A quality control method for pseudo-satellite observations for deformation monitoring

Abstract

The present invention relates to a pseudo-satellite observation quality control method for deformation monitoring, comprising the following steps: establishing GNSS / IMU combined positioning based on ZUPT constraints; fault detection, including detecting the position fault of the pseudo-satellite itself and introducing the Euclidean distance matrix to detect the pseudo-satellite observation value fault; executing decisions on the pseudo-satellite based on the detection results; introducing the predicted observation value correction amount to correct the pseudo-range and carrier phase observation values; RTK solution. The present invention uses deep learning to model the pseudo-satellite and GNSS satellite multipath errors using historical data, establishes the connection between the multipath error and the satellite position, the satellite signal quality and the environmental water level, realizes the prediction and correction of the observation value error, improves the positioning accuracy of the pseudo-satellite GNSS combination system, and enhances the deformation monitoring capability and reliability.

Description

Technical Field

[0001] The invention relates to the field of surveying and mapping science and technology, and in particular to a pseudo-satellite observation quality control method for deformation monitoring. Background Art

[0002] GNSS deformation monitoring technology has been widely used in engineering implementation and safety monitoring of bridges, dams, landslides, buildings, etc. due to its advantages of high degree of automation and high-precision all-weather monitoring. However, due to the GNSS signal obstruction in the high mountain canyon environment where some bridges, dams, and landslides are located, it will lead to adverse effects such as poor satellite visibility, GNSS signal multipath interference, and non-line-of-sight reception, which ultimately limits the RTK positioning accuracy and causes certain damage to the GNSS deformation monitoring capability.

[0003] As an important supplement to navigation satellites, pseudolites can improve the geometric distribution of satellites and thus improve positioning accuracy in poor GNSS satellite observation conditions or GNSS denial environments. For this purpose, ground-based pseudolites are usually deployed to supplement low-altitude angle observation information, optimize positioning capabilities and improve positioning accuracy. However, the position of pseudolites themselves may change in long-term monitoring operations in environments such as dams, bridges, and mountain slopes, resulting in abnormal deviations in observation values. In addition, the observation environment often causes serious multipath errors near water areas, resulting in observation value errors and other problems, which reduces the positioning accuracy and reliability of ground-based pseudolites augmentation systems, and brings certain obstacles to the development of high-precision deformation monitoring of GNSS pseudolites. Summary of the invention

[0004] In view of the above-mentioned shortcomings of the prior art, an object of the present invention is to provide a pseudo-satellite observation quality control method for deformation monitoring to solve one or more problems in the prior art.

[0005] To achieve the above object, the technical solution of the present invention is as follows:

[0006] A pseudo-satellite observation quality control method for deformation monitoring comprises the following steps:

[0007] Establish GNSS / IMU combined positioning based on ZUPT constraints;

[0008] Fault detection, including detecting the position fault of pseudolite itself and introducing Euclidean distance matrix to detect the fault of pseudolite observation value;

[0009] executing decisions on pseudolites based on the detection results;

[0010] Introduce predicted observation value correction to correct pseudorange and carrier phase observation values;

[0011] RTK solution.

[0012] Furthermore, the steps of establishing GNSS / IMU combined positioning based on ZUPT constraints are as follows:

[0013] Perform RTK positioning on GNSS satellite data to obtain the pseudo-satellite receiving end position P PL ;

[0014] The pseudo-satellite inertial measurement unit (IMU) is used to collect information and mechanically arrange it to obtain the estimated position P. IMU , speed V IMU , attitude information A IMU Then it is fused through the extended Kalman filter EKF.

[0015] Furthermore, the detection of the position failure of the pseudolite itself comprises the following steps:

[0016] Calibrate the initial pseudo-satellite receiving antenna position P ini ;

[0017] Solving for P ini and the position P of the pseudo-satellite signal receiving antenna after EKF k The pseudo-satellite's suspected position △P;

[0018] Execute the judgment, if △P ≥ 5cm, it is considered that the pseudo-satellite has its own displacement, otherwise it does not exist.

[0019] Furthermore, the method of introducing the Euclidean distance matrix to detect pseudolite observation value failures includes the following steps:

[0020] The Euclidean distance matrix is ​​constructed based on the pseudo-satellite transmitter position and the pseudo-range observations obtained by the monitoring station receiver.

[0021] Transform Euclidean distance matrix is the corresponding Gram matrix

[0022] For Gram matrix Perform singular value decomposition;

[0023] Execute judgment. If the test statistic is greater than the detection threshold, it indicates that there are faulty pseudolites among the existing pseudolites. It is necessary to detect and exclude the faulty pseudolites. Repeat the detection and exclusion until the test statistic of the remaining pseudolites is lower than the detection threshold. It is considered that there are no pseudolites with observation value faults among the remaining pseudolites.

[0024] Furthermore, the transformed Euclidean distance matrix is the corresponding Gram matrix The conversion formula is as follows:

[0025]

[0026] Where: L represents the geometric center matrix, I represents the identity matrix, and 1 represents the column vector of 1.

[0027] Furthermore, the judgment formula for executing the judgment is as shown below:

[0028]

[0029] Where: i For the matrix The singular values ​​of , d is the dimension of the state space, is the test statistic, β thre It is a manually set detection threshold.

[0030] Further, the decision making on the pseudolite based on the detection result includes the following situations:

[0031] For pseudo-satellites that have position faults but no observation faults, the observation value retention operation is allowed, but the alarm information needs to be sent through wireless transmission;

[0032] For pseudolites with faulty positions and observation values, these pseudolites are directly excluded from the solution.

[0033] For pseudolites whose positions are not faulty but whose observation values ​​are faulty, they are given a downgraded weighting operation and their observation value weights are adjusted to 0.5 during the solution.

[0034] Furthermore, the step of introducing the correction amount of the predicted observation value is as follows:

[0035] Build data sets based on historical data;

[0036] The samples are calibrated using the inverted multipath error value;

[0037] An observation value correction prediction model is established based on integrated bagged regression trees.

[0038] Furthermore, the inverted multipath error value includes pseudorange residuals and carrier phase fractional part residuals of the GNSS satellite and pseudosatellite respectively.

[0039] Furthermore, the establishment of the observation value correction prediction model based on the integrated bagged regression tree includes the following steps:

[0040] Subsample single regression tree model generation;

[0041] Generate a training model based on all regression trees;

[0042] Prediction models are established for different GNSS constellations and pseudo-satellites.

[0043] Compared with the prior art, the beneficial technical effects of the present invention are as follows:

[0044] (I) The existing methods for detecting and eliminating pseudo-satellite faults in indoor positioning environments cannot well meet the needs of detecting and eliminating pseudo-satellite faults in outdoor environments such as bridges, dams, and mountain slopes under deformation monitoring applications. This patent performs pseudo-satellite displacement self-checking based on zero-speed correction constraints and implements pseudo-satellite fault detection based on the Euclidean distance matrix. It performs mutual check of satellite observations without the need for reference satellites, and downgrades or eliminates pseudo-satellites with different fault conditions, thereby realizing observation quality control of the pseudo-satellite GNSS combination system.

[0045] (ii) In response to the problem that the current ground-based pseudo-satellite augmentation system does not model and correct the observation value errors, the present invention uses deep learning to model the multipath errors of pseudo-satellite and GNSS satellites using historical data, establishes the relationship between multipath errors and satellite positions, satellite signal quality and environmental water levels, realizes the prediction and correction of observation value errors, improves the positioning accuracy of the pseudo-satellite GNSS combination system, and enhances the deformation monitoring capability and reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic flow chart of a pseudolite observation quality control method for deformation monitoring according to an embodiment of the present invention is shown.

[0047] Figure 2 A schematic diagram of a Euclidean distance matrix of a pseudo-satellite observation quality control method for deformation monitoring according to an embodiment of the present invention is shown.

[0048] Figure 3 A process diagram of an integrated bagged regression tree algorithm of a pseudolite observation quality control method for deformation monitoring according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical scheme and advantages of the present invention clearer, the following is a further detailed description of a pseudo-satellite observation quality control method for deformation monitoring proposed by the present invention in combination with the accompanying drawings and specific embodiments. According to the following description, the advantages and features of the present invention will be clearer. It should be noted that the accompanying drawings are in a very simplified form and use non-precise proportions, which are only used to conveniently and clearly assist in explaining the purpose of the embodiments of the present invention. In order to make the purpose, features and advantages of the present invention more obvious and easy to understand, please refer to the accompanying drawings. It should be noted that the structure, proportion, size, etc. illustrated in the drawings of this specification are only used to match the content disclosed in the specification for people familiar with this technology to understand and read, and are not used to limit the limiting conditions for the implementation of the present invention, so they have no technical substantive significance. Any modification of the structure, change of the proportional relationship or adjustment of the size, without affecting the effect that the present invention can produce and the purpose that can be achieved, should still fall within the scope of the technical content disclosed by the present invention.

[0050] See also Figure 1 , a pseudo-satellite observation quality control method for deformation monitoring, comprising the following steps:

[0051] Step 1: Establish GNSS / IMU combined positioning based on ZUPT constraints. The specific steps are as follows:

[0052] Step 1.1: Use the GNSS satellite data obtained by the reference station and the pseudo-satellite receiver to perform RTK positioning to obtain the pseudo-satellite receiver position P PL .

[0053] Step 1.2: Collect specific force and acceleration information through the inertial measurement unit (IMU) installed on the pseudo-satellite and perform mechanical arrangement to obtain the IMU-calculated position P IMU , speed V IMU , attitude information A IMU , and the extended Kalman filter is used for GNSSRTK / IMU filter fusion.

[0054] Furthermore, the 21-dimensional system error input into the EKF is shown in the following formula 1:

[0055]

[0056] Where: r , δ v , δ ψ are the three-dimensional position, velocity and attitude error vectors in the Earth-centered Earth-fixed coordinate system, and b g is the gyroscope three-axis zero bias, b a is the accelerometer three-axis zero bias, s g is the three-axis scale error of the gyroscope, s ais the accelerometer scale error. Through EKF recursion, the system error state X of the previous epoch is k-1 Derived the system error state X for the next epoch k The three-dimensional position error in the Earth-centered Earth-fixed coordinate system included therein can be used to accumulate on the basis of the position of the previous epoch to obtain the required pseudo-satellite receiving end position coordinates.

[0057] Furthermore, the recursive formula of the extended Kalman filter EKF is shown in the following equations 2 to 7:

[0058] X k,k-1 =Φ k,k-1 X k-1 (2)

[0059]

[0060] V k =Z k -H k X k,k-1 (4)

[0061]

[0062] X k =X k-1 +K k V k (6)

[0063]

[0064] In the above formulas 2 to 7, X k,k-1 is the one-step prediction state vector, Φ k,k-1 is the state transfer matrix, Σ k,k-1 is the one-step prediction state covariance matrix, B k-1 is the variance matrix of the motion model process error, H k is the observation matrix, V k is the prediction residual, K k is the gain matrix, R k is the observation noise matrix, X k is the estimated state vector, Σ k is the covariance matrix of the estimated state, Z k The observation vector is a loose combination in the present invention, and the input observation vector is the difference between the output state of the GNSS and the estimated state of the IMU.

[0065] Furthermore, the observation vector Z k It is expressed as shown in the following formula 8:

[0066] Z k =[P IMU -PPL v IMU -v PL ] T (8)

[0067] Where: P IMU is the IMU estimated position, P PL Pseudo-satellite receiving end position, V IMU is the IMU estimated velocity, v PL is the displacement speed of the pseudo-satellite receiving end.

[0068] ZUPT is a zero-speed correction, that is, due to the long-term offset of the IMU-calculated position, the speed is limited to zero according to the actual situation. The difference between the three-axis speed output by the IMU and the ZUPT speed, i.e., zero speed, is used as measurement information to limit the divergence of the IMU in the EKF.

[0069] Furthermore, Z after ZUPT constraint k It is expressed as shown in the following formula 9:

[0070] Z k =[P IMU -P PL v IMU ] T (9)

[0071] Where: P IMU is the IMU estimated position, P PL Pseudo-satellite receiving end position, V IMU The velocity is calculated by IMU.

[0072] The system state X obtained by recursion k The first three elements are the pseudo-satellite receiving antenna positions P estimated by the integrated navigation system. k .

[0073] Step 2: Fault detection, including detecting the position fault of the pseudolite itself and introducing the Euclidean distance matrix to detect the pseudolite observation value fault.

[0074] For detecting the position failure of the pseudo-satellite itself, the specific steps are as follows:

[0075] Step 1: Assume that the initial pseudo-satellite receiving antenna position P is calibrated ini .

[0076] Step 2: Solve for P ini and the pseudo-satellite signal receiving antenna position P after EKF k The pseudo-satellite suspected displacement △P is calculated by the pseudo-satellite suspected displacement △P as shown in the following formula 10:

[0077] △P=||P k -P ini|| (10)

[0078] Where: P ini To assume the initial pseudo-satellite receiving antenna position, P k is the position of the pseudo-satellite signal receiving antenna after EKF.

[0079] Step 3: Execute the judgment. If △P≥5cm, it is considered that the pseudo-satellite has its own displacement and the position broadcast in its ephemeris is unreliable. An alarm is sent to the receiver through wireless signal transmission. Otherwise, it does not exist.

[0080] The specific steps for introducing the Euclidean distance matrix to detect pseudolite observation failures are as follows:

[0081] Step 1: Construct a Euclidean distance matrix based on the pseudo-satellite transmitter position and the pseudo-range observations obtained by the monitoring station receiver N PL Indicates the number of pseudo-satellites that the receiver sees in common. The first row and column in the matrix are the pseudo-ranges measured by the receiver. The Euclidean distance matrix is ​​composed of: m is the number of rows, n is the number of columns, the matrix diagonal elements are 0, and the remaining elements are filled with the distances between the known pseudo-satellite positions. The Euclidean distance matrix is ​​composed of: Figure 2 shown.

[0082] for Figure 2 As shown, it is worth noting that the relationship between three pseudo-satellites and the receiver is used as a schematic diagram here. In fact, at least five pseudo-satellites are required to detect the faulty pseudo-satellite through observation redundancy. R1 is the receiver, and PL1~PL3 represent different pseudo-satellites. The elements a on the diagonal of the matrix 11 、a 22 、a 33 、a 44 All are 0, the first row element a of the matrix 1m with the matrix element a in the first column n1 is the observed pseudorange between the pseudosatellite and the receiver and The remaining positions mn is the distance between the two satellites in Indicates the position of the pseudo-satellite represented by the m-1th row, Indicates the position of the pseudo-satellite represented by the n-1th column.

[0083] Step 2: Convert the Euclidean distance matrix is the corresponding Gram matrix The conversion formula is shown in Formula 11 below:

[0084]

[0085] Where: L represents the geometric center matrix, I represents the identity matrix, and 1 represents the column vector of 1. The faulty pseudo-satellite observation value can be checked by using the property that the rank of the Gram matrix is ​​equal to the dimension d of the state space, that is,

[0086] Step 3: Gram matrix Perform singular value decomposition, the specific formula is shown in equation 12 below:

[0087]

[0088] Where: U is the singular vector on the left, V is the singular vector on the right, Σ is a diagonal matrix containing The singular values ​​σ1,σ2,…,σ 1+l .

[0089] Step 4: Execute the judgment. If the test statistic is greater than the detection threshold, it indicates that there are faulty pseudolites among the existing pseudolites, and the faulty pseudolites need to be detected and excluded. For example, it is known that the dimension of the state space in the three-dimensional space is d=3, so the d+1th column of the U matrix can be directly set as u4, and the d+1th column of the V matrix can be set as v4. The elements in u4 and v4 are sorted according to the absolute value, and the element corresponding to the largest absolute value represents the faulty pseudolites. Repeat the above judgment and sorting operations: after eliminating the faulty pseudolites, recalculate the test statistic and judge whether it exceeds the detection threshold. When it exceeds the detection threshold, re-sort the U matrix and the V matrix and determine if it exceeds the detection threshold. Re-sort and re-determine and exclude the faulty satellites until the test statistic of the remaining pseudolites is lower than the detection threshold. It is considered that there are no pseudolites with observation value faults among the remaining pseudolites.

[0090] The judgment formula for executing the judgment is shown in Formula 13:

[0091]

[0092] Where: i For the matrix The singular value of , d is the dimension of the state space, in this embodiment d = 3, is the test statistic, β thre It is a manually set detection threshold.

[0093] Step 3: Executing a decision on the pseudolite based on the detection result, wherein the executing a decision on the pseudolite based on the detection result includes the following situations:

[0094] The first case: For a pseudo-satellite that has a position fault but no observation fault, the observation value is retained, but the alarm information needs to be sent through wireless transmission.

[0095] The second case: For a pseudo-satellite with a faulty position and a faulty observation value, the pseudo-satellite is directly excluded from the solution.

[0096] The third case: For pseudo-satellites with no faults in their own positions but only in their observation values, they are given a downgrade operation and their observation value weights are adjusted to 0.5 during the solution.

[0097] Step 4: Introduce the predicted observation value correction to correct the pseudorange and carrier phase observation values.

[0098] The specific steps of introducing the correction of the predicted observation value are as follows:

[0099] Step 1: Build a dataset based on historical data:

[0100] The feature quantities selected when constructing the data set are: satellite signal elevation angle (el), signal-to-noise ratio (C / N0), satellite azimuth (az), reservoir water level height (H w ) and pseudorange error (△ρ), carrier phase error (△φ). The defined data set is Q = (x, y), where x = [el, az, C / N0, H w ], y = [△ρ, △φ], △ρ is the pseudorange residual obtained by inverting the pseudorange through the determined receiver position, and △φ is the decimal part of the carrier phase residual inverted by the same method. The receiver position coordinates are obtained by position coordinate traction by the total station or by baseline solution by the base station, and then the receiver is used to obtain the GNSS signal to solve the elevation angle, carrier-to-noise ratio, azimuth and observation value residual of the satellite signal. Since the position of the pseudosatellite is fixed, the receiver only needs to obtain the carrier-to-noise ratio and observation value residual of the pseudosatellite signal. The reservoir water level height data is manually input regularly by the observer.

[0101] Step 2: Calibrate the sample using the inverted multipath error value:

[0102] The inverted multipath error values ​​include pseudorange residuals and carrier phase fractional part residuals of the GNSS satellite and pseudosatellite respectively.

[0103] For GNSS satellites, the inverted pseudorange ρ of GNSS satellites res As shown in formula 14:

[0104]

[0105] Where: (x s ,y s ,z s ) represents the GNSS satellite position, t r is the receiver clock error, t sis the satellite clock error, I is the ionospheric delay correction calculated by the Klobuchar model, and T is the tropospheric delay correction calculated by the Hopfield model.

[0106] Furthermore, the calculation formula of the pseudorange residual of GNSS is as shown in the following formula 15:

[0107] △ρ=ρ-ρ res (15)

[0108] Where: ρ represents the actual pseudorange observation value obtained.

[0109] Furthermore, the calculation formula of the fractional residual of the GNSS carrier phase is as shown in the following formula 16:

[0110] △φ={φ}-{ρ res / λ} (16)

[0111] Where: φ is the actual carrier phase observation value, {i} represents the decimal operation, and λ is the wavelength of the GNSS satellite signal.

[0112] For pseudolites, the inverted pseudorange of pseudolites As shown in Formula 17 below:

[0113]

[0114] Where: (x PL ,y PL ,z PL ) is the pseudo-satellite position, t PL is the pseudo-satellite clock error, and T is the tropospheric delay correction calculated by the Hopfield model.

[0115] Similarly, the pseudo-satellite pseudo-range residual calculation formula and the carrier fractional part residual calculation formula are shown in the following equations 18 and 19 respectively:

[0116]

[0117] In the above equations 18 and 19, ρ represents the actual pseudorange observation value, φ represents the actual carrier phase observation value, {i} represents the decimal operation, and λ is the wavelength of the pseudosatellite signal.

[0118] Step 3: Establish an observation correction prediction model based on the integrated bagged regression tree. For the principle of the integrated bagged regression tree algorithm, please refer to Figure 3 , the specific steps are as follows:

[0119] The sub-sample single regression tree model is generated to determine the optimal segmentation variable J and solve the segmentation point G. The minimization formula is shown in Equation 20:

[0120]

[0121] Where: c1 = ave(y i |x i ∈Reg1(J,G)), c2=ave(y i |x i ∈Reg2(J,G)), ave(·) is the average function, Reg is the sub-region divided by the segmentation variable J and the segmentation point G, and then the segmentation variable J is traversed, and the segmentation point G is scanned and the segmentation variable J is fixed, and finally the minimization formula (J,G) is obtained, which is the above formula 20.

[0122] Based on (J, G), the sub-regions are divided and the output values ​​of each sub-region are calculated, as shown in equations 21 to 23 below:

[0123] Reg1(J,G)={x|x (J) ≤G} (21)

[0124] Reg2(J,G)={x|x (J) >G} (22)

[0125]

[0126] In the above formulas 21 to 23: Reg1(J, G) and Reg2(J, G) represent two divided regions respectively, N g is the number of subsets in the region, are the output values ​​corresponding to the two regions. Each time the region is divided, two new nodes are generated. The nodes are composed of the data in the sub-sample and the mapping rules.

[0127] Repeat the above operation for each sub-region until the node sample size is less than the specified threshold. Finally, the input space is divided into Q Reg regions and generate the final regression tree, as shown in Equation 24:

[0128]

[0129] Where: h(x) is the predicted value of the regression tree, q Reg is the region number, IF(·) is the indicative function.

[0130] Furthermore, a training model is generated based on all regression trees:

[0131] The final output is the average of the predicted values ​​of all regression trees i=1,2,…,I established above, expressed as shown in the following formula 25:

[0132]

[0133] Where: h bag(x) is the final prediction value of regression tree i when the regression tree is input x.

[0134] Furthermore, prediction models are established for different GNSS constellations and pseudo-satellites.

[0135] The multipath error prediction rules of different constellations are different. Therefore, the GNSS constellations and pseudo-satellites are divided into independent data sets for training to obtain the corresponding observation correction prediction model h bag,GPS (x), h bag,BDS (x) and h bag,PL (x), etc., represent the prediction models of the observation correction values ​​of GPS satellites, BDS satellites, and pseudo-satellites, etc.

[0136] After the observation correction prediction model is established through the above steps, the real-time state quantity x is input to perform multipath error prediction and correction for each pseudo-satellite and GNSS satellite involved in the solution.

[0137] For pseudo-satellites, the multipath error prediction correction value is as shown in the following equation 26:

[0138]

[0139] That is, after inputting the state quantity x into the pseudo-satellite prediction model, the pseudo-range correction prediction value output by the model can be obtained: and carrier correction value

[0140] For Beidou satellites, the multipath error prediction correction value is shown in the following equation 27:

[0141]

[0142] That is, after inputting the state quantity x into the BDS satellite prediction model, the pseudo-range correction prediction value output by the model can be obtained: and carrier correction value

[0143] Then, the original pseudorange and the original carrier phase are corrected to obtain the corrected pseudorange ρ ad and the corrected carrier phase φ ad They are shown in equations 28 and 29 respectively:

[0144]

[0145]

[0146] Step 5: RTK solution.

[0147] The subsequent RTK solution is performed based on the corrected pseudorange and carrier phase.

[0148] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0149] The above-mentioned embodiments only express several implementation methods of the present invention, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A pseudo-satellite observation quality control method for deformation monitoring, characterized by: The steps include: Establish GNSS / IMU combined positioning based on ZUPT constraints; Fault detection includes detecting the position fault of the pseudolite itself and introducing the Euclidean distance matrix to detect the pseudolite observation value fault, wherein the introduction of the Euclidean distance matrix to detect the pseudolite observation value fault includes the following steps: The Euclidean distance matrix is ​​constructed based on the pseudo-satellite transmitter position and the pseudo-range observations obtained by the monitoring station receiver. R is the representation of the real number field in mathematics, N PL Indicates the number of pseudo-satellites that the receiver sees in common. Transform Euclidean distance matrix is the corresponding Gram matrix The transformed Euclidean distance matrix is the corresponding Gram matrix The conversion formula is as follows: Where: L represents the geometric center matrix, I represents the identity matrix, 1 represents the column vector of 1, For Gram matrix Perform singular value decomposition, Execute judgment. If the test statistic is greater than the detection threshold, it indicates that there are faulty pseudolites among the existing pseudolites. It is necessary to detect and exclude the faulty pseudolites. Repeat the detection and exclusion until the test statistic of the remaining pseudolites is lower than the detection threshold. It is considered that there are no pseudolites with observation value faults among the remaining pseudolites. The judgment formula for executing judgment is as follows: Where: i For the matrix The singular values ​​of , d is the dimension of the state space, is the test statistic, β thre The detection threshold is set artificially; executing decisions on pseudolites based on the detection results; Introduce predicted observation value correction to correct pseudorange and carrier phase observation values; RTK solution.

2. A pseudo-satellite observation quality control method for deformation monitoring as claimed in claim 1, characterized in that: The steps for establishing GNSS / IMU combined positioning based on ZUPT constraints are as follows: Perform RTK positioning on GNSS satellite data to obtain the pseudo-satellite receiving end position P PL ; The pseudo-satellite inertial measurement unit (IMU) is used to collect information and mechanically arrange it to obtain the estimated position P. IMU , speed V IMU , attitude information A IMU Then it is fused through the extended Kalman filter EKF.

3. A pseudo-satellite observation quality control method for deformation monitoring as claimed in claim 2, characterized in that: The detection of pseudolite position failure comprises the following steps: Calibrate the initial pseudo-satellite receiving antenna position P ini ; Solving for P ini and the pseudo-satellite signal receiving antenna position P after EKF k The pseudo-satellite's suspected position △P; Execute the judgment, if △P ≥ 5cm, it is considered that the pseudo-satellite has its own displacement, otherwise it does not exist.

4. The pseudo-satellite observation quality control method for deformation monitoring according to claim 3, characterized in that: The decision-making on pseudolites based on the detection results includes the following situations: For pseudo-satellites that have position faults but no observation faults, the observation value retention operation is allowed, but the alarm information needs to be sent through wireless transmission; For pseudolites with faulty positions and observation values, these pseudolites are directly excluded from the solution. For pseudolites whose positions are not faulty but whose observation values ​​are faulty, they are given a downgraded weighting operation and their observation value weights are adjusted to 0.5 during the solution.

5. The pseudo-satellite observation quality control method for deformation monitoring according to claim 4, characterized in that: The steps of introducing the correction amount of the predicted observation value are as follows: Build data sets based on historical data; The samples are calibrated using the inverted multipath error value; An observation value correction prediction model is established based on integrated bagged regression trees.

6. The pseudo-satellite observation quality control method for deformation monitoring according to claim 5, characterized in that: The inverted multipath error values ​​include pseudorange residuals and carrier phase fractional part residuals of the GNSS satellite and pseudosatellite respectively.

7. The pseudo-satellite observation quality control method for deformation monitoring according to claim 6, characterized in that: The step of establishing the observation value correction prediction model based on the integrated bagged regression tree includes the following steps: Subsample single regression tree model generation; Generate a training model based on all regression trees; Prediction models are established for different GNSS constellations and pseudo-satellites.

Images