A low-cost PPP random model adaptive adjustment method for complex environment

Abstract

The application discloses a low-cost PPP random model adaptive adjustment method for complex environment, and relates to the technical field of satellite navigation and positioning. The specific steps of the method comprise the following steps: performing strict quality control on GNSS original observation data of a low-cost terminal in a complex environment; obtaining observation value errors based on a code reduction phase model and fitting to obtain a prior random model based on a carrier-to-noise ratio; and adaptively adjusting the weights of pseudorange and carrier phase observation values through PPP posterior residual error to obtain more robust positioning results. The application fully utilizes the correlation of multi-system and multi-frequency GNSS measurement errors, and adaptively adjusts the random model in combination with a posterior residual threshold test, so that the convergence speed and positioning accuracy of the low-cost terminal PPP in a complex urban environment are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of satellite navigation, and in particular to a low-cost PPP random model adaptive adjustment method for complex environments. BACKGROUND

[0002] GNSS signals are seriously affected by various modeled and non-modeled errors during propagation. Modeled errors such as satellite orbits, clock errors, differential code biases, ionosphere, troposphere, etc. can be well eliminated through observation domain differencing or state domain correction. However, non-modeled errors such as measurement noise, multipath, radio frequency interference, etc. usually have no unified correction model, and their impact on GNSS positioning performance becomes more and more serious. Especially with the continuous development of modernization, the urban environment becomes more and more complex. The shielding and reflection of various buildings to GNSS signals, the radio frequency interference generated by different wireless network facilities, and the poor anti-multipath interference ability of low-cost terminals all make it extremely challenging to achieve accurate and reliable positioning in complex environments. An important factor affecting positioning accuracy and reliability is how to reasonably allocate the weights of multi-system, multi-frequency GNSS observations.

[0003] Traditional GNSS observation weighting models usually use equal weight models or elevation angle weighting models. The equal weight model considers that the ranging accuracy of observations of all satellite navigation systems and all frequencies is equal, but this model is not applicable in most real environments. The elevation angle weighting model considers that the measurement error of GNSS observations is mainly related to the height of the satellite, and a higher weight is allocated to satellites with higher elevation angles, and a lower weight is allocated to satellites with lower elevation angles. This model is widely used in some internationally renowned GNSS data processing software such as BERNESE, PANDA, GAMIT, etc., and has achieved good positioning results in open environments, but the effect is not good in urban complex environments with serious shielding of buildings and trees. Therefore, designing a reasonable multi-system, multi-frequency GNSS observation weighting model to adaptively adjust the observation weights of pseudoranges and carrier phases is a key factor to improve the convergence speed and positioning accuracy of low-cost terminals in complex urban environments. SUMMARY

[0004] The present application is provided to solve the above problems in the prior art. Therefore, a low-cost PPP random model adaptive adjustment method for complex environments, a device and a medium are needed, which can significantly improve the anti-roughness ability of low-cost navigation terminals in complex urban environments and effectively improve the convergence speed and positioning accuracy of PPP.

[0005] According to a first aspect of the present application, a low-cost PPP random model adaptive adjustment method for complex environments is provided, the method comprising:

[0006] According to the quality control parameters, the GNSS raw observation data of the low-cost terminal in the complex environment is subjected to quality control to obtain GNSS observation values, and the quality control parameters include pseudorange rough error detection, carrier cycle slip detection and carrier ratio threshold test;

[0007] The observation value error is obtained through the code-subtraction phase model, and the prior random model based on the carrier-to-noise ratio is fitted according to the correlation between the observation value error and the carrier-to-noise ratio;

[0008] The PPP posterior residual is obtained through filtering, and the weights of the pseudorange and carrier phase observation values are adaptively adjusted according to the residual threshold test to obtain the positioning result.

[0009] Further, the GNSS raw observation data of the low-cost terminal in the complex environment is subjected to quality control according to the quality control parameters to obtain GNSS observation values, and the quality control parameters include:

[0010] According to the relationship between the pseudoranges between epochs and the Doppler, the following pseudorange rough error detection model is constructed:

[0011] G P =(P k -P k-1 )-(D k +D k-1 )·λ / 2,k≥2 (33)

[0012] Wherein, P k and P k-1 represent the pseudorange observation values at the kth and k-1th epochs, D k and D k-1 represent the Doppler shift observation values at the kth and k-1th epochs, G P is a pseudorange rough error test quantity, according to the variance-covariance error propagation criterion, if |G P |>3σ P , it is considered that the current pseudorange observation value has a rough error, at this time, the weight of the pseudorange observation value is reduced or the observation value is removed, σ P represents the mean error of the pseudorange measurement error;

[0013] The following carrier phase integer ambiguity test formula is constructed through high-precision Doppler:

[0014]

[0015] Wherein, ΔN represents the change quantity of the carrier phase integer ambiguity at the kth and k-1th epochs, represents the noise of the carrier phase observation value, and ε D represents the noise of the Doppler observation value;

[0016] The threshold value for high-precision Doppler carrier phase cycle slip detection is set as follows:

[0017]

[0018] wherein ξ represents the threshold value for carrier phase cycle slip detection; when the observation value sampling interval is 1 second, ξ is set to 1 cycle, and a cycle slip of more than 1 cycle can be detected by high-precision Doppler;

[0019] The carrier-to-noise ratio threshold test formula S is constructed obs Satellite signals with poor observation quality are directly rejected:

[0020]

[0021] wherein C / N0 represents the carrier-to-noise ratio of the received satellite signal, and Ele represents the elevation angle of the satellite; if the elevation angle of the satellite is greater than 15° and the carrier-to-noise ratio is greater than 30 dB-Hz, the current observation value is considered valid, and the observation data outside the valid range indicates that the satellite signal cannot be effectively tracked and is rejected.

[0022] Further, the observation value error is obtained by the code-minus-carrier phase model, and the prior random model based on the carrier-to-noise ratio is fitted according to the correlation between the observation value error and the carrier-to-noise ratio, and specifically includes:

[0023] The code-minus-carrier phase model is constructed by the GNSS basic observation equation to obtain the pseudorange measurement error;

[0024] The basic observation equation on the GNSS propagation path from satellite signal transmission to reception is as follows:

[0025]

[0026] wherein ρ represents the geodetic distance, c represents the speed of light in vacuum, dt represents the receiver clock error, dT represents the satellite clock error, I represents the ionospheric delay of the slant path, T represents the tropospheric delay of the slant path, λ represents the carrier frequency of the satellite transmitted signal, N represents the integer ambiguity of the carrier phase, and M P represents the pseudorange multipath error, represents the carrier phase multipath error, and ε P represents the pseudorange measurement noise, represents the pseudorange measurement noise;

[0027] According to formula (5), the expression of the code-minus-carrier phase model (CMC, Code-minus-Carrier phase) is obtained:

[0028]

[0029] The code-minus-carrier phase residual is obtained by taking the average value in the continuous arc segment:

[0030]

[0031] A prior stochastic model based on the carrier-to-noise ratio (CNR) is obtained by fitting the correlation between pseudorange measurement error and CNR; the expression of the prior stochastic model based on CNR is as follows:

[0032]

[0033] in, and The empirical values ​​are 0.1–0.3 and 0.001–0.003; v and c are the coefficients to be fitted for the carrier-to-noise ratio stochastic model, calibrated based on the correlation between pseudorange measurement error CMCR and C / N0. The function expression to be fitted is:

[0034]

[0035] Among them, CMCR and C / N0 are obtained from actual measurement data, and v and c are parameters to be fitted. In order to make the random model coefficients v and c based on the carrier-to-noise ratio as consistent as possible with the quality of low-cost terminal GNSS observation data, the best fitting coefficients are searched based on the minimum residual sum of squares criterion during the fitting process.

[0036] Taking the logarithm of both sides of equation (9), we obtain the objective function e in the following equation. 2 :

[0037]

[0038] Where k and N represent the k-th measurement and the total number of measurements, respectively, when the objective function e 2 The optimal fitting parameters for v and c are obtained when the minimum value is reached.

[0039] Furthermore, the step of obtaining the PPP post-hoc residual through filtering and adaptively adjusting the weights of pseudorange and carrier phase observations based on the residual threshold test to obtain the positioning result specifically includes:

[0040] The observations are assigned initial weights by a prior random model, and the PPP a priori residuals are obtained after filtering and solving.

[0041] Linearizing the basic GNSS observation equation (5) yields the error equation for the observation vector at the k-th epoch:

[0042]

[0043] Among them, B k Represents the design matrix. L represents the vector of parameters to be estimated. k V represents the vector of observed values ​​minus calculated values.k represents the posteriori residual vector solved by filtering. According to the principle of weighted least squares, formula (11) is solved to obtain:

[0044]

[0045] In the formula, P k is the priori weight matrix (diagonal matrix, diagonal elements are p kk , k = 1, 2, … n) calculated by formula (8):

[0046]

[0047] In the formula, n represents the total number of observation values, the larger the variance of the observation value, the more random errors contained in the measurement data, and the smaller the weight of the corresponding observation value;

[0048] According to the residual threshold, the weights of the pseudo-range and carrier phase observation values are adaptively adjusted to obtain a more robust positioning result;

[0049] In order to realize the adaptive adjustment of the pseudo-range and carrier phase observation value weights, a weight factor w is introduced to control the weight:

[0050]

[0051] In the formula, represents the equivalent weight matrix (diagonal elements are ); considering formula (12)-(14), the adaptive robust estimation is obtained:

[0052]

[0053] The posteriori residual V k is normalized to and a piecewise function is constructed to adaptively adjust the weight factor:

[0054]

[0055] In the formula, c0 and c1 represent the lower bound and upper bound of the normalized residual; for a low-cost GNSS navigation terminal, the empirical values of the two coefficients are: c0 = 1.0-1.5, c1 = 2.5-5.0.

[0056] According to the second scheme of the application, a low-cost PPP random model adaptive adjustment device for complex environment is provided, and the device comprises:

[0057] a quality control unit configured to perform quality control on GNSS raw observation data of the low-cost terminal in a complex environment according to quality control parameters to obtain GNSS observation values, the quality control parameters including pseudorange gross error detection, carrier cycle slip detection, and carrier ratio threshold test;

[0058] a priori random model construction unit configured to obtain observation value error through a code-minus-phase model and fit a priori random model based on carrier-to-noise ratio according to a correlation between the observation value error and the carrier-to-noise ratio;

[0059] a positioning result calculation unit configured to obtain PPP posteriori residual through filtering and adaptively adjust weights of pseudorange and carrier phase observation values according to residual threshold test to obtain a positioning result.

[0060] Further, the quality control unit is further configured to:

[0061] According to a relationship between interepoch pseudoranges and Doppler, a pseudorange gross error detection model is constructed as follows:

[0062] G P = (P k -P k-1 )-(D k +D k-1 )·λ / 2,k≥2 (49)

[0063] wherein P k and P k-1 represent pseudorange observation values at k and k-1 epochs respectively, D k and D k-1 represent Doppler shift observation values at k and k-1 epochs respectively, G P is a pseudorange gross error test quantity, according to variance-covariance error propagation criterion, if |G P |>3σ P , it is considered that the current pseudorange observation value has a gross error, at this time, the weight of the pseudorange observation value is reduced or the observation value is removed, σ P represents a mean error of pseudorange measurement error;

[0064] A carrier phase integer ambiguity test formula is constructed as follows through high-precision Doppler:

[0065]

[0066] wherein ΔN represents a change quantity of carrier phase integer ambiguity at k and k-1 epochs, σ represents noise of carrier phase observation value, and ε D represents noise of Doppler observation value;

[0067] The threshold value for high-precision Doppler carrier phase cycle slip detection is set as follows:

[0068]

[0069] wherein ξ represents the threshold value for carrier phase cycle slip detection; when the observation value sampling interval is 1 second, ξ is set to 1 cycle, and a cycle slip of more than 1 cycle can be detected by high-precision Doppler;

[0070] The carrier-to-noise ratio threshold test formula S is constructed obs The observation of poor satellite signals is directly rejected:

[0071]

[0072] wherein C / N0 represents the carrier-to-noise ratio of the received satellite signal, and Ele represents the elevation angle of the satellite; if the elevation angle of the satellite is greater than 15° and the carrier-to-noise ratio is greater than 30 dB-Hz, the current observation value is considered valid, and the observation data that is not within the valid range indicates that the satellite signal cannot be effectively tracked and is rejected.

[0073] Further, the prior random model construction unit is further configured to:

[0074] The code-minus-carrier phase model is constructed by the GNSS basic observation equation to obtain the pseudorange measurement error;

[0075] The basic observation equation on the GNSS propagation path from satellite signal transmission to reception is as follows:

[0076]

[0077] wherein ρ represents the geodetic distance, c represents the speed of light in vacuum, dt represents the receiver clock error, dT represents the satellite clock error, I represents the ionospheric delay of the slant path, T represents the tropospheric delay of the slant path, λ represents the carrier frequency of the satellite transmitted signal, N represents the integer ambiguity of the carrier phase, and M P represents the pseudorange multipath error, represents the carrier phase multipath error, and ε P represents the pseudorange measurement noise, represents the pseudorange measurement noise;

[0078] According to formula (5), the expression of the code-minus-carrier phase model (CMC, Code-minus-Carrier phase) is obtained:

[0079]

[0080] The code-minus-carrier phase residual is obtained by taking the average in the continuous arc segment:

[0081]

[0082] A prior stochastic model based on the carrier-to-noise ratio (CNR) is obtained by fitting the correlation between pseudorange measurement error and CNR; the expression of the prior stochastic model based on CNR is as follows:

[0083]

[0084] in, and The empirical values ​​are 0.1–0.3 and 0.001–0.003; v and c are the coefficients to be fitted for the carrier-to-noise ratio stochastic model, calibrated based on the correlation between pseudorange measurement error CMCR and C / N0. The function expression to be fitted is:

[0085]

[0086] Among them, CMCR and C / N0 are obtained from actual measurement data, and v and c are parameters to be fitted. In order to make the random model coefficients v and c based on the carrier-to-noise ratio as consistent as possible with the quality of low-cost terminal GNSS observation data, the best fitting coefficients are searched based on the minimum residual sum of squares criterion during the fitting process.

[0087] Taking the logarithm of both sides of equation (9), we obtain the objective function e in the following equation. 2 :

[0088]

[0089] Where k and N represent the k-th measurement and the total number of measurements, respectively, when the objective function e 2 The optimal fitting parameters for v and c are obtained when the minimum value is reached.

[0090] Furthermore, the positioning result calculation unit is further configured as follows:

[0091] The observations are assigned initial weights by a prior random model, and the PPP a priori residuals are obtained after filtering and solving.

[0092] Linearizing the basic GNSS observation equation (5) yields the error equation for the observation vector at the k-th epoch:

[0093]

[0094] Among them, B k Represents the design matrix. L represents the vector of parameters to be estimated. k V represents the vector of observed values ​​minus calculated values. k Let represent the post-hoc residual vector obtained by filtering. Solving equation (11) according to the weighted least squares principle yields:

[0095]

[0096] wherein P k is a prior weight matrix (diagonal matrix, diagonal elements are p kk , k = 1, 2, … n) calculated by formula (8) :

[0097]

[0098] wherein n represents the total number of observation values, the greater the variance of the observation values, the more random errors contained in the measurement data, and the smaller the weight of the corresponding observation values;

[0099] According to the residual threshold, the weights of the pseudo-range and carrier phase observation values are adaptively adjusted to obtain a more robust positioning result.

[0100] In order to adaptively adjust the weights of the pseudo-range and carrier phase observation values, a weight factor w is introduced to control the weights:

[0101]

[0102] wherein represents an equivalent weight matrix (diagonal elements are ); considering formula (12)-(14), an adaptive robust estimation is obtained:

[0103]

[0104] The posterior residual V k is normalized as and a piecewise function is constructed to adaptively adjust the weight factor:

[0105]

[0106] wherein c0 and c1 represent the lower bound and upper bound of the normalized residual; for a low-cost GNSS navigation terminal, the empirical values of the two coefficients are: c0 = 1.0-1.5, c1 = 2.5-5.0.

[0107] According to a third aspect of the present application, a non-transitory computer readable storage medium storing instructions is provided, when the instructions are executed by a processor, the method according to the embodiments of the present application is executed.

[0108] The adaptive adjustment method of the low-cost PPP random model for complex environment according to the various aspects of the present application has at least the following technical effects:

[0109] The application makes full use of the correlation of the measurement error of multi-system and multi-frequency GNSS observation values, constructs a carrier-to-noise ratio prior weight random model based on a code-minus-phase model, is more in line with the observation data quality change of a low-cost navigation terminal in a city environment, adaptively adjusts the observation value weights of the pseudo-range and carrier phase through a posterior residual threshold test, can significantly improve the anti-roughness ability of the model in a complex scene, effectively improves the positioning accuracy and convergence speed of the low-cost terminal PPP in a complex city environment, and guarantees the accuracy and continuity of positioning in real-time navigation services. BRIEF DESCRIPTION OF DRAWINGS

[0110] In the drawings, which are not necessarily drawn to scale, like numerals can describe similar components in different views. Like numerals having different letter suffixes can represent different instances of the components. The drawings illustrate generally by way of example, and not by way of limitation, various embodiments discussed herein, as presently described herein, and with reference to the claims. Wherever possible, the same reference numerals are used throughout the drawings to refer to the same or like components. Such embodiments do not represent all embodiments according to the disclosure, but rather they are illustrative thereof.

[0111] Figure 1 A flowchart of a low-cost PPP random model adaptive adjustment method for complex environments according to an embodiment of the application.

[0112] Figure 2 A carrier-to-noise ratio prior random model obtained by fitting a code-minus-phase model by a Huawei P40 low-cost terminal according to an embodiment of the application.

[0113] Figure 3 A carrier-to-noise ratio prior random model obtained by fitting a code-minus-phase model by a Xiaomi 8 low-cost terminal according to an embodiment of the application.

[0114] Figure 4 Positioning error sequences in east, north, and sky of a Huawei P40 low-cost terminal using adaptive random model PPP and traditional PPP according to an embodiment of the application.

[0115] Figure 5 Positioning error sequences in east, north, and sky of a Xiaomi 8 low-cost terminal using adaptive random model PPP and traditional PPP according to an embodiment of the application.

[0116] Figure 6 A structure diagram of a low-cost PPP random model adaptive adjustment device for complex environments according to an embodiment of the application. DETAILED DESCRIPTION

[0117] For those skilled in the art to better understand the technical solutions of the present application, the present application will be described in detail below in combination with the drawings and specific embodiments. The embodiments of the present application will be further described in detail below in combination with the drawings and specific embodiments, but not as a limitation of the present application. The order in which each step is described herein as an example should not be considered as a limitation, and those skilled in the art should know that the order can be adjusted as long as the logic between them is not destroyed and the whole process cannot be realized.

[0118] The embodiment of the present application provides a low-cost PPP random model adaptive adjustment method for complex environment, as shown in the figure, Figure 1 The present application first strictly controls the GNSS raw observation data of the low-cost navigation terminal in the complex environment, including pseudorange gross error detection, carrier cycle slip detection, carrier-to-noise ratio threshold test, etc., so as to obtain relatively clean GNSS observation values. Secondly, the observation value error is obtained through the code minus phase model, and the prior random model based on the carrier-to-noise ratio is obtained according to the correlation fitting of the code minus phase residual and the carrier-to-noise ratio; then, the PPP posterior residual is obtained by filtering, and the weights of the pseudorange and carrier phase observation values are adaptively adjusted according to the posterior residual threshold test to obtain more robust positioning results.

[0119] In order to make the technical solutions, objects and advantages of the present application clearer, the present embodiment uses two low-cost terminals to solve the GNSS observation data in the complex urban environment by the adaptive random model PPP method, and compares with the traditional PPP method. Specifically, a low-cost PPP random model adaptive adjustment method for complex environment is described, which comprises the following steps S1-S3:

[0120] S1: Strictly control the GNSS raw observation data of the low-cost terminal in the complex environment, including pseudorange gross error detection, carrier cycle slip detection and carrier-to-noise ratio threshold test, etc., so as to obtain clean GNSS observation values.

[0121] In some embodiments, the specific process of step S1 is:

[0122] S1-1: Due to the shielding of trees and buildings in the complex urban environment, the pseudorange observation value of the low-cost terminal is easy to produce large gross error. The Doppler shift is independent of the pseudorange and carrier phase observation value, which can ensure the continuity and stability of the data. According to the relationship between the pseudorange and the Doppler between epochs, the following pseudorange gross error detection model is constructed:

[0123] G P = (P k -P k-1 )-(D k +Dk-1 )·λ / 2,k≥2 (65)

[0124] where P k and P k-1 denote the pseudorange observations at epoch k and k-1, respectively, D k and D k-1 denote the Doppler shift observations at epoch k and k-1, respectively, G P is the pseudorange roughness test quantity. According to the variance-covariance error propagation rule, if |G P |>3σ P , the current pseudorange observation is considered to be rough. In this case, the weight of the pseudorange observation should be reduced or the observation should be rejected, σ P denotes the mean error of the pseudorange measurement.

[0125] S1-2: Low-cost terminals are equipped with low-end microstrip antennas, which are prone to cause the carrier phase tracking loop to lose lock frequently. Efficiently detecting carrier phase cycle slips is an important prerequisite to ensure the smooth convergence of PPP. The following carrier phase integer ambiguity test formula is constructed by high-precision Doppler:

[0126]

[0127] where ΔN denotes the change of the carrier phase integer ambiguity at epoch k and k-1, and denotes the noise of the carrier phase observation, and ε D denotes the noise of the Doppler observation.

[0128] Since the noises of the Doppler and the carrier phase are both in the order of centimeters, they can be ignored in the test process. Therefore, the threshold for detecting carrier phase cycle slips by high-precision Doppler is set as follows:

[0129]

[0130] where ξ denotes the threshold for detecting carrier phase cycle slips; when the observation sampling interval is 1 second, ξ can be set to 1 cycle, and a cycle slip of more than 1 cycle can be detected by high-precision Doppler.

[0131] S1-3: The carrier-to-noise ratio is one of the effective indicators that directly reflect the quality of GNSS data. The carrier-to-noise ratio threshold test formula S obs is constructed to directly reject satellite signals with poor observation quality:

[0132]

[0133] Wherein, C / N0 represents the carrier-to-noise ratio of the received satellite signal, and Ele represents the elevation angle of the satellite; if the elevation angle of the satellite is greater than 15° and the carrier-to-noise ratio is greater than 30 dB-Hz, it is considered that the current observation value is effective, and the observation data not in the effective range indicates that the satellite signal cannot be effectively tracked and is eliminated.

[0134] S2: Obtain the observation value error through the code-minus-carrier phase model, and fit the prior random model based on the carrier-to-noise ratio according to the correlation thereof.

[0135] In some embodiments, the specific process of step S2 is as follows:

[0136] S2-1: Construct a code-minus-carrier phase model through a GNSS basic observation equation to obtain a pseudorange measurement error;

[0137] The basic observation equation on the GNSS propagation path from satellite signal emission to reception is as follows:

[0138]

[0139] Wherein, ρ represents the geodetic distance, c represents the speed of light in a vacuum, dt represents the receiver clock error, dT represents the satellite clock error, I represents the ionospheric delay of the slant path, T represents the tropospheric delay of the slant path, λ represents the carrier frequency of the satellite transmitted signal, N represents the integer ambiguity of the carrier phase, and M P represents the pseudorange multipath error, represents the carrier phase multipath error, and ε P represents the pseudorange measurement noise, represents the pseudorange measurement noise;

[0140] According to formula (5), the expression of the code-minus-carrier phase model (CMC) can be obtained:

[0141]

[0142] Since the carrier phase multipath and noise are both in the order of centimeters, they can be ignored compared to the noise and multipath of the pseudorange observation value. In addition, the ionosphere hardly changes in a short time, and the carrier phase ambiguity can be regarded as a constant when there is no cycle slip, so the code-minus-carrier phase residual (CMCR) can be obtained by taking the average value in the continuous arc segment:

[0143]

[0144] S2-2: Obtain the prior random model based on the carrier-to-noise ratio according to the correlation fitting of the pseudorange measurement error and the carrier-to-noise ratio;

[0145] Carrier-to-Noise Ratio (C / N0) is an effective indicator of satellite signal quality. When the C / N0 of the received satellite signal is low, it indicates that the measurement error of the current satellite observation is large. Therefore, the prior random model expression based on C / N0 is constructed as follows:

[0146]

[0147] wherein, and The empirical values of and are 0.1-0.3 and 0.001-0.003, respectively, and can also be calibrated by zero baseline station double-difference; v and c are the coefficients to be fitted of the C / N0 random model, which can be calibrated according to the correlation between the pseudorange measurement error CMCR and C / N0, and the function expression to be fitted is:

[0148]

[0149] wherein, CMCR and C / N0 are obtained from actual measurement data, and v and c are the fitting parameters; in order to make the random model coefficients v and c based on C / N0 as much as possible to meet the GNSS observation data quality of low-cost terminals, the best fitting coefficients are searched based on the least squares criterion in the fitting process. Taking the logarithm of both sides of equation (9), the objective function e 2 of the following equation is obtained:

[0150]

[0151] wherein, k and N represent the kth measurement value and the total measurement value, respectively, and the optimal fitting parameters of v and c can be obtained when the objective function e 2 is the minimum. Huawei P40 and Xiaomi 8 low-cost navigation terminals are selected for experiments, and the recorded multi-system GNSS observations and C / N0 are fitted according to equations (9) and (10) to obtain the random model coefficients. Figure 2 and Figure 3 are the C / N0 prior random model calibration results of Huawei P40 and Xiaomi 8, respectively. It can be seen from the relationship between code-minus-phase residuals and C / N0 that the measurement error of low-cost navigation terminals has a strong correlation with C / N0, and the higher the C / N0, the smaller the corresponding measurement error. In addition, the measurement accuracy of each satellite system is not consistent, and GPS, BDS-2, BDS-3 and Galileo systems have obvious anti-roughness ability compared with GLONASS system, which also shows the advantage of the method proposed in the present application over the traditional empirical random model, and the method can better reflect the characteristics of multi-system and multi-frequency GNSS observations of low-cost terminals.

[0152] S3: obtain the PPP posterior residuals by filtering, and adaptively adjust the weights of pseudorange and carrier phase observations according to the residual threshold to obtain more robust positioning results.

[0153] In some embodiments, the specific process of step S3 is as follows:

[0154] S3-1: Assign an initial weight to the observation value through a priori random model, and obtain the PPP posterior residual after filtering solution;

[0155] The error equation of the kth epoch observation vector is obtained after linearizing the GNSS basic observation equation (5):

[0156]

[0157] where B k represents the design matrix, represents the parameter vector to be estimated, L k represents the observation value minus the calculated value vector, V k represents the posterior residual vector obtained by filtering solution. According to the weighted least squares principle, the solution of equation (11) can be obtained:

[0158]

[0159] In the formula, P k is the a priori weight matrix (diagonal matrix, the diagonal element is p kk , k = 1, 2, … n) calculated by equation (8):

[0160]

[0161] In the formula, n represents the total number of observations. The larger the variance of the observation value, the more random errors the measurement data contains, and the smaller the weight of the corresponding observation value;

[0162] S3-2: Adjust the weights of pseudorange and carrier phase observation values adaptively according to the residual threshold to obtain more robust positioning results;

[0163] In order to realize the adaptive adjustment of the weights of pseudorange and carrier phase observation values, a weight factor w is introduced to control the weight:

[0164]

[0165] In the formula, represents the equivalent weight matrix (the diagonal element is ); considering equations (12)-(14), the adaptive robust estimation can be obtained:

[0166]

[0167] The key to robust positioning results lies in the dynamic adjustment of the value of the weight factor. In addition, in order to uniformly process the weights of pseudorange and carrier phase, the posterior residual Vk normalized as and construct a piecewise function to adaptively adjust the weight factor:

[0168]

[0169] wherein c0 and c1 represent the lower bound and upper bound of the normalized residual; the empirical values of the two coefficients for low-cost GNSS navigation terminals are: c0 = 1.0-1.5, c1 = 2.5-5.0.

[0170] In this embodiment, c0 and c1 take values of 1.0 and 3.0 respectively, and the traditional random model takes values of v = 10 and c = 150 according to experience 2 . The coefficients v and c of the adaptive random model are calibrated according to equations (9) and (10) (such as Figure 2 and Figure 3 ), and the weight factor w is dynamically changed according to equation (16) in real-time calculation to realize the adjustment of the adaptive random model, so as to more reasonably allocate the GNSS error weight. In this embodiment, Huawei P40 and Xiaomi 8 low-cost terminals are used to perform adaptive random model PPP method calculation in urban complex environment, and the convergence speed and positioning accuracy are compared with the traditional PPP method. Figure 4 and Figure 5 are the PPP positioning error sequences of the two low-cost terminals in a selected urban environment using the adaptive random model and the traditional random model respectively. Obviously, compared with the traditional random model, the positioning accuracy and convergence speed of the adaptive random model proposed in the present application in the east, north and sky three-dimensional directions are significantly improved and the convergence speed is faster. In addition, the three-dimensional positioning accuracy of the adaptive random model PPP after convergence is better than 0.5 meters, which can meet the sub-meter positioning demand of low-cost navigation terminals in urban complex environment.

[0171] In summary, by using the positioning method proposed in the present application, the terminal can obtain more robust PPP calculation results without the need of modifying the hardware. Compared with the traditional PPP method, the adaptive random model PPP method disclosed in the present application has significant improvement in positioning accuracy and convergence speed in urban complex environment, which can further improve the urban navigation and positioning experience of the general public.

[0172] As shown in Figure 6 , the embodiment of the present application also provides a low-cost PPP random model adaptive adjustment device for complex environment, which comprises:

[0173] a quality control unit 601 configured to perform quality control on GNSS raw observation data of the low-cost terminal in a complex environment according to quality control parameters to obtain GNSS observation values, the quality control parameters including pseudorange gross error detection, carrier cycle slip detection, and carrier ratio threshold test;

[0174] a priori random model construction unit 602 configured to obtain observation value error through a code-minus-phase model and fit a priori random model based on carrier-to-noise ratio according to a correlation between the observation value error and the carrier-to-noise ratio;

[0175] a positioning result calculation unit 603 configured to obtain PPP posteriori residual through filtering and adaptively adjust weights of pseudorange and carrier phase observation values according to residual threshold test to obtain a positioning result.

[0176] In some embodiments, the quality control unit 601 is further configured to:

[0177] According to the relationship between interepoch pseudorange and Doppler, a pseudorange gross error detection model is constructed as follows:

[0178] G P = (P k -P k-1 )-(D k +D k-1 )·λ / 2,k≥2 (81)

[0179] wherein P k and P k-1 represent pseudorange observation values at the kth and (k-1)th epochs, D k and D k-1 represent Doppler shift observation values at the kth and (k-1)th epochs, G P is a pseudorange gross error test quantity, according to the variance-covariance error propagation rule, if |G P |>3σ P , it is considered that the current pseudorange observation value has a gross error, at which time the weight of the pseudorange observation value is reduced or the observation value is removed, and σ P represents the mean error of the pseudorange measurement error;

[0180] A carrier phase integer ambiguity test formula is constructed as follows through high-precision Doppler:

[0181]

[0182] wherein ΔN represents the change of carrier phase integer ambiguity at the kth and (k-1)th epochs, and represents the noise of the carrier phase observation value, and ε D represents the noise of the Doppler observation value;

[0183] The threshold value for high-precision Doppler carrier phase cycle slip detection is set as follows:

[0184]

[0185] wherein ξ represents the threshold value for carrier phase cycle slip detection; when the observation value sampling interval is 1 second, ξ is set to 1 cycle, and a cycle slip of more than 1 cycle can be detected by high-precision Doppler;

[0186] The carrier-to-noise ratio threshold test formula S is constructed obs Satellite signals with poor observation quality are directly rejected:

[0187]

[0188] wherein C / N0 represents the carrier-to-noise ratio of the received satellite signal, and Ele represents the elevation angle of the satellite; if the elevation angle of the satellite is greater than 15° and the carrier-to-noise ratio is greater than 30 dB-Hz, the current observation value is considered valid, and observation data outside the valid range indicates that the satellite signal cannot be effectively tracked and is rejected.

[0189] In some embodiments, the prior random model construction unit 602 is further configured to:

[0190] The code-minus-carrier phase model is constructed through the GNSS basic observation equation to obtain the pseudorange measurement error;

[0191] The basic observation equation on the GNSS propagation path from satellite signal transmission to reception is as follows:

[0192]

[0193] wherein ρ represents the geodetic distance, c represents the speed of light in a vacuum, dt represents the receiver clock error, dT represents the satellite clock error, I represents the ionospheric delay of the slant path, T represents the tropospheric delay of the slant path, λ represents the carrier frequency of the satellite transmitted signal, N represents the integer ambiguity of the carrier phase, and M P represents the pseudorange multipath error, represents the carrier phase multipath error, and ε P represents the pseudorange measurement noise, represents the pseudorange measurement noise;

[0194] According to formula (5), the expression of the code-minus-carrier phase model (CMC, Code-minus-Carrier phase) is obtained:

[0195]

[0196] The code-minus-carrier phase residual is obtained by taking the average value in the continuous arc segment:

[0197]

[0198] A prior stochastic model based on the carrier-to-noise ratio (CNR) is obtained by fitting the correlation between pseudorange measurement error and CNR; the expression of the prior stochastic model based on CNR is as follows:

[0199]

[0200] in, and The empirical values ​​are 0.1–0.3 and 0.001–0.003; v and c are the coefficients to be fitted for the carrier-to-noise ratio stochastic model, calibrated based on the correlation between pseudorange measurement error CMCR and C / N0. The function expression to be fitted is:

[0201]

[0202] Among them, CMCR and C / N0 are obtained from actual measurement data, and v and c are parameters to be fitted. In order to make the random model coefficients v and c based on the carrier-to-noise ratio as consistent as possible with the quality of low-cost terminal GNSS observation data, the best fitting coefficients are searched based on the minimum residual sum of squares criterion during the fitting process.

[0203] Taking the logarithm of both sides of equation (9), we obtain the objective function e in the following equation. 2 :

[0204]

[0205] Where k and N represent the k-th measurement and the total number of measurements, respectively, when the objective function e 2 The optimal fitting parameters for v and c are obtained when the minimum value is reached.

[0206] In some embodiments, the positioning result calculation unit 603 is further configured to:

[0207] The observations are assigned initial weights by a prior random model, and the PPP a priori residuals are obtained after filtering and solving.

[0208] Linearizing the basic GNSS observation equation (5) yields the error equation for the observation vector at the k-th epoch:

[0209]

[0210] Among them, B k Represents the design matrix. L represents the vector of parameters to be estimated. k V represents the vector of observed values ​​minus calculated values. k Let represent the post-hoc residual vector obtained by filtering. Solving equation (11) according to the weighted least squares principle yields:

[0211]

[0212] In the formula, P k is a priori weight matrix (diagonal matrix, diagonal elements are p kk , k = 1, 2, … n) calculated by formula (8):

[0213]

[0214] In the formula, n represents the total number of observation values, the greater the variance of the observation values, the more random errors contained in the measurement data, and the smaller the corresponding observation value weight;

[0215] According to the residual threshold, the weights of the pseudo-range and carrier phase observation values are adaptively adjusted to obtain a more robust positioning result;

[0216] In order to adaptively adjust the weights of the pseudo-range and carrier phase observation values, a weight factor w is introduced for weight control:

[0217]

[0218] In the formula, represents an equivalent weight matrix (diagonal elements are ); considering formula (12)-(14), an adaptive robust estimation is obtained:

[0219]

[0220] The posterior residual V k is normalized as and a piecewise function is constructed to adaptively adjust the weight factor:

[0221]

[0222] Wherein, c0 and c1 represent the lower bound and upper bound of the normalized residual; for a low-cost GNSS navigation terminal, the empirical values of the two coefficients are: c0 = 1.0-1.5, c1 = 2.5-5.0.

[0223] It should be noted that the device provided by the embodiment of the application and the method provided in the prior art belong to the same technical idea, have the same technical principle and can play the same technical effect, and details are not repeated here.

[0224] The embodiment of the application also provides a non-transitory computer readable storage medium storing instructions, when the instructions are executed by a processor, the method according to the embodiments of the application is executed.

[0225] Furthermore, although example embodiments have been described herein, the scope includes any and all embodiments having equivalent elements, modifications, omissions, combinations (e.g., of

[0226] The foregoing description is intended to be illustrative and not exclusive. For example, the above-described examples (or one or more aspects thereof) can be used in combination with each other. Other embodiments can be utilized, such as would be apparent to one of ordinary skill in the art, upon reading the above description. Additionally, in the specific description of embodiments above, various features can be grouped together in one or more embodiments. This should not be interpreted as an intention that an unclaimed application claim requires features of a claim to group the claimed application's features. Rather, inventive subject matter can be less than all features of a specific embodiment. Accordingly, the claims as follows are hereby expressly incorporated into this detailed description, with each claim standing on its own as a separate embodiment, and it is made expressly clear that such embodiments can be combined with each other in various combinations or permutations. The scope of the application should be determined by reference to the appended claims, along with the full scope of equivalents to which such claims are entitled.

Claims

1. A method for adaptive adjustment of a low-cost PPP stochastic model for complex environments, characterized in that, The method comprises: According to the quality control parameters, the GNSS original observation data of the low-cost terminal in the complex environment is subjected to quality control to obtain GNSS observation values, and the quality control parameters comprise pseudo-range rough error detection, carrier phase jump detection and carrier ratio threshold test; The observation value error is obtained through the code-minus-phase model, and a priori random model based on the carrier-to-noise ratio is fitted according to the correlation between the observation value error and the carrier-to-noise ratio; PPP posteriori residuals are obtained through filtering solution, and the weights of the pseudo-range and carrier phase observation values are adaptively adjusted according to the residual threshold test to obtain a positioning result; The quality control parameters are used to control the GNSS original observation data of the low-cost terminal in the complex environment to obtain GNSS observation values, and the quality control parameters comprise pseudo-range rough error detection, carrier phase jump detection and carrier ratio threshold test. A pseudo-range rough error detection model is constructed according to the relationship between the pseudo-range and the Doppler between epochs as follows: G P = (P k -P k-1 ) - (D k + D k-1 ) · λ / 2, k ≥ 2 (1) wherein P k and P k-1 represent the pseudo-range observations at the kth and (k-1)th epochs, respectively, D k and D k-1 represent the Doppler shift observations at the kth and (k-1)th epochs, respectively, G P is a pseudo-range gross error test quantity, according to the variance-covariance error propagation rule, if |G P | > 3σ P , it is considered that the current pseudo-range observation has a gross error, at which time the weight of the pseudo-range observation is reduced or the observation is rejected, and σ P represents the mean error of the pseudo-range measurement error; A carrier phase integer ambiguity test formula is constructed through high-precision Doppler as follows: where ΔN denotes the change of the carrier phase integer ambiguity between the kth and (k-1)th epochs, denotes the noise of the carrier phase observation, ε D denotes the noise of the Doppler observation; The threshold for detecting the carrier phase jump through high-precision Doppler is set as follows: Wherein, ξ represents the threshold for detecting the carrier phase jump; when the observation value sampling interval is 1 second, ξ is set to 1 cycle, and a jump of more than one cycle can be detected through high-precision Doppler; Constructing the carrier-to-noise ratio threshold test formula S obs Directly rejecting satellite signals with poor observation quality: Wherein, C / N0 represents the carrier ratio of the received satellite signal, and Ele represents the elevation angle of the satellite; if the elevation angle of the satellite is greater than 15° and the carrier-to-noise ratio is greater than 30 dB-Hz, it is considered that the current observation value is effective, and the observation data that is not within the effective range is considered to be unable to effectively track the satellite signal and is eliminated; The observation value error is obtained through the code-minus-phase model, and a priori random model based on the carrier-to-noise ratio is fitted according to the correlation between the observation value error and the carrier-to-noise ratio. The pseudo-range measurement error is obtained through the code-minus-phase model constructed by the GNSS basic observation equation; The basic observation equation on the GNSS propagation path from the satellite signal transmission to the reception process is as follows: where p represents the geodetic distance, c represents the speed of light in vacuum, dt represents the receiver clock error, dT represents the satellite clock error, I represents the ionospheric delay of the slant path, T represents the tropospheric delay of the slant path, l represents the carrier frequency of the satellite transmitted signal, N represents the integer ambiguity of the carrier phase, M P represents the pseudorange multipath error, represents the carrier phase multipath error, e P represents the pseudorange measurement noise; The expression of the code-minus-phase model is obtained according to formula (5): The code-minus-phase residual is obtained by taking the mean value in the continuous arc segment: The priori random model based on the carrier-to-noise ratio is fitted according to the correlation between the pseudo-range measurement error and the carrier-to-noise ratio; the expression of the priori random model based on the carrier-to-noise ratio is as follows: wherein, and The empirical values of 0.1-0.3 and 0.001-0.003; v and c are the coefficients to be fitted of the carrier-to-noise ratio random model, which are calibrated according to the correlation between the pseudo-range measurement error CMCR and C / N0, and the function expression to be fitted is: Wherein, CMCR and C / N0 are obtained from the actual measurement data, and v and c are to-be-fitted parameters; in order to make the random model coefficients v and c based on the carrier-to-noise ratio as much as possible to conform to the GNSS observation data quality of the low-cost terminal, the best fitting coefficients are searched based on the residual sum of squares minimum criterion during the fitting process; Taking the logarithm of both sides of equation (9), the objective function e of the following equation can be obtained 2 : where k and N represent the kth measurement and the total number of measurements, respectively, and the objective function e 2 The optimal fitting parameters v and c are obtained at the minimum. PPP posteriori residuals are obtained through filtering solution, and the weights of the pseudo-range and carrier phase observation values are adaptively adjusted according to the residual threshold test to obtain a positioning result. The initial weight of the observation value is given through the priori random model, and the PPP posteriori residuals are obtained after the filtering solution; The error equation of the kth epoch observation vector is obtained after the GNSS basic observation equation (5) is linearized: where B k denotes the design matrix, denotes the parameter vector to be estimated, L k denotes the observation minus computed value vector, V k denotes the posterior residual vector obtained by filtering, and solving equation (11) according to the weighted least squares principle can obtain: where P k is the prior weight matrix calculated from equation (8): In the formula, n represents the total number of observation values, the larger the variance of the observation value, the more random errors the measurement data contains, and the smaller the weight of the corresponding observation value. The weight of the pseudo-range and carrier phase observation value is adaptively adjusted according to a residual threshold value to obtain a more robust positioning result; In order to adaptively adjust the weight of the pseudo-range and carrier phase observation value, a weight factor w is introduced to control the weight: where denotes the equivalent weight matrix, with diagonal elements Taking into account equations (12)-(14) gives the adaptive robust estimate: The post-test residual V k is normalized to and a piecewise function is constructed to adaptively adjust the weight factor: Wherein, c0 and c1 represent the lower bound and upper bound of the normalized residual; the empirical values of the two coefficients for the low-cost GNSS navigation terminal are: c0=1.0~1.5, c1=2.5~5.

0.

2. A low-cost PPP stochastic model adaptive adjustment device for complex environments, characterized in that, The device comprises: A quality control unit configured to perform quality control on GNSS raw observation data of the low-cost terminal in a complex environment according to a quality control parameter to obtain GNSS observation values, the quality control parameter comprising pseudo-range gross error detection, carrier cycle slip detection and carrier ratio threshold value test; A priori random model construction unit configured to obtain observation value error through a code-minus-phase model and to obtain a priori random model based on carrier-to-noise ratio (C / N0) by fitting the correlation between the observation value error and the C / N0; A positioning result calculation unit configured to obtain PPP posteriori residual by filtering and to adaptively adjust the weight of the pseudo-range and carrier phase observation value according to a residual threshold value to obtain a positioning result; The quality control unit is further configured to: According to the relationship between the inter-epoch pseudo-range and the Doppler, a pseudo-range gross error detection model is constructed as follows: G P \(P k -P k-1 )-(D k +D k-1 )·λ / 2,k≥2 (1) wherein P k and P k-1 represent the pseudo-range observations at the kth and (k-1)th epochs, respectively, D k and D k-1 represent the Doppler shift observations at the kth and (k-1)th epochs, respectively, G P is a pseudo-range rough error check quantity, according to the variance-covariance error propagation rule, if |G P | > 3σ P , it is considered that the current pseudo-range observation has a rough error, at which time the weight of the pseudo-range observation is reduced or the observation is rejected, and σ P represents the mean error of the pseudo-range measurement error. A carrier phase integer ambiguity test formula is constructed as follows through high-precision Doppler: where ΔN denotes the change of the carrier phase integer ambiguity between the kth and (k-1)th epochs, denotes the noise of the carrier phase observation, ε D denotes the noise of the Doppler observation; The threshold value for detecting carrier phase cycle slip through high-precision Doppler is set as follows: Wherein, ξ represents the threshold value for carrier phase cycle slip detection; when the observation value sampling interval is 1 second, ξ is set to 1 cycle, and a cycle slip of more than 1 cycle can be detected through high-precision Doppler; Constructing the carrier-to-noise ratio threshold test formula S obs Directly rejecting satellite signals with poor observation quality: Wherein, C / N0 represents the carrier ratio of the received satellite signal, and Ele represents the elevation angle of the satellite; if the elevation angle of the satellite is greater than 15° and the carrier-to-noise ratio is greater than 30 dB-Hz, the current observation value is considered to be effective, and the observation data that is not within the effective range is considered to be invalid and is excluded; The a priori random model construction unit is further configured to: A code-minus-phase model is constructed through the GNSS basic observation equation to obtain the pseudo-range measurement error; The basic observation equation in the GNSS propagation path from the satellite signal transmission to the reception process is as follows: where p represents the geodetic distance, c represents the speed of light in vacuum, dt represents the receiver clock error, dT represents the satellite clock error, I represents the ionospheric delay of the slant path, T represents the tropospheric delay of the slant path, l represents the carrier frequency of the satellite transmitted signal, N represents the integer ambiguity of the carrier phase, M P represents the pseudorange multipath error, represents the carrier phase multipath error, e P represents the pseudorange measurement noise; The expression of the code-minus-phase model is obtained according to formula (5): The code-minus-phase residual is obtained by taking the mean value in the continuous arc segment: An a priori random model based on C / N0 is obtained by fitting the correlation between the pseudo-range measurement error and the C / N0; the expression of the a priori random model based on C / N0 is as follows: wherein, and The empirical values of 0.1-0.3 and 0.001-0.003; v and c are the coefficients to be fitted of the carrier-to-noise ratio random model, which are calibrated according to the correlation between the pseudo-range measurement error CMCR and C / N0, and the function expression to be fitted is: Wherein, CMCR and C / N0 are obtained from actual measurement data, and v and c are to-be-fitted parameters; in order to make the random model coefficients v and c based on C / N0 as much as possible to conform to the GNSS observation data quality of the low-cost terminal, the best fitting coefficients are searched based on the least square sum of residual in the fitting process; Taking the logarithm of both sides of equation (9), the objective function e of the following equation can be obtained 2 : where k and N represent the kth measurement and the total number of measurements, respectively, and the objective function e 2 The optimal fitting parameters v and c are obtained at the minimum. The positioning result calculation unit is further configured to: An initial weight is given to the observation value through the a priori random model, and the PPP posteriori residual is obtained after filtering; The error equation of the kth epoch observation vector is obtained after linearizing the GNSS basic observation equation (5): where B k denotes the design matrix, denotes the parameter vector to be estimated, L k denotes the observation minus computed value vector, V k denotes the posterior residual vector obtained by filtering, and solving equation (11) according to the weighted least squares principle can obtain: where P k is the prior weight matrix calculated from equation (8): In the formula, n represents the total number of observation values, the greater the variance of the observation values, the more random errors the measured data contains, and the smaller the corresponding observation value weight is; According to the residual threshold, the weights of the pseudo-range and carrier phase observation values are adaptively adjusted to obtain a more robust positioning result; In order to adaptively adjust the weights of the pseudo-range and carrier phase observation values, a weight factor w is introduced for weight control: where denotes the identity matrix (diagonal elements are ) and taking into account equations (12)-(14) the adaptive robust estimate is obtained: The post-test residual V k is normalized to and a piecewise function is constructed to adaptively adjust the weight factor: Wherein, c0 and c1 represent the lower bound and upper bound of the normalized residual; for a low-cost GNSS navigation terminal, the empirical values of the two coefficients are: c0=1.0~1.5, c1=2.5~5.

0.

3. A non-transitory computer-readable storage medium storing instructions that, when executed by a processor, perform the method of claim 1.

Images