strategies:

[0172] (1) Scheme 1: The classical strategy, where the correction uncertainty is ignored;

[0173] (2) Scheme 2: The proposed strategy, where the correction and its uncertainty are incorporated into the user filter.

[0174] First, the correction interval of the satellite clock is selected as 20 seconds, the ionospheric delay is 1 minute, and the correction interval of

[0175] Figure 2 The clock error prediction errors of the inter-satellite single differences of two schemes are shown. Both schemes exhibit obvious noise. Since the satellite clock error varies significantly over time, the prediction error accumulates to several centimeters within a time interval of 20, at which point the satellite clock error is unavailable and the user needs to independently estimate the satellite clock error. However, as can be seen from the figure, the prediction error of Scheme 2 is generally smaller than that of Scheme 1, and Scheme 2 performs better.

[0176] Figure 3 The performance of the ionospheric delay prediction errors of two schemes is shown. Although the ionospheric error is broadcast to the user every minute, as can be seen from the figure, for Scheme 1, the cumulative satellite single-difference ionospheric prediction error exceeds 4 cm, while the prediction error of Scheme 2 performs better and is always below 2 cm.

[0177] Figure 4 The inter-satellite single-difference combined correction prediction errors of two schemes are shown. As Figure 4 shown, the satellite-to-satellite SD comprehensive correction prediction error in Scheme 2 is significantly smaller than that in Scheme 1. Due to the neglect of the uncertainties of the time-varying satellite clock and ionospheric delay prediction errors, the comprehensive prediction error of Scheme 1 is more noisy. In addition, both Scheme 1 and Scheme 2 show cumulative errors over time (every 5 minutes), affected by the SPB prediction error. In contrast, the comprehensive prediction error of Scheme 2 is significantly more stable.

[0178] The above description is only the preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiment. Any equivalent modification or change made by those of ordinary skill in the art according to the content disclosed by the present invention shall be included in the protection scope recorded in the claims.

Claims

1. A strategy method for processing PPP-RTK time delay correction, characterized in that: The strategy method for processing PPP-RTK time delay correction includes the following steps: First, obtain the GNSS observation vector through the receiver, then define the augmented design matrix, combine the GNSS observation vector to construct the network-side observation model, then define the transition matrix, combine the GNSS observation vector to construct the network-side dynamic model, and then obtain the network-side filter according to the network-side observation model and the network-side dynamic model; Second, divide the area where the receiver is located into multiple grids, each grid corresponding to a grid number. Then the receiver obtains its own position information through initial positioning to obtain the grid number corresponding to the receiver, and filters out the correction information related to this grid according to the corresponding grid number to construct the user-side filter; Third, when performing position correction, the network-side filter first generates correction information, and then the network-side filter transmits the generated correction information to the user-side filter; Fourth, first, the user-side filter screens the transmitted correction information to obtain relevant correction information, and then processes the screened relevant correction information to obtain real-time updated vector data. After a preset time interval, the user-side filter uploads the measured and updated vector data to the server to obtain real-time position information.

2. The strategy method for processing PPP-RTK time-delay correction according to claim 1, characterized in that: In the first step, the specific construction of the network-side observation model is as follows: Assume that the network-side GNSS observation vector is constrained by a pseudo-observation value with zero mean constraint; The enhanced observation vector ; Pseudo-observations , where is the corresponding noise vector, and the coefficient matrix D is used to construct the relationship between the unknown parameters and . Subsequently, an augmented design matrix is defined, and . At this time, the network-side observation model is as follows: 。 3. A strategy method for processing PPP-RTK time delay correction according to claim 1, characterized in that: The construction of the network-side dynamic model is specifically as follows: Define the transition matrix ; Enhanced process noise ; At this time, the network-side dynamic model is expressed as: ； Pseudo-observation vector At epoch k, a constraint is imposed on the parameter vector The dynamic model imposes constraints on the parameter vectors and between two consecutive epochs k-1 and k.

4. A strategy method for processing PPP-RTK time delay correction according to claim 3, characterized in that: The transfer matrix further satisfies: ; A set of pseudo-observation vectors and whose combination forms part of the dynamic model, and left-multiplying the dynamic model by a one-to-one transformation matrix ; Among them, ; Here, E is a matrix with full column rank, satisfying , and at the same time is invertible; the variance matrix of the process noise is denoted by , that is, we get: 。 5. A strategy method for processing PPP-RTK time delay correction according to claim 4, characterized in that: After the above-mentioned repeated constraints are avoided, only a part in the dynamic model needs to be used in the network-side filter, and this part is as follows: ; Weighted constrained noise vector is the process noise vector function; According to the error propagation law, the variance matrix of the noise vector is: 。 6. A strategy method for processing PPP-RTK time delay correction according to claim 2, characterized in that: When repetitive constraints occur, an ionospheric weighting model is adopted to incorporate the spatial dependence of the inter-station ionospheric delay into the network-side assisted user-side observation model.

7. A strategy method for processing PPP-RTK time delay correction according to claim 1, characterized in that: In the second step, the user filter is constructed as follows: The recursive estimation of the user parameter vector depends on the correction vector provided by the network-side filter, and the correction is connected across epochs, that is: ； Among them, represents the associated zero-shot pseudo-observation vector, which can predict the correction value from epoch to epoch . Therefore, the observation model of the user filter is expressed as: ； Among them, the approximation symbol is because the uncertainty of the enhanced observation value is affected by the process noise and the estimation error .

8. A strategy method for processing PPP-RTK time delay correction according to claim 7, characterized in that: By defining the transition matrix and the augmented process noise vector , the user filter dynamic model is as follows: 。 9. A strategy method for processing PPP-RTK time delay correction according to claim 8, characterized in that: When the variance matrix of the enhanced observation value is equal to 0, then: 。 10. A strategy method for processing PPP-RTK time delay correction according to claim 9, characterized in that: In the time update step, the parameter vector for enhancing the time prediction of the user filter , and the user solves it using the least squares principle , and the solution of the time update is used as an additional observation value to participate in the solution calculation.

Images