Dam deformation monitoring method for water conservancy construction process

Abstract

The present application relates to the technical field of precision engineering surveying, and discloses a dam body deformation monitoring method for water conservancy construction process, comprising: arranging a main station coaxial with a total station and a satellite receiver, synchronously obtaining an optical-electrical observation vector and a satellite baseline vector, and constructing a mixed observation equation group with an atmospheric refraction coefficient as an unknown parameter; establishing a distance scalar constraint condition by using a satellite baseline vector length, and synchronously obtaining a refraction coefficient value by adjustment calculation; and correcting zenith distance and calculating an absolute deformation value according to the refraction coefficient value, wherein the present application uses the physical characteristics of microwave ranging which is not affected by light refraction, internally inverts the refraction parameter under the condition of no external meteorological sensor, eliminates the interference of the non-uniform atmospheric field on the trigonometric leveling, and improves the physical authenticity of the monitoring data in a complex environment.

Description

Technical Field

[0001] This invention relates to a method for monitoring dam deformation during water conservancy construction, belonging to the field of precision engineering measurement technology. Background Technology

[0002] Currently, during the construction period of ultra-high arch dams or gravity dams, it is crucial to establish a high spatiotemporal resolution deformation monitoring system to ensure project safety. Conventional monitoring technology systems employ automatic total stations to construct photoelectric edge and corner observation networks or deploy external deformation control networks in conjunction with global navigation satellite systems. They utilize polar coordinate methods or intersection methods to obtain the three-dimensional coordinate changes of key parts of the dam body. The calculation logic of such high-precision measurement methods relies rigorously on the mathematical assumption of the absolute stability of the coordinates of the measuring stations or backsight reference points, and the control points on both sides of the dam abutment are identified as fixed geometric origins.

[0003] In deep canyon terrain, water conservancy construction sites are inevitably subject to millimeter-level or even centimeter-level intrinsic rheological changes and relaxations due to the combined effects of rock excavation unloading, high-pressure grouting, and the periodic water storage load of the reservoir. At this time, the control benchmark network established on both banks is actually in a nonlinear dynamic drift state. This causes the deformation data of the dam body calculated based on the fixed benchmark assumption to be mixed with the systematic error of the benchmark point's own displacement, failing to truly reflect the structural response of the dam body relative to the geological environment. Existing technologies attempt to introduce satellite positioning technology to correct benchmark drift, but in the environment of high mountains and deep canyons, satellite signals are easily interfered with by multipath effects, and the accuracy of the elevation component is far lower than that of the plane component, failing to meet the needs of millimeter-level settlement monitoring. Simultaneously, the complex and non-uniform micro-meteorological environment within the canyon causes drastic fluctuations in the atmospheric refractive index along the photoelectric measurement path, making conventional meteorological data unsuitable. The correction model is difficult to accurately describe the bending law of the light path, resulting in cumulative deviations in the trigonometric leveling. In order to correct the refraction interference in complex environments, existing technologies have attempted to introduce auxiliary observation methods, but they are often limited to using external asynchronous data sources and are difficult to accurately describe the non-uniform atmospheric field on long-distance light paths. For example, Chinese invention patent with publication number CN108765432A discloses a corner network deformation monitoring method based on single meteorological station data correction. This method relies on point data provided by a single or sparse meteorological station for correction, which has defects in deep canyon terrain: the local non-uniform temperature and humidity gradient on the long observation path makes the data of a single station unable to represent the path average true value of the atmospheric refraction coefficient, resulting in the deviation between the calculated correction and the actual value, and ultimately leaving systematic errors in the trigonometric leveling results.

[0004] Therefore, how to construct a deformation calculation system that does not rely on physically fixed points, inherently eliminates the effects of non-uniform atmospheric refraction, and has the ability to resist multipath interference has become the technical problem to be solved by this invention. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: A method for monitoring dam deformation during water conservancy construction, comprising the following steps:

[0006] Multiple benchmark points were set up on the rock slopes on both sides of the dam in the water conservancy construction area, and multiple monitoring points were set up on the dam surface. A total station and a GNSS receiver were coaxially set up at the points whose field of view covered the monitoring points to form the main station, and the remaining benchmark points were defined as auxiliary stations. Within the monitoring epoch, the total station of each main station is driven to traverse the auxiliary stations and monitoring points within the visible range to obtain photoelectric observation vectors including slant range, horizontal angle and zenith distance, and simultaneously obtain the GNSS baseline vectors between each main station. A hybrid observation equation set is constructed, which includes the photoelectric observation vector error equation and the GNSS baseline vector constraint equation. In the photoelectric observation vector error equation, the atmospheric refractive index of the current optical path is set as an unknown parameter that varies with the station position. Calculate the modulus of the GNSS baseline vector and establish a distance constraint equation based on the modulus. Use the distance constraint equation to ensure that the solution value of the slant range obtained by the total station after correction by the atmospheric refraction coefficient converges to the modulus. The mixed observation equations were adjusted using a rank-deficient free network, and the relative geometric coordinates and atmospheric refractive index of the main station, auxiliary station and monitoring points were calculated simultaneously using the least squares criterion. The calculated atmospheric refractive index is used to correct the optical path curvature of the zenith distance obtained by the total station, and the absolute deformation value of each monitoring point relative to the reference point is calculated based on the corrected zenith distance.

[0007] Preferably, before constructing the hybrid observation equation set, the method further includes: performing independent free network adjustment using the photoelectric observation vectors of the total station to obtain the initial relative geometric structure of the observation network consisting of the main station, auxiliary station, and monitoring points; calculating the scale ratio between the distance between each main station in the initial relative geometric structure and the modulus of the corresponding GNSS baseline vector; constructing a statistical distribution model based on the scale ratio between all main stations, identifying anomalous GNSS baseline vectors whose scale ratio deviates from the confidence interval of the statistical distribution model, and removing the constraint equations corresponding to the anomalous GNSS baseline vectors when constructing the hybrid observation equation set.

[0008] Preferably, after performing rank-deficient free network adjustment on the mixed observation equation set, the method further includes: based on the statistical characteristics of the relative displacement of each benchmark point between the current monitoring epoch and the initial monitoring epoch, using an iterative weighting method to select a quasi-stable benchmark group from the benchmark points; calculating the displacement vector of each benchmark point in the quasi-stable benchmark group relative to the centroid of the quasi-stable benchmark group; and using the centroid of the quasi-stable benchmark group as a constraint, transforming the relative geometric coordinates to the engineering coordinate system using the benchmark transformation method.

[0009] Preferably, the step of selecting a pseudo-stable benchmark group from the benchmark points includes: constructing a benchmark topology connection network based on the spatial location of the benchmark points; calculating the linear strain value of each topological edge in the benchmark topology connection network according to the result of the rank-deficient free network adjustment; calculating the sum of the linear strain energies of all topological edges associated with each benchmark point, and using this sum as a quantitative index for evaluating the local geological structural stability of each benchmark point; including benchmark points with quantitative indices lower than a preset threshold into the pseudo-stable benchmark group, and setting the pseudo-stable benchmark group to have a higher weight than other benchmark points in coordinate transformation calculations.

[0010] Preferably, the steps for constructing a hybrid observation equation set including the photoelectric observation vector error equation and the GNSS baseline vector constraint equation include: calculating the posterior variance component of the GNSS baseline vector; dynamically allocating the weight matrix of the GNSS baseline vector constraint equation in the adjustment solution based on the posterior variance component and the nominal accuracy of the total station ranging; and reducing the weight value of the constraint equation corresponding to the GNSS baseline vector in the hybrid observation equation set when the posterior variance component is greater than a preset accuracy threshold.

[0011] Preferably, the step of correcting the zenith distance obtained by the total station for optical path curvature using the calculated atmospheric refractive index follows the following correction logic: Substitute the calculated atmospheric refractive index into the trigonometric leveling model to calculate the elevation correction caused by optical path curvature; wherein the calculation of the elevation correction follows the following mathematical relationship: ,in, This is the elevation correction value. The calculated atmospheric refractive index, The slope distance is measured by a total station. The mean radius of curvature of the Earth is given; the elevation correction is superimposed on the original elevation difference observation to obtain the precise elevation difference after eliminating the optical path bending error.

[0012] Preferably, before performing adjustment and solution of the mixed observation equation set, the method further includes: identifying closed geometric loops formed by the observation paths of the total station in the observation network; calculating the elevation difference closure error of each closed geometric loop based on the original observation values; establishing a set of conditional equations with the atmospheric refractive index as the unknown parameter and the elevation difference closure error as the observation value; solving the set of conditional equations to obtain the initial value of the regional equivalent atmospheric refractive index of the current survey area, and using the initial value as the iterative initial value of the atmospheric refractive index in the mixed observation equation set.

[0013] Preferably, in the step of constructing a hybrid observation equation set including the photoelectric observation vector error equation and the GNSS baseline vector constraint equation, for each survey line that simultaneously possesses total station observation data and GNSS baseline vector, a light wave ranging error equation and a microwave ranging constraint equation are established respectively; a proportional error term proportional to the distance is introduced as an unknown parameter into the light wave ranging error equation; the proportional error term is numerically locked using the microwave ranging constraint equation to separate the photoelectric ranging system error caused by atmospheric refraction.

[0014] Preferably, the method further includes: monitoring the time series change rate of the atmospheric refractive index in real time during the monitoring process; when the time series change rate exceeds a preset gradient threshold, generating an instruction to increase the number of observations; and driving the total station to perform encrypted observations on points within the visible range according to the instruction; and using the encrypted observation data to update the hybrid observation equation set and re-solve it.

[0015] Preferably, the steps of coaxially deploying a total station and a GNSS receiver to form a main station by selecting a point from the reference point that covers the field of view of the monitoring point include: calibrating the principal optical axes of the total station and the GNSS receiver to keep them coincident in the vertical direction; measuring the spatial deviation between the optical center of the total station and the phase center of the GNSS receiver; storing the spatial deviation as a fixed parameter in the solution model; and using the spatial deviation to center the GNSS baseline vector when constructing the hybrid observation equation set.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. In the deformation of dam body during water conservancy construction, a hybrid observation rank-deficient free network adjustment model based on baseline vector constraints of the Global Navigation Satellite System is constructed to avoid dependence on physically fixed starting points. The satellite observation spatial baseline vector is introduced as a rigid geometric constraint into the photoelectric measurement equation to lock the scale and orientation skeleton of the measurement network. Combined with the iterative weighted quasi-stable benchmark screening logic, the overall drift component of the benchmark point caused by geological unloading and the structural deformation component of the monitoring point are separated from the statistical distribution. A calculation method for obtaining high-confidence relative deformation data in deep canyon environments lacking absolute fixed points is established to avoid systematic coordinate deviations introduced by the displacement of the starting benchmark itself.

[0017] 2. Establish a closed-loop compensation mechanism based on the refractive index inversion of microwave slant range scalar optical path. Utilizing the physical characteristic that the propagation path of microwave signals is not affected by atmospheric refraction, the satellite baseline modulus is introduced as a true distance constraint into the photoelectric observation equations. By minimizing the heterogeneous ranging residual, the real-time equivalent refractive parameters of the current optical path are solved in reverse. The parameters are used to correct the optical path curvature of the total station zenith distance observation, transforming the measurement system itself into an atmospheric environment sensing unit. Without relying on external meteorological sensors, the problem of trigonometric elevation transmission distortion caused by non-uniform temperature and humidity fields is solved, realizing the transformation of elevation measurement accuracy from empirical model correction to physical parameter inversion.

[0018] 3. Establish a scale consistency verification barrier for rigid microwave signals based on photoelectric angle mesh. Construct a relative geometric benchmark using a high internal coincidence accuracy angle observation network of a total station. Perform scale factor statistical verification on the satellite-derived baseline vector. Identify and block abnormal microwave baselines contaminated by canyon multipath effects from participating in the adjustment based on the optical mesh geometric closure condition. Ensure that the final coordinate calculation system has both the absolute scale advantage of microwave measurement and the local mesh construction accuracy of photoelectric measurement. Based on the bidirectional verification method of heterogeneous data geometric constraints, maintain the geometric stability of the monitoring system in complex terrain and electromagnetic environment, and prevent external signal noise from inducing distortion of the high-precision measurement network. Attached Figure Description

[0019] Figure 1 This is a flowchart of the deformation monitoring method based on GNSS constraint and refractive inversion of the present invention; Figure 2 This is a curve comparing the convergence of the closed loop elevation difference closure error before and after the present invention. Figure 3 This is a hardware layout topology diagram of the dam monitoring system with heterogeneous monitoring stations in this invention. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0021] This invention discloses a method for monitoring dam deformation during water conservancy construction. Multiple reference points are established on the rock slopes on both sides of the dam in the construction area, and multiple monitoring points are established on the dam surface. Points covering the monitoring points are selected from the reference points. A total station and a GNSS receiver are coaxially deployed to form the main station, and the remaining reference points are defined as auxiliary stations. During system initialization, the principal optical axes of the total station and the GNSS receiver are calibrated to be aligned vertically. The spatial deviation between the optical center of the total station and the phase center of the GNSS receiver is measured and stored as a fixed parameter in the solution model. Within the monitoring epoch, the total station at each main station traverses the auxiliary stations and monitoring points within its visible range, acquiring data including slant distance. Horizontal angle and zenith distance The photoelectric observation vector, and the GNSS baseline vector acquired by the GNSS receivers between the main stations. The system extracts the magnitude of the GNSS baseline vector. As a distance scalar constraint, to ensure the reliability of the observation data, before constructing the hybrid observation equation system, independent free network adjustment was performed using the photoelectric observation vectors of the total station to obtain the initial relative geometric structure of the observation network consisting of the main station, auxiliary stations, and monitoring points. The distances between the main stations in the initial relative geometric structure were then calculated. Modulus of the corresponding GNSS baseline vector scale ratio between A statistical distribution model is constructed based on the scale ratios among all main stations. Abnormal GNSS baseline vectors whose scale ratios deviate from the confidence interval of the statistical distribution model are identified. If the scale ratio of a certain baseline deviates from the mean by more than three times the mean error, it is determined that the GNSS baseline vector is affected by the multipath effect. The constraint equations corresponding to the abnormal GNSS baseline vector are then removed when constructing the hybrid observation equation set.

[0022] To address the impact of non-uniform atmospheric fields on elevation measurements, this method constructs a hybrid set of observation equations, including photoelectric observation vector error equations and GNSS baseline vector constraint equations. For each survey line that simultaneously possesses total station observation data and GNSS baseline vectors, optical ranging error equations and microwave ranging constraint equations are established separately. In the photoelectric observation vector error equations, the atmospheric refractive index of the current optical path is included. The unknown parameters are set as varying with the station location. Distance constraints are established using microwave ranging constraint equations to ensure that the calculated slope range obtained by the total station, after correction for atmospheric refraction, converges to the modulus of the GNSS baseline vector. A rank-deficient free network adjustment is performed on the mixed observation equations, and the relative geometric coordinates and atmospheric refraction of the main station, auxiliary station, and monitoring points are simultaneously calculated using the least squares criterion. In the adjustment process, the posterior variance component of the GNSS baseline vector is calculated. Based on the posterior variance component and the nominal accuracy of the total station ranging, the weight matrix of the GNSS baseline vector constraint equations in the adjustment solution is dynamically allocated. When the posterior variance component exceeds a preset accuracy threshold, the weight of the constraint equation corresponding to that GNSS baseline vector in the mixed observation equation set is reduced. This performs optical path curvature correction on the zenith distance acquired by the total station. The principle is based on a trigonometric leveling model, which treats optical path curvature as an atmospheric refraction phenomenon, causing elevation correction. and observation distance It is proportional to the square of the value and is related to the atmospheric refractive index. Earth's mean radius of curvature Related, elevation correction numbers The calculation follows the mathematical relationship below: ,in, The elevation correction caused by the bending of the optical path is expressed in the dimension of length (m) and its range is all real numbers. The atmospheric refractive index, obtained by solving, is a dimensionless parameter, and its value range is typically between [value range missing]. to between; The slope distance measured by the total station is in the dimension of length (m) and its range is real numbers greater than zero. Let be the Earth's mean radius of curvature, with dimensions of length (m), and the constant value be approximately . kilometers, elevation correction number By superimposing the original elevation difference observations, a precise elevation difference is obtained after eliminating optical path bending errors.

[0023] Using the calculated atmospheric refractive index Zenith distance obtained by total station The optical path curvature correction follows the mathematical relationship described below: Substituting the calculated atmospheric refractive index into the trigonometric leveling model, the elevation correction caused by the optical path curvature is calculated. The calculation formula is: ,in, The slope distance is measured by a total station. To obtain a precise elevation difference after eliminating optical path bending errors, the elevation correction is superimposed on the original elevation difference observation value, using the Earth's mean radius of curvature as the reference point. For local areas lacking GNSS constraints, closed geometric loops formed by the observation paths of total stations are identified in the observation network. The elevation difference closure error of each closed geometric loop based on the original observation value is calculated. A set of conditional equations is established with the atmospheric refractive index as the unknown parameter and the elevation difference closure error as the observed value. The conditional equations are solved to obtain the initial value of the regional equivalent atmospheric refractive index of the current survey area, and this initial value is used as the initial iterative value of the atmospheric refractive index in the hybrid observation equations. After obtaining the relative geometric coordinates, a benchmark stability analysis is performed. Based on the statistical characteristics of the relative displacement of each benchmark point between the current monitoring epoch and the initial monitoring epoch, an iterative weighted method is used to select a quasi-stable benchmark group from the benchmark points. This selection process includes: constructing a benchmark topology connection network based on the spatial location of the benchmark points; and calculating the linear strain value of each topological edge in the benchmark topology connection network based on the results of the rank-deficient free network adjustment. ; Calculate the sum of the linear strain energies of all topological edges associated with each reference point. The sum is used as a quantitative index to evaluate the local geological structural stability of each benchmark point. Benchmark points with quantitative indices below a preset threshold are included in the pseudo-stable benchmark group. Finally, using the centroid of the pseudo-stable benchmark group as a constraint, the relative geometric coordinates are transformed to the engineering coordinate system using the benchmark transformation method. The absolute deformation value of the monitoring point relative to the benchmark point is calculated. During the monitoring process, the time series change rate of the atmospheric refractive index is monitored in real time. When the time series change rate exceeds the preset gradient threshold, an instruction to increase the number of observations is generated. Based on this instruction, the total station is driven to perform intensified observations on points within the visible range. The intensified observation data is used to update the hybrid observation equation set and recalculate it.

[0024] This scheme is used to calculate the topological strain energy for evaluating the local geological structural stability of benchmark points. The principle is based on the theory of relative deformation and energy density in elasticity, treating the observation network as an elastic body connected by topological edges, with linear strain values... The calculation uses the relative change in geometric distance between two points over two monitoring epochs, for connecting reference points. and Topological edges Linear strain The calculation formula is as follows: This represents the change in distance between two reference points between the current monitoring epoch and the initial monitoring epoch. The distance between the two points is calculated using the currently solved relative geometric coordinates. Subtract the initial epoch distance get Reference point for initial monitoring epoch. and The distance between them, linear strain energy The calculation is based on the principle that the deformation energy stored in a unit volume of an elastic body is proportional to the strain, and the linear strain energy... Defined as associated with a reference point The weighted sum of the squares of the linear strain values ​​of all topological edges: , indicating all points relative to the reference point The set of directly connected reference points is used for calculation. As a quantitative benchmark Indicators of local geological stability will Benchmark points below a preset threshold are included in the quasi-stable benchmark group. This threshold is set based on statistical analysis results of empirical values ​​of critical strain for microcrack propagation in rock masses during geotechnical engineering, used to distinguish between elastic deformation and plastic slip in rock masses, ensuring the selected benchmark group possesses the highest geometric stability. To ensure the numerical stability of the adjustment system in the initial stage and prevent weight divergence due to random error fluctuations, this method presets an initial weight allocation and weight damping procedure based on prior accuracy before performing adaptive weighting iterations, according to the factory-specified accuracy of the total station, such as... Typical solution accuracy of GNSS receivers in static mode, such as horizontal. The ratio of the prior standard error of the two observations at the current range length is calculated, and the reciprocal of the square of this ratio is used as the initial weight ratio of the heterogeneous observations, thus providing a prior constraint consistent with physical facts for the first iteration. A weight damping factor is introduced during the iteration process. The value range is typically from 0.7 to 0.9, using the formula... Update the weight matrix, where The theoretical weights calculated in the current step. This represents the weight of the previous iteration.

[0025] Example 1: In a water conservancy construction area with high mountain canyons and deep troughs, the rock masses on both sides of the dam abutment are subjected to continuous high-intensity excavation unloading and periodic water storage loads. The monitoring network faces the dilemma of lacking an absolutely stable benchmark. At this time, the benchmark points on both sides undergo nonlinear creep displacement in both horizontal and vertical directions. Moreover, the complex and variable micro-meteorological environment in the canyon leads to changes in the atmospheric refractive index along the photoelectric measurement path. The data exhibits a non-uniform distribution, rendering conventional fixed-benchmark measurement modes and constant meteorological correction models ineffective. This prevents accurate separation of benchmark drift from the actual deformation of the dam body. Upon system startup, heterogeneous master stations deployed on the unstable rock mass simultaneously acquire photoelectric edge data and satellite baseline vectors. Addressing the satellite signal quality degradation caused by the strong multipath effect in the canyon environment, the system utilizes a high-precision angle observation network of a total station as a rigid verification sieve to check the scale consistency of the GNSS solution baseline, identifying and eliminating scale ratios. For abnormally low-quality baselines, only the verified GNSS baseline vectors are retained as distance scalar constraints for subsequent solutions. This mechanism uses the geometric rigidity of photoelectric angle measurement to compensate for the instability of microwave ranging in complex environments, achieving complementary advantages of heterogeneous data.

[0026] Based on the selected GNSS baseline model length The system establishes distance constraint equations to ensure that the slope distance observed by the total station is within acceptable limits. By converging towards the true value of the microwave distance, which is unaffected by optical refraction, the equivalent atmospheric refraction coefficient under the current optical path can be determined. This inversion process does not require the deployment of external meteorological sensor arrays. It directly utilizes the data closed loop of the measurement system itself to eliminate the trigonometric leveling system error caused by the non-uniform distribution of atmospheric density. The system incorporates the corrected photoelectric observation vector and GNSS baseline vector into the rank-deficient free network adjustment model and introduces a quasi-stable benchmark selection mechanism based on topological strain energy. The system calculates the linear strain of the topological side lengths between benchmark points. and the sum of their strain energies The method identifies a subset of reference points with the most stable geometric relationships as a pseudo-stable reference group. Using the S-reference transformation, the coordinate system of the monitoring network is anchored to the rigid framework of the center of gravity of the geological structure. Thus, even under extreme conditions where the reference points are displaced as a whole, the absolute deformation value of the dam body relative to the deep stable rock mass can still be calculated. This embodiment solves the dual problems of reference instability and refraction interference in the deep canyon environment by constructing a closed-loop processing system that verifies microwave scale by photoelectric angle, corrects photoelectric elevation by microwave distance, and locks geometric reference by topological strain. This method does not rely on a single physical fixed point, but establishes the measurement reference on the geometric consistency of multi-source data and the topological stability of the geological structure.

[0027] Example 2: This example aims to verify the deformation monitoring performance and data processing accuracy of the above-mentioned heterogeneous observation system in a real high mountain canyon environment. The test scenario was selected in a deep V-shaped canyon section with an elevation difference of over 1500 meters. This area not only suffers from severe topographic obstruction but is also affected by the monsoon climate, resulting in drastic diurnal variations in the atmospheric refractive index. The fundamental purpose of the experiment is to objectively evaluate the stability and accuracy of the method of the present invention under multiple interference sources such as dynamic obstruction, multipath effects, and atmospheric disturbances. The test platform was deployed from an elevation of 1850 meters to 2100 meters on the left bank. The system consists of three main measuring stations (coaxial integrated RTS and GNSS) on the rock mass, 12 monitoring points (prisms) distributed on the right bank and the surface of the dam, and a data processing server deployed in the on-site control room. The total station uses an industrial-grade surveying robot with a nominal angle measurement accuracy of 0.5 seconds and a distance measurement accuracy of 1mm+1ppm. The GNSS receiver uses a surveying receiver that supports multi-frequency and multi-satellite calculation. To simulate real construction interference, multiple tower cranes and cable cranes were randomly operating on-site during the test, causing dynamic obstruction of the line of sight to some measuring stations.

[0028] The core parameters of the experiment were set according to strict physical constraints. The GNSS baseline calculation duration was set to 2 hours. This parameter choice balances the convergence accuracy of the baseline calculation with the temporal resolution of deformation monitoring: too short a duration, such as 15 minutes, would lead to a decrease in ambiguity fixation rate and an increase in elevation component error; too long a duration would fail to capture short-term structural dynamic responses. Based on pre-conducted baseline calculation accuracy tests, a 2-hour observation window ensures that the baseline vector mean square error is better than 2 mm, meeting the accuracy threshold as a rigid constraint. The threshold for scale consistency verification was set at 3 times the mean square error. This setting is based on the statistical principle of low-probability events. Any deviation in scale ratio outside this range is considered gross error caused by multipath effects rather than random error. Finally, the strain energy threshold for screening the quasi-stable benchmark is set as follows: This threshold is derived from the empirical value of the critical strain for microcrack propagation in rock mechanics, and is used to distinguish between elastic deformation and plastic slip of rock mass. After the experiment was started, each main station performed a round of synchronous observation every 2 hours. During the continuous 72-hour monitoring period, the system collected heterogeneous data for 36 epochs. In order to intuitively demonstrate the effect of this method on the correction of atmospheric refraction error, a survey line with a long distance of about 800 meters across the valley was selected, and the original trigonometric leveling data and the leveling data corrected by the method of this invention were extracted. See Table 1. This table shows the comparison data of the uncorrected original height difference and the height difference corrected by the refraction coefficient inversion of this invention under different environmental conditions.

[0029] Table 1: Example Table of Elevation Difference Data for Cross-Valley Survey Lines

[0030] Under the low-temperature, statically stable conditions at 02:00, atmospheric refraction is relatively stable, and the error at the default k value (0.14) is still within a controllable range. However, during the high-temperature, strong convection period at 14:00, the huge temperature gradient causes the optical path to bend drastically, and the original height difference deviation reaches the decimeter level. The k value retrieved by this invention jumps to 0.38, and the corrected height difference quickly returns to near the true value with a deviation of +1.5mm. During the cooling phase at 20:00, even negative refraction occurs (k=-0.05), but this method can still accurately capture and correct it, proving that the GNS-based method... The S-distance-constrained refraction inversion mechanism can adaptively follow environmental changes and dynamically compensate for system errors. Furthermore, to verify the effectiveness of the pseudo-stable benchmark screening mechanism under benchmark instability conditions, a simulated displacement of 20mm was artificially applied to one of the benchmark points (TP-03). After introducing this disturbance, the coordinates of the dam monitoring points calculated by the conventional adjustment model showed a false displacement of approximately 15mm, falsely reporting dam deformation. However, the system of this invention, by calculating the topological strain energy, sensitively detected the sudden increase in linear strain at the TP-03 connecting edge. The method automatically eliminates pseudo-stable groups, and the final output of dam body monitoring point displacement fluctuation is only 1.2mm, successfully shielding the interference of benchmark point displacement. In summary, the experimental data objectively proves that the method of this invention has extremely high anti-interference ability and measurement accuracy in complex canyon environment. Through deep fusion of heterogeneous data and inversion of intrinsic parameters, it achieves high-precision and high-reliability monitoring of dam body deformation in water conservancy projects.

[0031] Example 3: This example combines Figures 1 to 3 This section describes methods for monitoring dam deformation during water conservancy construction, such as... Figure 1 As shown, heterogeneous stations are coaxially deployed, principal optical axis coincidence calibration and spatial deviation measurement are performed, and the data synchronous acquisition stage begins. On the one hand, photoelectric observation vectors are acquired, and on the other hand, input GNSS baselines are acquired. Through the scale consistency verification step, abnormal baselines with multipath effects are eliminated based on statistical distribution. The retained data enters the hybrid observation equation set construction stage. At this time, the GNSS baseline modulus is introduced as a rigid scale constraint and distance scalar constraint. Then, the rank-deficient free network adjustment calculation is performed. On the one hand, the relative geometric coordinates and atmospheric refractive index are solved synchronously. The output refractive index is used for optical path bending correction and the zenith distance is corrected by using the inverted refractive index to obtain the precise height difference. On the other hand, the preliminary relative coordinates are output to the quasi-stable benchmark set screening step. Based on the strain energy analysis of the topology connection network, the benchmark constraints are established. Finally, the S-benchmark transformation is performed by combining the precise height difference and benchmark constraints, and the quasi-stable benchmark centroid is used as a constraint to convert to the engineering coordinate system.

[0032] like Figure 2 As shown in the figure, the horizontal axis represents the closed geometric loop number, covering a continuous interval from loop 1 to loop 6, and the vertical axis represents the elevation difference closure error, in millimeters (mm). The solid line in the figure represents the elevation difference closure error before correction, exhibiting a large oscillation pattern between different loops; the dashed line represents the elevation difference closure error after correction, with the value remaining stable and converging near the zero axis. Figure 3 As shown, the physical implementation of this method relies on specific hardware deployment and data transmission topology. The top layer of the system provides satellite signals to the navigation satellite constellation. The ground part includes the left bank rock mass monitoring area and the right bank rock mass monitoring area, with main station A and main station B deployed respectively. Each main station integrates a satellite receiver and an automatic total station, and connects to auxiliary stations or reference points downwards through laser ranging. At the same time, it transmits laser ranging signals to monitoring prisms 01, 02 and 03 located in the central dam construction surface monitoring area. The data collected from each area is aggregated to the on-site control room or data center through wireless or fiber optic backhaul channels. The center is equipped with data processing servers and monitoring terminals for centralized data processing and status monitoring.

[0033] Example 4: In the actual working conditions of deformation monitoring during the construction period of a certain ultra-high arch dam, the monitoring area often faces the challenge of non-stationary environmental disturbances. For example, during the flood discharge period, the water mist that is constantly present in front of the dam affects the atmospheric refractive index along the photoelectric observation path. High-frequency random oscillations occur, and the local rock mass at the dam abutment may experience sudden, non-tectonic step displacements due to rainfall infiltration. Under such combined extreme conditions, conventional static weighted or single robust estimation strategies often fail to converge, leading to distorted solution results. To address this issue, this embodiment employs an iterative solution mechanism based on adaptive variance component estimation and the IGGIII robust scheme. After acquiring initial observation data, the system constructs a hybrid observation equation set including electro-optical ranging, angle measurement, and GNSS baseline vectors. In the initial adjustment stage, the prior unit weight variance ratio of various observation values ​​is set to 1:1. The residual statistics after preliminary solution show that the post-hoc unit weight mean square error of the electro-optical ranging observation values ​​deviates from the nominal accuracy, indicating strong environmental noise pollution.

[0034] To address this phenomenon, the system automatically triggers an iterative weighting procedure. In each iteration, the Helmert variance component estimation formula is used to calculate the actual posterior variance of various observations based on the current residual vector, and the weight matrix is ​​dynamically adjusted accordingly. For electro-optical ranging data severely affected by water fog, the system automatically reduces its relative weight in the adjustment model; while for GNSS baseline vectors less affected by the environment, their weight is maintained or increased. This statistically suppresses the distortion of the overall network shape by low-quality data. The system incorporates IGGIII during the iteration process. The robust estimation scheme performs statistical checks on each observation and benchmark coordinate. When the standardized residual of a benchmark exceeds a preset strong elimination threshold, such as 3.0 times the mean square error, the robust function automatically reduces its equivalent weight to zero. In this embodiment, this mechanism successfully identifies and eliminates anomalous benchmarks on the left bank that have experienced step displacement, preventing displacement deviations from being transmitted to the monitoring point coordinates through network adjustment. To ensure the uniqueness and reliability of the solution, the system sets a strict iteration termination criterion: when the difference norm of all unknown parameters, including coordinate corrections and refractive indices, is less than 1 / 3 in two consecutive iterations. When the maximum correction amount is less than 0.1 mm, the iteration is considered to have converged. In the actual operation of this embodiment, after 4 adaptive iterations, the system successfully achieved the optimal redistribution of weights and the precise isolation of gross errors. The final output results show that even under extreme conditions where the single-point displacement deviation reaches 12.5 mm and is accompanied by strong refraction interference, the system's calculation accuracy for key monitoring points of the dam body remains within 1.8 mm, and the unit weight error returns to the normal level after verification.

[0035] Example 5: In the actual deployment and operation of the water conservancy project monitoring system, to ensure the high consistency and reproducibility of heterogeneous observation data in terms of time synchronization, spatial reference, and physical parameter inversion, this method implements a strict system initialization calibration and parameter injection procedure. Before the start of the first monitoring epoch, the system performs a time reference unification operation between the GNSS receiver and the total station. Through the Network Time Protocol (NTP) or Pulses per Second (PPS) signal, the internal clocks of both are synchronized to the Coordinated Universal Time (UTC) reference, ensuring that the deviation between the photoelectric observation time and the satellite signal reception time is controlled within milliseconds. For coaxially mounted heterogeneous sensors, the system executes a spatial deviation calibration procedure, using a precision level and vernier calipers to measure the vertical deviation of the total station's optical center relative to the GNSS antenna phase center. and horizontal deviation vector These geometric eccentricities are written as fixed correction parameters into the configuration file of the solution model, so that centering correction is automatically performed in subsequent baseline vector constraint calculations.

[0036] Furthermore, for the initial value setting of the atmospheric refractive index inversion model, this method did not adopt commonly used empirical constants. Instead, it performed a baseline calibration procedure based on field micrometeorological characteristics. During the system trial operation phase, typical meteorological days covering the period of maximum diurnal temperature range and drastic temperature and humidity changes were selected. Continuous observations were conducted on multiple baselines of known length (calibrated using a precision rangefinder) using a total station, simultaneously recording temperature, air pressure, and humidity data. Based on these observation data, the system used least squares fitting to derive a basic atmospheric refractive index model suitable for the current geographical environment of the survey area. ,in These are temperature, air pressure, and water vapor pressure, respectively. This basic model is solidified as a priori constraints for the online inverse algorithm, serving as the basis for subsequent real-time refractive index based on GNSS distance constraints. The solution provides high-confidence initial values ​​for the iteration.

[0037] Example 6: In a high-altitude water conservancy monitoring network involving multiple deep river valleys, differences in geological structure, micro-meteorological conditions, and satellite visibility across different monitoring areas make it difficult for universal prior model parameters to adapt to all local environments. To address this issue, this method constructs an offline generation and verification procedure for an adaptive parameter matrix as a key configuration step before the system's formal operation. This procedure, based on historical geological survey data and satellite imagery, divides the monitoring area into several independent geographical units and deploys a temporary high-density verification network for each unit. This verification network includes at least two high-level GNSS reference stations and a set of photoelectric target points covering typical elevation gradients. During the data acquisition phase, the system runs continuously for no less than 72 hours, simultaneously acquiring high-frequency meteorological data (temperature, air pressure, humidity), raw GNSS observations, and automatic multi-recovery corner data from the total station. Using this set of multidimensional heterogeneous data, the system executes a parameter optimization algorithm to independently calculate the optimal atmospheric refractive index model parameters for each geographical unit. and GNSS multipath effect correction model coefficients The objective function for optimization is set as minimizing the sum of squared closure errors of all check baselines within the cell after parameter correction.

[0038] After the parameter calculation is completed, the system automatically generates a configuration file matrix containing the optimal parameter combination for each unit and injects it into the central processing unit of the monitoring system. In subsequent real-time operation, the system automatically calls the corresponding parameter set to correct the data according to the geographic unit ID to which the current station belongs. To verify the effectiveness of the parameter matrix, the procedure also requires a trial run of no less than 24 hours after injection. Only when the repeatability error of the coordinate calculation of each monitoring point is lower than the preset acceptance threshold, such as 1.5mm for plane and 2.0mm for elevation, is the parameter configuration of the area deemed to have passed the verification, and the system officially enters the online monitoring state. This procedure ensures that the monitoring system can always operate with the optimal parameter configuration when facing complex and ever-changing environmental conditions.

[0039] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0040] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for monitoring dam deformation during water conservancy construction, characterized in that, Includes the following steps: Multiple benchmark points were set up on the rock slopes on both sides of the dam in the water conservancy construction area, and multiple monitoring points were set up on the dam surface. A total station and a GNSS receiver were coaxially set up at the points whose field of view covered the monitoring points to form the main station, and the remaining benchmark points were defined as auxiliary stations. Within the monitoring epoch, the total station of each main station is driven to traverse the auxiliary stations and monitoring points within the visible range to obtain photoelectric observation vectors including slant range, horizontal angle and zenith distance, and simultaneously obtain the GNSS baseline vectors between each main station. A hybrid observation equation set is constructed, which includes the photoelectric observation vector error equation and the GNSS baseline vector constraint equation. In the photoelectric observation vector error equation, the atmospheric refractive index of the current optical path is set as an unknown parameter that varies with the station position. Calculate the modulus of the GNSS baseline vector and establish a distance constraint equation based on the modulus. Use the distance constraint equation to ensure that the solution value of the slant range obtained by the total station after correction by the atmospheric refraction coefficient converges to the modulus. The mixed observation equations were adjusted using a rank-deficient free network, and the relative geometric coordinates and atmospheric refractive index of the main station, auxiliary station and monitoring points were calculated simultaneously using the least squares criterion. The calculated atmospheric refractive index is used to correct the optical path curvature of the zenith distance obtained by the total station, and the absolute deformation value of each monitoring point relative to the reference point is calculated based on the corrected zenith distance.

2. The method for monitoring dam deformation during water conservancy construction according to claim 1, characterized in that, Before constructing the hybrid observation equation set, the following steps are also included: using the photoelectric observation vectors of the total station to perform independent free network adjustment, obtaining the initial relative geometric structure of the observation network consisting of the main station, auxiliary station and monitoring points; calculating the scale ratio between the distance between each main station in the initial relative geometric structure and the modulus of the corresponding GNSS baseline vector; constructing a statistical distribution model based on the scale ratio between all main stations, identifying anomalous GNSS baseline vectors whose scale ratio deviates from the confidence interval of the statistical distribution model, and removing the constraint equations corresponding to the anomalous GNSS baseline vectors when constructing the hybrid observation equation set.

3. The method for monitoring dam deformation during water conservancy construction according to claim 1, characterized in that, After performing rank-deficient free network adjustment on the mixed observation equations, the following steps are also included: based on the statistical characteristics of the relative displacement of each benchmark point between the current monitoring epoch and the initial monitoring epoch, an iterative weighted method is used to select a quasi-stable benchmark group from the benchmark points; the displacement vector of each benchmark point in the quasi-stable benchmark group relative to the centroid of the quasi-stable benchmark group is calculated; and the relative geometric coordinates are transformed to the engineering coordinate system using the benchmark transformation method with the centroid of the quasi-stable benchmark group as a constraint.

4. The method for monitoring dam deformation during water conservancy construction as described in claim 3, characterized in that, The steps for selecting a quasi-stable benchmark group from the benchmark points include: constructing a benchmark topology connection network based on the spatial location of the benchmark points; calculating the linear strain value of each topological edge in the benchmark topology connection network according to the results of the rank-deficient free network adjustment; calculating the sum of the linear strain energies of all topological edges associated with each benchmark point, and using the sum as a quantitative index for evaluating the local geological structural stability of each benchmark point; including benchmark points with quantitative indices lower than a preset threshold into the quasi-stable benchmark group, and setting a higher weight for the quasi-stable benchmark group than other benchmark points in coordinate transformation calculations.

5. A method for monitoring dam deformation during water conservancy construction as described in claim 1, characterized in that, The steps for constructing a hybrid observation equation set that includes the photoelectric observation vector error equation and the GNSS baseline vector constraint equation are as follows: calculate the apologetic variance component of the GNSS baseline vector; dynamically allocate the weight matrix of the GNSS baseline vector constraint equation in the adjustment solution based on the apologetic variance component and the nominal accuracy of the total station ranging; and reduce the weight value of the constraint equation corresponding to the GNSS baseline vector in the hybrid observation equation set when the apologetic variance component is greater than the preset accuracy threshold.

6. The method for monitoring dam deformation during water conservancy construction according to claim 1, characterized in that, The steps for correcting the optical path curvature of the zenith distance obtained by the total station using the calculated atmospheric refractive index follow the following correction logic: Substitute the calculated atmospheric refractive index into the trigonometric leveling model to calculate the elevation correction caused by optical path curvature; the calculation of the elevation correction follows the following mathematical relationship: ,in, This is the elevation correction value. The calculated atmospheric refractive index, The slope distance is measured by a total station. The mean radius of curvature of the Earth is given; the elevation correction is superimposed on the original elevation difference observation to obtain the precise elevation difference after eliminating the optical path bending error.

7. A method for monitoring dam deformation during water conservancy construction as described in claim 1, characterized in that, Before adjusting the mixed observation equations, the following steps are also included: identifying closed geometric loops formed by the observation paths of the total station in the observation network; calculating the elevation closure error of each closed geometric loop based on the original observation values; establishing a set of conditional equations with the atmospheric refractive index as the unknown parameter and the elevation closure error as the observed value; solving the set of conditional equations to obtain the initial value of the regional equivalent atmospheric refractive index of the current survey area, and using the initial value as the initial value of the atmospheric refractive index in the mixed observation equations.

8. A method for monitoring dam deformation during water conservancy construction as described in claim 1, characterized in that, In the process of constructing a hybrid observation equation set that includes photoelectric observation vector error equations and GNSS baseline vector constraint equations, for each survey line that simultaneously possesses total station observation data and GNSS baseline vectors, optical ranging error equations and microwave ranging constraint equations are established respectively. A proportional error term proportional to the distance is introduced as an unknown parameter into the optical ranging error equation. The proportional error term is numerically locked using the microwave ranging constraint equation to separate the photoelectric ranging system error caused by atmospheric refraction.

9. A method for monitoring dam deformation during water conservancy construction as described in claim 1, characterized in that, The method also includes: monitoring the time series change rate of atmospheric refractive index in real time during the monitoring process; when the time series change rate exceeds the preset gradient threshold, generating an instruction to increase the number of observations; and driving the total station to perform encrypted observations on points within the visible range according to the instruction; and using the encrypted observation data to update the hybrid observation equation set and re-solve it.

10. A method for monitoring dam deformation during water conservancy construction according to claim 1, characterized in that, The steps for selecting monitoring points from the reference points and coaxially deploying a total station and a GNSS receiver to form a main station include: calibrating the principal optical axes of the total station and the GNSS receiver to ensure they coincide in the vertical direction; measuring the spatial deviation between the optical center of the total station and the phase center of the GNSS receiver; storing the spatial deviation as a fixed parameter in the solution model; and using the spatial deviation to center the GNSS baseline vector when constructing the hybrid observation equation set.

Images