GNSS and leveling fusion ground deformation monitoring method and system with dynamic optimization weight

Abstract

The application discloses a GNSS and leveling fusion ground deformation monitoring method and system with dynamic optimization weight, and relates to the technical field of geodesy and deformation monitoring.The method comprises the following steps: analyzing leveling and GNSS rate observation data; setting the elevation correction number and subsidence rate parameter of a monitoring point as unknown parameters, constructing adjustment observation equations coupled with observation time and subsidence rate parameters, forming a design matrix, an observation vector and an initial weight matrix; performing dynamic weight optimization, introducing a global weight factor for adaptive scanning, and on the basis of the optimal global weight ratio, calculating a weight reduction factor by using robust estimation according to the standardized residual error of GNSS observation values, and obtaining the optimal weight matrix; and finally solving the optimal estimation value of unknown parameters based on the optimal weight matrix.The application realizes two-stage dynamic optimization of the weight, objectively optimizes the weight by data driving, effectively suppresses the influence of gross errors, and significantly improves the precision and reliability of multi-source data fusion deformation monitoring.

Description

Technical Field

[0001] This invention relates to the field of geodesy and engineering deformation monitoring technology, and in particular to a GNSS and leveling fusion ground deformation monitoring method and system for dynamically optimizing the weights of observation values ​​in multi-source data fusion adjustment. Background Technology

[0002] Currently, the main monitoring technologies include geometric leveling and continuous operation monitoring via Global Navigation Satellite System (GNSS). Leveling, as a traditional method for transferring elevation benchmarks, features high relative accuracy and reliability at single points, but its operational efficiency is low and its cost is high, making it difficult to achieve large-scale, high spatiotemporal resolution monitoring; it typically only acquires annual or periodic discrete point elevation data. In contrast, GNSS technology can acquire time series of three-dimensional coordinates of monitoring points automatically and around the clock, facilitating the revelation of long-term movement trends. However, its vertical observation accuracy is usually significantly lower than its horizontal accuracy and it is susceptible to errors such as multipath effects and atmospheric delay.

[0003] To overcome the limitations of single technologies, joint adjustment of long-term velocity information acquired from GNSS observations with high-precision elevation difference information acquired from leveling surveys has become an effective way to achieve complementary advantages and improve the overall performance of the monitoring network. However, in this type of multi-source data fusion method, a static weighting strategy based on prior accuracy indicators is commonly used, that is, fixing the weight ratio according to the leveling standard error and the long-term statistical mean square error of GNSS velocity. This method has some prominent problems: First, static weighting cannot reflect the dynamic fluctuations in the quality of observation data with time, space and environmental conditions in actual operations, resulting in a mismatch between the weighting matrix and the actual situation. Secondly, this method lacks robustness to unmodeled systematic errors or gross errors that may exist in GNSS time series, and low-quality observations may excessively affect the overall solution results due to their inappropriate fixed weights.

[0004] The aforementioned defects make it difficult for existing fusion models to achieve theoretically optimal estimates, reducing the accuracy and reliability of settlement parameters, especially the velocity field solution. Summary of the Invention

[0005] The purpose of this invention is to provide a GNSS and leveling fusion ground deformation monitoring method and system that can adaptively adjust observation weights, thereby solving the problem of suboptimal fusion results caused by existing static weighting.

[0006] A dynamically weighted GNSS and leveling fusion ground deformation monitoring method includes the following steps: Step S1, data acquisition and analysis: extract the leveling height difference observation values ​​between monitoring points with leveling observation attributes, the approximate elevation of each monitoring point, the observation time, the length of the measurement section and the benchmark constraint information, and extract the settlement rate observation values ​​and their prior mean square error of monitoring points with GNSS observation attributes. Step S2: Construct a fusion adjustment model by uniformly setting the elevation correction and settlement rate parameters of each monitoring point as unknown parameters, and constructing an adjustment observation equation coupling the observation time and settlement rate parameters, thereby forming the design matrix A, observation vector L, and initial weight matrix. ; Step S3, dynamic weight optimization, which includes global weight ratio optimization and local robustness optimization; The global weighting optimization involves introducing a global weighting factor. Adjust the variance of GNSS observations and scan the global weighting factor within a preset range. The values ​​of are determined by combining the design matrix A and the observation vector L for each global weight factor. The value of the initial weight matrix is ​​updated. And perform adjustment calculations, using effective unit weights for the mean square error. Minimization is used as the criterion to select the corresponding optimal candidate value as the optimal global weight ratio parameter. ; The local robustness optimization is based on the optimal global weighting parameters. Based on this, and using the standardized residuals of the settlement rate observations at the GNSS attribute monitoring points after adjustment, the weighting factor for each GNSS observation is calculated using a robust estimation function, resulting in the final weight matrix. ; Step S4, parameter calculation: Based on the elevation datum constraint, and using the design matrix A, observation vector L, and final weight matrix... Find the optimal estimate of the unknown parameters.

[0007] Preferably, in step S2, the adjustment observation equations are constructed as follows: For monitoring points with leveling observation attributes and , For the observation time The observed elevation difference of the leveling section , For monitoring points and Elevation correction, , For monitoring points and Settlement rate parameters, , For monitoring points and exist Approximate elevation at time, For reference time, For section measurement to The leveling elevation difference observation error; therefore, the leveling elevation difference observation equation is: ; For monitoring points with GNSS observation attributes The equation for direct observation of settling rate is: ; in, Monitoring points with GNSS observation attributes The observed settling rate For monitoring points The error in the observation of the settling rate.

[0008] Preferably, in step S3, the specific process of global weight ratio optimization is as follows: Set global weight factor The scanning range and step size; For each global weight factor The value updates the variance of the settlement rate observations at monitoring points with GNSS observation attributes to [value]. ,in The initial weighting matrix is ​​updated to reflect the prior mean square error of the settlement rate observations. Obtain the updated power array ; Combining the design matrix A, observation vector L, and weight matrix Perform adjustment calculations and calculate the effective unit weight error. The effective unit weight error The unit weighted mean square error is calculated for observations that meet only the valid determination criteria; the valid determination criteria include: the absolute value of its standardized residual is less than a preset threshold, and its weight is not lower than a preset minimum weight threshold.

[0009] Preferably, in step S3, the specific process of local robustness optimization is as follows: Using the optimal global weight ratio parameter The corresponding weight matrix is ​​used to calculate the standardized residuals of the settlement rate observations at each GNSS attribute monitoring point; The weighting factor for each GNSS observation is calculated using a robust estimation function. ,in Belongs to (0,1]; The final weights of the updated GNSS observations are , It is a constant factor.

[0010] Preferably, the robust estimation function is the Huber estimation function or the Tukey double-weight function.

[0011] Preferably, step S4, after solving for the optimal estimate of the unknown parameters, further includes: Generate a quality diagnostic report, which includes the number of valid observations, valid degrees of freedom, valid unit weighted mean square error, and the respective weighted percentages of leveling observations and GNSS observations, as well as the contribution percentage of the residual sum of squares.

[0012] A ground deformation monitoring system applying the aforementioned GNSS and leveling fusion ground deformation monitoring method with dynamically optimized weights includes: The data acquisition and analysis module is used to extract the leveling height difference observation values ​​between monitoring points with leveling observation attributes, the approximate elevation of each monitoring point, the observation time, the length of the measurement section and the benchmark constraint information, and to extract the settlement rate observation values ​​and their prior mean square error of monitoring points with GNSS observation attributes. A fusion adjustment model module is constructed to uniformly set the elevation correction and settlement rate parameters of each monitoring point as unknown parameters, and to construct the adjustment observation equations coupling the observation time and settlement rate parameters, thereby forming the design matrix A, the observation vector L, and the initial weight matrix. ; The dynamic weight optimization module includes global weight ratio optimization and local robustness optimization; The global weighting optimization involves introducing a global weighting factor. Adjust the variance of GNSS observations and scan the global weighting factor within a preset range. The values ​​of are determined by combining the design matrix A and the observation vector L for each global weight factor. The value of the initial weight matrix is ​​updated. And perform adjustment calculations, using effective unit weights for the mean square error. Minimization is used as the criterion to select the corresponding optimal candidate value as the optimal global weight ratio parameter. ; The local robustness optimization is based on the optimal global weighting parameters. Based on this, and using the standardized residuals of the settlement rate observations at the GNSS attribute monitoring points after adjustment, the weighting factor of each GNSS attribute monitoring point observation is calculated using a robust estimation function, resulting in the final weight matrix. ; The parameter calculation module is used to calculate the parameters based on the design matrix A, observation vector L, and final weight matrix, taking into account the elevation datum constraint. Find the optimal estimate of the unknown parameters.

[0013] The advantages and beneficial effects of this invention are as follows: 1. This invention achieves objectivity and adaptability in determining observation weights, overcoming the blindness of traditional experience-based weighting. Existing technologies often rely on fixed prior accuracy or manual experience to set the weight ratio of GNSS and leveling, making it difficult to adapt to changes in the actual observation environment. This invention introduces a global weighting factor. The automated scanning mechanism uses minimizing the error in effective unit weights as an objective criterion, and the selection of weight ratios is entirely driven by data quality. This not only eliminates the random and subjective errors of manual trial calculations, but also ensures that the fusion model is always in a statistically optimal or near-optimal state.

[0014] 2. Enhanced robustness of the fusion model to gross errors and anomalous disturbances. Addressing the issue of sudden gross errors in GNSS observations due to multipath effects, this invention further integrates a local robust estimation algorithm based on residual analysis, building upon global optimization. This mechanism can identify and automatically reduce (or eliminate) the weights of anomalous observations in real time, constructing a dual guarantee of global optimization and local defense, effectively preventing low-quality GNSS data from contaminating high-precision leveling network adjustment results.

[0015] 3. Significantly improved sensitivity and reliability in capturing subtle deformation trends. Through the aforementioned dynamic weight optimization, this invention maximizes the value of high-quality observation data while suppressing data interference with low signal-to-noise ratios. Compared with traditional static weighting methods, under the same observation conditions, the ground subsidence rate and elevation change calculated by this invention have higher accuracy and confidence, and have significant engineering practical value for timely detection and early warning of subtle deformation characteristics in the early stages of geological disasters. Attached Figure Description

[0016] For those skilled in the art, other related figures can be obtained from the above figures without any creative effort.

[0017] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0018] To enable those skilled in the art to better understand the present invention, the technical solution of the present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0019] like Figure 1 As shown, the GNSS and leveling fusion ground deformation monitoring method with dynamically optimized weights of the present invention includes the following steps: Step S1: Data Acquisition and Analysis. Read the leveling observation file, GNSS rate observation file, and monitoring network information. Extract the observation data from each monitoring point in the deformation monitoring network. This includes extracting the leveling elevation difference between monitoring points with leveling observation attributes, the approximate elevation of the monitoring points, observation time, section length, and benchmark constraint information. It also includes extracting the settlement rate observations and their prior mean square errors for monitoring points with GNSS observation attributes (GNSS attribute monitoring points). Specifically, read the leveling observation file, GNSS rate observation file, and monitoring network information configuration file from the storage medium. Analyze and extract the section information of the leveling observations, the leveling elevation difference observations, the approximate elevation of each monitoring point, the observation time, section length and grade, the settlement rate observations and their prior mean square errors for GNSS attribute monitoring points, and the monitoring point number, name, and known constraints.

[0020] Step S2: Construct a fusion adjustment model, setting the elevation correction and settlement rate parameters of the monitoring points as unknown parameters, and constructing the adjustment observation equations that couple the observation time and settlement rate parameters. This includes the leveling difference observation equation and the GNSS settlement rate direct observation equation, forming the design matrix A, the observation vector L, and the initial weight matrix. ; Based on the analyzed data, a unified parametric fusion adjustment model is constructed. The elevation correction and settlement rate parameters of each monitoring point are treated as unknown parameters. Using the leveling difference observation equation (which includes a coupling term between the settlement rate parameter and the observation time) and the GNSS rate direct observation equation, the overall design matrix A and the observation vector L are formed.

[0021] First, at each monitoring point Define two unknown parameters: one is the elevation correction at the reference time. Secondly, the settling rate parameter This settling rate parameter It will be used as an unknown quantity in both the leveling elevation difference observation equation and the GNSS rate direct observation equation.

[0022] Assuming at time Relative time was observed From monitoring points with leveling observation attributes to The leveling elevation difference, the theoretical elevation difference of the leveling section. It can be expressed by the approximate elevation of each point and the unknown parameters as follows: ; in, , It is a monitoring point and monitoring points exist Approximate elevation at time, , For each monitoring point and monitoring points Elevation correction, , For monitoring points and monitoring points The corresponding settling rate parameters. Let... The actual observed elevation difference of the leveling section; introducing the leveling section to Leveling height difference observation error Based on the mathematical relationship between actual observed values ​​and theoretical values ​​(i.e. Substituting these equations into the aforementioned theoretical elevation difference formula and rearranging them, we obtain the leveling elevation difference observation equation and the leveling elevation difference error equation, which are expressed as follows: ; ; For the corresponding monitoring point in the observation vector L and monitoring points The elements are the leveling error constant term. Furthermore, since the elevation correction in the leveling difference observation equation appears in difference form, to ensure the solvability of the normal equation, the fusion adjustment model also introduces elevation datum constraints; these constraints include fixing the elevation correction of one datum monitoring point to zero and applying known constraints to multiple stable monitoring points.

[0023] The direct observation equations and error equations for GNSS settlement rate are as follows: ; ; In the formula GNSS attribute monitoring points The observed settling rate GNSS attribute monitoring points The rate observation error, For the monitoring points in the observation vector L The corresponding element is the GNSS error constant term.

[0024] Based on the analyzed data, the system constructs a unified unknown parameter vector (containing elevation corrections and settlement rate parameters for all monitoring points), a design matrix A, and an observation vector L across the entire network. While constructing the design matrix A and observation vector L, the system also adds constraint terms based on baseline constraint information to eliminate baseline deficits in the elevation corrections. The design matrix A represents the linear mapping relationship between the observation vector L and the unknown parameter vector. Specifically, each row of the design matrix A corresponds to an independent observation equation, and each column corresponds to an unknown parameter to be estimated. The assignment rules for its matrix elements are as follows: For the row vector of the corresponding leveling difference observation equation: at its corresponding starting point elevation correction The column element is assigned a value of -1, corresponding to the endpoint elevation correction. The column element is assigned a value of 1, corresponding to the initial settlement rate parameter. The column element is assigned a value At the corresponding endpoint settlement rate parameter The column element is assigned a value All other elements in the row are 0; For the row vector of the direct observation equation of the settlement rate at the corresponding GNSS attribute monitoring point: in its corresponding settlement rate parameters The column element is assigned a value of 1, and all other elements in the row are 0. Furthermore, the observation vector L is composed of constant terms corresponding to all error equations. A one-dimensional column vector is formed by vertically combining elements in the same row order as the design matrix A.

[0025] To intuitively illustrate the construction logic of the design matrix A, let's take a micro-deformation monitoring network with three monitoring points (denoted as monitoring point 1, monitoring point 2, and monitoring point 3) as an example. Assume that the following four observations are obtained: (1) Leveling observation 1: During the observation time The observed leveling elevation difference from monitoring point 1 to monitoring point 2; (2) Leveling observation 2: During the observation time The observed leveling elevation difference between monitoring point 2 and monitoring point 3; (3) GNSS observation 1: Settlement rate observation value of monitoring point 1 with GNSS attributes; (4) GNSS observation 2: Settlement rate observation of monitoring point 3 with GNSS attributes.

[0026] Based on the above observation conditions, the unknown parameter vector of the entire network It can be represented as: ; in, , The elevation correction values ​​for monitoring points 1, 2, and 3. , These are the settlement rate parameters for monitoring points 1, 2, and 3. The corresponding observation vectors are... (That is, the vector of constant terms in the error equation) is represented as follows: ; Based on the aforementioned rules for assigning values ​​to matrix elements, the design matrix A can now be strictly expressed as: ; The system calculates the initial variances of leveling observations and GNSS observations respectively, and constructs the initial variance-covariance matrix D and its inverse matrix (i.e., the initial weight matrix) accordingly. The specific calculation rules are as follows: For leveling observations: based on the leveling measurement grade and its corresponding random error per kilometer. Combined with the length of the leveling section The initial variance of the leveling observations was determined to be... It corresponds to the monitoring point. to monitoring point The main diagonal weight elements corresponding to the test segment The calculation formula is: ; For GNSS observations: based on the extracted GNSS monitoring points prior standard error and global weight factor The variance of GNSS observations was determined to be... When constructing the initial weight matrix, let the global weight factor... =1, its corresponding main diagonal weight element The calculation formula is: ; Based on the independently calculated initial variances, a diagonal initial variance-covariance matrix D is constructed, whose main diagonal elements are formed by sequentially arranging the initial variances of all observations in the network according to the order of the observation vector L. Then, the inverse of matrix D is obtained to acquire the initial weight matrix. The initial weight matrix It is also a diagonal matrix, and its main diagonal elements are determined by the initial weights of the above observations (i.e., the corresponding...). or Composed of, i.e. .

[0027] Step S3: Dynamic weight optimization, including global weight ratio optimization and local robustness optimization; The global weighting optimization involves introducing a global weighting factor. To adjust the parameters of the variance of GNSS observations, the global weighting factor is scanned within a preset range. The values ​​of are determined by combining the design matrix A and the observation vector L for each global weight factor. The value of the initial weight matrix is ​​updated. Then, adjustment calculations are performed, and within the preset candidate interval, the effective unit weight mean error is selected. smallest ; The local robustness optimization is based on the optimal global weighting parameters. Based on this, and using the standardized residuals of the adjusted GNSS settlement rate observations, the weighting factor of each GNSS attribute monitoring point observation is calculated using a robust estimation function, resulting in the final weight matrix. ; Specifically, in traditional fusion adjustment, the observations of GNSS attribute monitoring points are assigned a fixed prior error based on long-term statistics. This invention introduces a global weighting factor. Express its variance as In typical application scenarios, the number of leveling observation points is usually significantly greater than the number of GNSS observation points. The purpose of this invention is to introduce some GNSS values ​​when adjusting the leveling network to avoid abnormal deformation of the leveling network due to accumulated errors. By adjusting the adjustment factor, the weights of GNSS and leveling are controlled, and the final optimal weight ratio is determined by the effective weight error.

[0028] Specifically, first input the initial design matrix A, the observation vector L, and the initial weight matrix. Within the preset range The inner loop takes values ​​at pre-step intervals to update the GNSS observation weights in the initial weight matrix. , ; According to the updated power array The system performs constrained least-squares adjustment. To eliminate the baseline deficit in elevation corrections, the system constructs elevation constraints based on baseline constraint information and uses constrained adjustment and additional method equations to solve for the optimal estimates of the unknown parameters. The weight matrix... As shown below: ; in The weight representing the leveling observation value, Representative introduction The weights corresponding to the GNSS values. During the adjustment cycle, based on the current values... For the initial weight matrix The updated weight matrix is ​​obtained as follows. Specifically, the diagonal elements corresponding to the leveling observation weights are kept unchanged, while the diagonal elements corresponding to the GNSS observation weights are updated to... All off-diagonal elements are 0.

[0029] Then calculate the effective unit weight error. Obtain the optimal global weight ratio factor The effective unit weight error It is a unit weighted variance estimate calculated based on effective observations with significant weights. The criteria for determining an effective observation include: its standardized residual absolute value is less than a preset threshold, and its weight is not lower than a preset minimum weight threshold. This unit weighted variance estimate, calculated solely from effective observations with significant weights, excludes the distortion of overall accuracy caused by observations deemed to have abnormal weights or residuals during iteration. It reflects the intrinsic consistency of reliable observations better than traditional unit weighted variance. This is achieved through systematic scanning. The values ​​of are taken and the corresponding effective unit weight error is calculated. Within the preset candidate interval, the effective unit weight error is located. Minimum optimal global weight ratio factor This step is a key step in the present invention, in which the system enters an optimization loop: First, the global weight factor is scanned with a fixed step size within a preset range (e.g., 0.5 to 3.0). Value; for each global weight factor Update the GNSS observation variance and reweight it, perform adjustment, and calculate and record the corresponding effective unit weight mean square error. After the loop ends, the system selects the effective unit weight error. The minimum value is used as the optimal global weight ratio parameter. This allowed for the determination of the optimal candidate weight ratio between GNSS and leveling observations globally.

[0030] Subsequently, the effective unit weight error is calculated. Specifically, observations whose standardized residual absolute values ​​are less than a preset threshold and whose weights are not lower than the minimum weight threshold are selected to form a valid observation set. The calculation model is as follows: ; In the formula, In order to effectively observe the corresponding residual vector, Here is the corresponding weight matrix. For the number of valid observations, The total number of unknown parameters. To determine the effective degrees of freedom of the adjustment system, select the values ​​that make the system more efficient. To reach the minimum The value serves as the optimal global weight ratio parameter. .

[0031] In the local robust estimation stage, the system is based on the optimal global weight ratio factor. Perform an adjustment to obtain the residuals of the settlement rate observations at each GNSS attribute monitoring point. and its corresponding prior mean square error Then, the standardized residuals of the settlement rate observations at each GNSS attribute monitoring point are calculated. , ; For each monitoring point with GNSS observation attributes after adjustment, The standardized residuals of the settlement rate observations This represents the original residual of the settlement rate observations at the GNSS attribute monitoring point, obtained through adjustment calculations.

[0032] Further application of robust functions (such as Huber or Tukey functions) based on the standardized residuals Calculate the settlement rate observations for each GNSS attribute monitoring point. weighting factor The weight of the settlement rate observation value of the GNSS attribute monitoring point is then adjusted to... , This is a constant factor used to maintain consistent final weight scales or meet weight matrix normalization requirements; it is typically set to 1. This process is equivalent to adaptively reducing the weights of local outlier observations based on global optimality. Then, the final weight matrix is ​​obtained: ; Compared to the traditional classical least squares adjustment method, which is sensitive to gross errors and weight configuration and easily distorted by individual residuals or extremely low weights, this method optimizes the objective function to exclude observations with abnormal residuals or weights from the overall accuracy, obtains the optimized effective unit weight mean square error, and finally selects the global weight factor that minimizes the objective function value within a preset candidate interval. The value is used as the optimal global weight ratio parameter. Secondly, local robust optimization is performed. Based on the determined optimal global weight ratio, the robust estimation principle is used to dynamically calculate the weight reduction factor of each GNSS observation according to the standardized residuals of each GNSS rate observation after adjustment, and automatically reduce the weight of observations with very large residuals, thereby suppressing the influence of gross errors.

[0033] Step S4: Parameter calculation. Based on the elevation datum constraint, and using the design matrix A, observation vector L, and final weight matrix... The optimal estimates of the unknown parameters are obtained. This can be achieved by solving the equations using least squares adjustment or robust estimation methods to find the optimal estimates of the elevation corrections and settlement rate parameters for all monitoring points. In existing static weighting methods, the weight matrix P is fixed and based on prior experience. However, the final weight matrix in this invention... It is dynamically generated through a two-stage process of global weight ratio optimization and local robustness optimization. It is a data-driven, quality-adaptive optimal weight matrix.

[0034] Meanwhile, after solving for the optimal estimate of the unknown parameters, step S4 also includes calculating the accuracy information of the optimal estimate of the parameters and generating a key quality diagnostic report, including the number of effective observations, effective degrees of freedom, effective unit weight error, and the respective weighted sum of leveling observations and GNSS observations and the contribution of the residual sum of squares, in order to quantitatively evaluate the actual contribution and data quality of various types of data in the fusion, thereby comprehensively evaluating the quality and reliability of data fusion.

[0035] Furthermore, the calculated elevation corrections, settlement rate parameters, corresponding accuracy information, and quality diagnostic reports can be output as readable text files and structured data files in a predetermined format for storage, visualization, or further analysis.

[0036] This invention proposes a joint adjustment method based on dynamically optimized weights, solving the problem in existing technologies where fixed prior weight ratios cannot adapt to actual fluctuations in data quality, leading to suboptimal fusion results and poor robustness. Compared to traditional static weighted fusion methods, this model introduces an adjustable global weight factor. And residual-based locally robust estimation, global weighting factor The system systematically explores and determines the optimal weighting ratio of GNSS observation data relative to leveling observation data in this adjustment; robust estimation identifies and suppresses potential random gross errors or local anomalies in the GNSS data. In other words, it achieves two-level dynamic optimization of the weights. This enables the system to automatically find the optimal weighting ratio between the two types of observations and suppress the influence of gross errors.

[0037] Furthermore, this invention also discloses a ground deformation monitoring system that applies the aforementioned GNSS and leveling fusion ground deformation monitoring method with dynamically optimized weights, comprising: The data acquisition and parsing module is used to read and transform observation data; The model building module is used to create the design matrix and observation vectors; The dynamic weight optimization module is used to execute global weight factors. Scanning and robust estimation weighting; The parameter calculation module is used to output the optimal estimate of deformation parameters. The system is typically implemented as a physical server or cloud server and is responsible for running the core algorithm of this invention. This system can receive observation data results from the acquisition end via a data transmission network and store the processed deformation results and diagnostic reports in an associated database. Finally, users can access and visualize the monitoring results through a results display terminal, such as a personal computer or professional workstation. The dynamic weighting fusion processing system and the database can be deployed on the same server or in a distributed architecture; this invention does not limit this.

[0038] In other words, this invention encapsulates the complete processing flow (including data reading, model building, dynamic weighting, and solution diagnosis) into an automated software system that can be executed via command line, script, or configured tasks. Users only need to prepare a standard-format input file and start the service to automatically obtain high-precision deformation parameters and quality assessment reports.

[0039] When performing fusion calculations based on the aforementioned dynamic weighting model, the optimization of GNSS observation weights by the system can be implemented through different strategies, including but not limited to the following two methods: First type: Fully automatic optimization mode 1) The system uses a preset global weight factor. The scanning range and step size are determined automatically to optimize the global weight ratio. This process is entirely algorithm-driven, using the minimization of the error per unit weight as the sole objective criterion, requiring no manual intervention.

[0040] 2) Determining the optimal global weighting factor Then, the system automatically activates the built-in robust estimation module to perform local weight adjustments on the settlement rate observations of all GNSS attribute monitoring points.

[0041] 3) The system uses the optimized weight matrix to complete all subsequent calculations and outputs.

[0042] The second type is the interactive semi-automatic optimization mode. 1) The system completes the global weighting factor. After scanning, it not only provides the optimal global weight ratio factor. At the same time, the effective unit weight error With global weight factor The changing curve is displayed to the user. An acceptable value can be selected near the optimal value recommended by the algorithm. Value, denoted as Even when the algorithm has already recommended the optimal value, users are allowed to manually fine-tune it. The main advantage is that it broadens the applicable scenarios of the solution and enhances its practicality.

[0043] 2) For robust estimation, the type of robust function can be manually adjusted. Adjusting the type of robust function includes choosing Huber or Tukey and the threshold parameter (c). Huber is more conservative, retaining some of the influence of large residuals; Tukey is more aggressive, completely eliminating large residuals. Choose Tukey when there is suspicion of destructive gross errors.

[0044] 3) The system, based on user confirmation... With robust parameter configuration, the final weighting and calculation are completed. This mode provides users with algorithm-recommended decision intervention points, suitable for research scenarios or scenarios requiring fine-grained control over specific projects.

[0045] The present invention has been described above by way of example. It should be noted that any simple modifications, alterations or other equivalent substitutions that can be made by those skilled in the art without creative effort without departing from the core of the present invention fall within the protection scope of the present invention.

Claims

1. A GNSS and leveling fusion ground deformation monitoring method with dynamically optimized weights, characterized in that, Includes the following steps, Step S1, data acquisition and analysis: extract the leveling height difference observation values ​​between monitoring points with leveling observation attributes, the approximate elevation of each monitoring point, the observation time, the length of the measurement section and the benchmark constraint information, and extract the settlement rate observation values ​​and their prior mean square error of monitoring points with GNSS observation attributes. Step S2: Construct a fusion adjustment model by uniformly setting the elevation correction and settlement rate parameters of each monitoring point as unknown parameters, and constructing an adjustment observation equation coupling the observation time and settlement rate parameters, thereby forming the design matrix A, observation vector L, and initial weight matrix. ; Step S3, dynamic weight optimization, which includes global weight ratio optimization and local robustness optimization; The global weighting optimization involves introducing a global weighting factor. Adjust the variance of GNSS observations and scan the global weighting factor within a preset range. The values ​​of are determined by combining the design matrix A and the observation vector L for each global weight factor. The value of the initial weight matrix is ​​updated. And perform adjustment calculations, using effective unit weights for the mean square error. Minimization is used as the criterion to select the corresponding optimal candidate value as the optimal global weight ratio parameter. ; The local robustness optimization is based on the optimal global weighting parameters. Based on this, and using the standardized residuals of the settlement rate observations at the GNSS attribute monitoring points after adjustment, the weighting factor for each GNSS observation is calculated using a robust estimation function, resulting in the final weight matrix. ; Step S4, parameter calculation: Based on the elevation datum constraint, and using the design matrix A, observation vector L, and final weight matrix... Find the optimal estimate of the unknown parameters.

2. The GNSS and leveling fusion ground deformation monitoring method with dynamically optimized weights according to claim 1, characterized in that, In step S2, the adjustment observation equations are constructed as follows: For monitoring points with leveling observation attributes and , For the observation time The observed elevation difference of the leveling section , For monitoring points and Elevation correction, , For monitoring points and Settlement rate parameters, , For monitoring points and exist Approximate elevation at time, For reference time, For section measurement to The leveling elevation difference observation error; therefore, the leveling elevation difference observation equation is: ； For monitoring points with GNSS observation attributes The equation for direct observation of settling rate is: ； in, Monitoring points with GNSS observation attributes The observed settling rate For monitoring points The error in the observation of the settling rate.

3. The GNSS and leveling fusion ground deformation monitoring method with dynamically optimized weights according to claim 2, characterized in that, In step S3, the specific process of global weight ratio optimization is as follows: Set global weight factor The scanning range and step size; For each global weight factor The value updates the variance of the settlement rate observations at monitoring points with GNSS observation attributes to [value]. ,in The initial weighting matrix is ​​updated to reflect the prior mean square error of the settlement rate observations. Obtain the updated power array ; Combining the design matrix A, observation vector L, and weight matrix Perform adjustment calculations and calculate the effective unit weight error. ; The effective unit weight error The unit weighted mean square error is calculated for observations that meet only the valid determination criteria; the valid determination criteria include: the absolute value of its standardized residual is less than a preset threshold, and its weight is not lower than a preset minimum weight threshold.

4. The GNSS and leveling fusion ground deformation monitoring method with dynamically optimized weights according to claim 3, characterized in that, In step S3, the specific process of local robustness optimization is as follows: Using the optimal global weight ratio parameter The corresponding weight matrix is ​​used to calculate the standardized residuals of the settlement rate observations at each GNSS attribute monitoring point; The weighting factor for each GNSS observation is calculated using a robust estimation function. ,in Belongs to (0,1]; The final weights of the updated GNSS observations are , It is a constant factor.

5. The GNSS and leveling fusion ground deformation monitoring method with dynamically optimized weights according to claim 4, characterized in that, The robust estimation function is either the Huber estimation function or the Tukey double-weight function.

6. The GNSS and leveling fusion ground deformation monitoring method with dynamically optimized weights according to claim 1, characterized in that, Step S4, after solving for the optimal estimate of the unknown parameters, also includes: Generate a quality diagnostic report, which includes the number of valid observations, valid degrees of freedom, valid unit weighted mean square error, and the respective weighted percentages of leveling observations and GNSS observations, as well as the contribution percentage of the residual sum of squares.

7. A ground deformation monitoring system applying the GNSS and leveling fusion ground deformation monitoring method according to any one of claims 1 to 6, characterized in that, include: The data acquisition and analysis module is used to extract the leveling height difference observation values ​​between monitoring points with leveling observation attributes, the approximate elevation of each monitoring point, the observation time, the length of the measurement section and the benchmark constraint information, and to extract the settlement rate observation values ​​and their prior mean square error of monitoring points with GNSS observation attributes. A fusion adjustment model module is constructed to uniformly set the elevation correction and settlement rate parameters of each monitoring point as unknown parameters, and to construct the adjustment observation equations coupling the observation time and settlement rate parameters, thereby forming the design matrix A, the observation vector L, and the initial weight matrix. ; The dynamic weight optimization module includes global weight ratio optimization and local robustness optimization; The global weighting optimization involves introducing a global weighting factor. Adjust the variance of GNSS observations and scan the global weighting factor within a preset range. The values ​​of are determined by combining the design matrix A and the observation vector L for each global weight factor. The value of the initial weight matrix is ​​updated. And perform adjustment calculations, using effective unit weights for the mean square error. Minimization is used as the criterion to select the corresponding optimal candidate value as the optimal global weight ratio parameter. ; The local robustness optimization is based on the optimal global weighting parameters. Based on this, and using the standardized residuals of the settlement rate observations at the GNSS attribute monitoring points after adjustment, the weighting factor of each GNSS attribute monitoring point observation is calculated using a robust estimation function, resulting in the final weight matrix. ; The parameter calculation module is used to calculate the parameters based on the design matrix A, observation vector L, and final weight matrix, taking into account the elevation datum constraint. Find the optimal estimate of the unknown parameters.

Images