Dam slope monitoring comprehensive observation method, system and equipment based on Beidou GNSS and medium

Abstract

The invention discloses a Beidou GNSS-based dam slope monitoring comprehensive observation method, system and device and a medium, and belongs to the technical field of dam slope monitoring, and the method comprises the steps: deploying a plurality of Beidou GNSS reference stations at a stable bedrock at the periphery of a hydropower station dam area, and receiving satellite signals; collecting displacement data in real time, and synchronously collecting environmental factor data; performing time synchronization on the acquired displacement data and environmental factor data, performing hard synchronization through a PPS pulse signal, and uniformly converting the data after hard synchronization to a dam local coordinate system; de-noising the synchronized displacement data and environmental factor data by using a Kalman filtering technology, and carrying out data fusion; historical displacement data and real-time data are combined, and a machine learning algorithm is used for predicting the future displacement change trend of the dam slope. According to the invention, through combination of multi-base-station networking and Beidou / GNSS positioning technology, the three-dimensional monitoring precision of the dam slope is improved.

Description

Technical Field

[0001] This invention relates to the field of dam slope monitoring technology, specifically to a comprehensive observation method, system, equipment, and medium for dam slope monitoring based on BeiDou GNSS. Background Technology

[0002] Currently, common methods for monitoring dam slopes include leveling, trigonometric leveling, and GPS monitoring. Leveling and trigonometric leveling offer high accuracy, but they are static monitoring methods, unable to achieve real-time, continuous monitoring, and are labor-intensive, inefficient, and susceptible to adverse weather conditions such as rain and fog. GPS monitoring offers advantages such as high real-time performance and automation, but it also has certain drawbacks in practical applications. For example, it may be affected by signal blockage in canyon terrain, and multipath effects can cause millimeter-level errors, especially near the dam's metal structure. Furthermore, its accuracy in the vertical direction is relatively poor.

[0003] Compared with other GNSS systems (such as GPS, GLONASS, and Galileo), the BeiDou Navigation Satellite System has unique advantages. However, existing dam slope monitoring methods based on BeiDou / GNSS still have shortcomings in terms of monitoring station deployment, data processing, and early warning mechanisms. Existing monitoring systems process multi-source data in isolation, lack a fusion arbitration mechanism, and the use of static threshold early warning models leads to a high false alarm rate, making it difficult to meet the needs of high-precision, real-time, and comprehensive monitoring of dam slopes. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by this invention is that traditional monitoring methods, such as leveling and trigonometric leveling, although highly accurate, cannot achieve real-time and continuous monitoring and are subject to environmental factors such as rain and fog. GPS monitoring, while having the advantages of real-time and automation, is prone to multipath effects and poor accuracy in canyon terrain and near metal structures, especially in vertical monitoring, making it difficult to meet the needs of high-precision, large-scale, real-time monitoring of dam slopes.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a comprehensive observation method for dam slope monitoring based on BeiDou GNSS, comprising,

[0007] Multiple BeiDou GNSS reference stations were deployed in the stable bedrock outside the hydropower station dam area to receive satellite signals and determine the coordinates of the reference stations through static observation.

[0008] Based on the coordinates of the base station, multiple monitoring stations were set up at different locations on the dam slope to collect displacement data in real time and environmental factor data simultaneously.

[0009] The collected displacement data and environmental factor data are synchronized in time using PPS pulse signals for hard synchronization, and the hard-synchronized data is then uniformly converted to the local coordinate system of the dam.

[0010] Kalman filtering was used to denoise the synchronized displacement and environmental factor data, and then the data was fused.

[0011] Based on the data fusion, the displacement warning threshold is calculated and adjusted according to changes in environmental factors.

[0012] By combining historical displacement data with real-time data, machine learning algorithms are used to predict the future displacement trend of the dam slope.

[0013] Based on the comparison of real-time monitoring data and prediction results, an early warning mechanism is triggered according to preset rules, and the monitoring frequency is adjusted and safety measures are taken for regulation according to the early warning level.

[0014] As a preferred embodiment of the comprehensive observation method for dam slope monitoring based on BeiDou GNSS described in this invention, the step of receiving satellite signals and determining the coordinates of the reference station through static observation includes recording the carrier phase, pseudorange, and signal arrival time.

[0015] The collected satellite signal data is preprocessed to remove noise signals and extract the carrier phase, pseudorange, and timestamp information of each satellite.

[0016] The coordinates of the first reference station were determined by using the least squares method based on the received carrier phase and pseudorange data, by constructing observation equations and performing iterative calculations.

[0017] By using differential technology, the carrier phases of multiple reference stations are differentially divided to eliminate errors and obtain the coordinates of the second reference station.

[0018] The beneficial effects of this preferred technical solution are as follows: By deploying multiple BeiDou GNSS reference stations in the dam slope area and employing static observation technology for high-precision coordinate calculation, the geographical deviation and local error problems inherent in single reference stations during dam slope monitoring are effectively solved. Data acquisition using carrier phase, pseudorange, and signal arrival time, followed by calculation using the least squares method, ensures high accuracy of the reference station coordinates, thus providing a precise benchmark for subsequent displacement monitoring. Through multi-reference station differential technology, this invention effectively eliminates the influence of atmospheric delay, orbital errors, and other factors, avoiding the error accumulation caused by environmental factors in traditional single-point positioning methods. This ensures monitoring accuracy in complex environments, especially in remote areas or areas with severe signal obstruction.

[0019] As a preferred embodiment of the comprehensive observation method for dam slope monitoring based on BeiDou GNSS described in this invention, the step of synchronizing the collected displacement data and environmental factor data in time, performing hard synchronization through PPS pulse signals, and uniformly converting the hard-synchronized data to the dam local coordinate system includes connecting the displacement data acquisition device and the environmental factor data acquisition device to the synchronization signal source.

[0020] A hardware clock generator is used to receive PPS pulse signals, adjust the internal clock of the device, and calibrate the device time through a clock comparison algorithm, so that the time reference of the displacement data acquisition device and the environmental factor data acquisition device are completely consistent.

[0021] Align the clock signal of each device with the PPS pulse signal, and correct clock drift error through a delay compensation mechanism;

[0022] The time-synchronized displacement data is merged with environmental factor data, and the data is mapped to a unified time reference through a time interpolation algorithm.

[0023] The merged synchronous data is subjected to coordinate transformation, and the synchronous data is uniformly transformed to the local coordinate system of the dam through a coordinate transformation algorithm.

[0024] The beneficial effects of this preferred technical solution are as follows: By employing PPS pulse signal hard synchronization technology, the problem of data inconsistency caused by clock differences among multiple sensors is solved. A hardware clock generator accurately receives the PPS signal, and a clock calibration algorithm ensures that the time bases of all monitoring devices are completely consistent. To address potential clock drift between devices, this invention further introduces a delay compensation mechanism, effectively correcting data time errors caused by clock deviations. This synchronization technology ensures that displacement data and environmental factor data are collected under the same time base, improving the accuracy of multi-source data fusion. The synchronized data is uniformly mapped to a unified time base using a time interpolation algorithm, resolving the problem of time misalignment during multi-device data acquisition and ensuring data consistency.

[0025] As a preferred embodiment of the comprehensive observation method for dam slope monitoring based on BeiDou GNSS described in this invention, the step of using Kalman filtering technology to denoise the synchronized displacement data and environmental factor data, and performing data fusion, includes:

[0026] Using the synchronized displacement data and environmental factor data as input, a Kalman filter model is established, defining the state variables as displacement change and environmental factor change.

[0027] Based on the Kalman filter algorithm, the system state is predicted and the covariance matrix is ​​calculated. The Kalman gain is used to weight each observation.

[0028] The synchronized data is denoised, and the optimal estimate is calculated iteratively using Kalman filtering. The displacement data and environmental factor data are then fused into a unified estimate.

[0029] As a preferred embodiment of the comprehensive observation method for dam slope monitoring based on BeiDou GNSS described in this invention, the calculation of the displacement early warning threshold, and the adjustment of the displacement early warning threshold according to changes in environmental factors, includes the following:

[0030]

[0031]

[0032] Where Δh is the daily water level fluctuation, H is the dam height, R is the three-day cumulative rainfall, and R max The highest three-day rainfall in history, Δt is the temperature difference between the inside and outside of the dam, T is the internal stable temperature, and α, β, and γ are the water level coefficient, rainfall coefficient, and temperature coefficient, respectively.

[0033] As a preferred embodiment of the comprehensive observation method for dam slope monitoring based on BeiDou GNSS described in this invention, the step of using machine learning algorithms to predict the future displacement trend of the dam slope includes...

[0034] Collect historical displacement data, real-time displacement data, and environmental factor data;

[0035] The historical displacement data is detrended, and the real-time data is differencing the historical data.

[0036] Build an LSTM model, using the preprocessed data as input;

[0037] A time series forecasting method is adopted, which trains an LSTM model on past displacement data and environmental factors.

[0038] During training, gradient descent is used to optimize network weights, sliding window technique is used to handle the time dependence of data, and backpropagation of error is used to adjust the weights and biases in the network until the loss function converges to the minimum.

[0039] The LSTM model is used to predict the displacement of the dam slope and output the predicted displacement change.

[0040] As a preferred embodiment of the comprehensive observation method for dam slope monitoring based on BeiDou GNSS described in this invention, the hard synchronization via PPS pulse signals is represented as follows:

[0041] [X,Y,Z] T =R(α,β,γ)·[x,y,z] T+[ΔX,ΔY,ΔZ]

[0042] Where, [X,Y,Z] T Represents hard-synchronized coordinates, [x, y, z] T Represents the coordinates in the three directions of the original measurement coordinate system, [ΔX,ΔY,ΔZ] represents the coordinate deviation, and R(α,β,γ) represents the rotation matrix containing the azimuth angle α of the dam axis, the slope inclination angle β of the dam, and the coordinate system twist angle γ.

[0043] This invention provides a comprehensive observation system for dam slope monitoring based on BeiDou GNSS.

[0044] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a comprehensive observation system for dam slope monitoring based on BeiDou GNSS, comprising: a base station deployment module, used to deploy multiple BeiDou GNSS base stations at stable bedrock outside the hydropower station dam area, receive satellite signals, and determine the coordinates of the base stations through static observation;

[0045] The monitoring station deployment module is used to deploy multiple monitoring stations at different locations on the dam slope based on the coordinates of the base station, to collect displacement data in real time and collect environmental factor data simultaneously.

[0046] The data conversion module is used to synchronize the collected displacement data and environmental factor data in time. It performs hard synchronization through PPS pulse signals and converts the hard-synchronized data to the local coordinate system of the dam.

[0047] The data fusion module is used to denoise the synchronized displacement data and environmental factor data using Kalman filtering technology, and then perform data fusion.

[0048] The threshold calculation module is used to calculate the displacement warning threshold based on the data after data fusion, and adjust the displacement warning threshold according to changes in environmental factors.

[0049] The trend prediction module combines historical displacement data with real-time data and uses machine learning algorithms to predict the future displacement trend of the dam slope.

[0050] The safety control module is used to trigger an early warning mechanism according to preset rules based on the comparison of real-time monitoring data and prediction results, and to adjust the monitoring frequency and take safety measures for control according to the early warning level.

[0051] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the comprehensive observation method for dam slope monitoring based on BeiDou GNSS.

[0052] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that, when the computer program is executed by a processor, it implements the steps of the comprehensive observation method for dam slope monitoring based on BeiDou GNSS.

[0053] The beneficial effects of this invention are as follows: By combining multi-reference station networking with BeiDou / GNSS positioning technology, the three-dimensional monitoring accuracy of dam slopes is improved. The deployment of multiple reference stations effectively eliminates factors such as atmospheric delay, orbital errors, and multipath effects, ensuring high accuracy and stability of displacement data, especially in complex terrain and near metal structures. Furthermore, the use of the Kalman filter algorithm improves data fusion and shortens data convergence time, thereby increasing the response speed of the early warning system. By combining it with a dynamic threshold model for environmental factors, the early warning threshold can be automatically adjusted based on real-time data, reducing false alarms and improving the accuracy and timeliness of early warnings. Combined with LSTM prediction technology, future displacement trends can be predicted more accurately, reducing early warning errors in practical applications. Attached Figure Description

[0054] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 The above is a flowchart of the overall observation method for dam slope monitoring based on BeiDou GNSS provided in one embodiment of the present invention. Detailed Implementation

[0056] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0057] Example 1, referring to Figure 1 This is one embodiment of the present invention, which provides a comprehensive observation method for dam slope monitoring based on BeiDou GNSS, including:

[0058] Step 1: Deploy multiple BeiDou GNSS reference stations in the stable bedrock outside the hydropower station dam area to receive satellite signals and determine the coordinates of the reference stations through static observation;

[0059] Step 2: Based on the coordinates of the base station, multiple monitoring stations are set up at different locations on the dam slope to collect displacement data in real time and environmental factor data simultaneously.

[0060] Step 3: Synchronize the collected displacement data with the environmental factor data in time. Perform hard synchronization using PPS pulse signals and convert the hard-synchronized data to the local coordinate system of the dam.

[0061] Step 4: Use Kalman filtering to denoise the synchronized displacement data and environmental factor data, and then perform data fusion;

[0062] Step 5: Based on the data after data fusion, calculate the displacement warning threshold and adjust the displacement warning threshold according to changes in environmental factors;

[0063] Step 6: Combine historical displacement data with real-time data and use machine learning algorithms to predict the future displacement trend of the dam slope;

[0064] Step 7: Based on the comparison of real-time monitoring data and prediction results, trigger the early warning mechanism according to preset rules, adjust the monitoring frequency according to the early warning level, and take safety measures for regulation.

[0065] Existing dam slope monitoring technologies suffer from insufficient accuracy, poor real-time performance, and poor adaptability to complex environments. While traditional leveling and trigonometric leveling offer high accuracy, their inability to provide real-time monitoring and their susceptibility to environmental influences prevent continuous monitoring of dam slopes. Existing GPS monitoring, though real-time and highly automated, suffers from signal obstruction and multipath effects near canyons or metal structures, leading to decreased accuracy, particularly in the vertical direction. Current early warning systems typically employ static threshold models, which are prone to false alarms or missed alarms and are ill-suited to the dynamic changes and complex environments of dam slopes.

[0066] This embodiment of the method improves the accuracy of three-dimensional monitoring by deploying multiple BeiDou GNSS reference stations and employing reference station networking technology. Each reference station accurately determines its coordinates through static observation, providing a reliable benchmark for subsequent data processing. PPS pulse signal hard synchronization technology ensures data time synchronization at each monitoring point, avoiding inaccurate monitoring caused by synchronization errors.

[0067] The deployment of multiple BeiDou GNSS reference stations ensures the accuracy and stability of dam slope monitoring. Kalman filtering is used to denoise the synchronized data, significantly reducing errors and ensuring data reliability. By combining real-time monitoring data with historical data using an LSTM model, the dam displacement trend for the next few hours can be accurately predicted.

[0068] Example 2, an embodiment of the present invention, provides a comprehensive observation method for dam slope monitoring based on BeiDou GNSS, based on the previous embodiment, including:

[0069] Step 1: Deploying multiple BeiDou GNSS reference stations in stable bedrock outside the hydropower station dam area to receive satellite signals and determine the coordinates of the reference stations through static observation includes the following steps A1-A4:

[0070] A1: Record carrier phase, pseudorange, and signal arrival time;

[0071] A2: Preprocess the collected satellite signal data to remove noise signals and extract the carrier phase, pseudorange, and timestamp information for each satellite;

[0072] A3: The coordinates are calculated using the least squares method. Based on the received carrier phase and pseudorange data, the coordinates of the first reference station are determined by constructing observation equations and performing iterative calculations.

[0073] A4: Using differential technology, the carrier phases of multiple reference stations are differentially divided to eliminate errors and obtain the coordinates of the second reference station.

[0074] In this embodiment of the application, in step 1, the coordinates of the base station are determined by static observation through:

[0075] Multiple GNSS reference stations are deployed in stable bedrock surrounding the hydropower station dam area. When selecting reference stations, it is essential to ensure that their locations are far from water flow, groundwater, vibration sources, and other factors that may affect measurement accuracy. The deployment of reference stations should cover the entire dam slope monitoring area, and the distance between base stations should meet the relative position accuracy requirements. Each reference station is equipped with high-precision GNSS receiving equipment to receive BeiDou, GPS, and other satellite signals.

[0076] Each base station continuously observes for at least 72 hours using a GNSS receiver, recording information such as carrier phase, pseudorange, and signal arrival time of the satellite signal. To improve accuracy, the measurement equipment employs a high-sensitivity receiver to ensure a stable and sufficiently strong signal. During the observation process, the GNSS receiving equipment acquires data once per second and stores it as a time-series file.

[0077] The collected satellite signal data undergoes preprocessing to remove noise. This process includes: removing random noise and intermittent interference signals from the signal, calibrating systematic errors in the pseudorange data, correcting the carrier phase data, and eliminating deviations caused by electromagnetic interference. After data cleaning, the system automatically extracts the carrier phase, pseudorange, and timestamp information for each satellite.

[0078] The preprocessed data is input into the Least Squares (LSM) calculation program to solve for the coordinates of each reference station. The steps are as follows: By constructing observation equations, the carrier phase and pseudorange data of each satellite are used to iteratively calculate the observation equations, minimizing the calculation error. The reference station coordinates are optimized after each iteration. Finally, the precise coordinates of each reference station are obtained through the convergence process of the least squares method.

[0079] To further improve coordinate accuracy, the reference stations correct calculation errors using differential techniques. The steps are as follows: Differential measurements are performed between multiple reference stations. By comparing the carrier phases of different reference stations, atmospheric delay, orbital errors, and other error sources are corrected in the comparison results. Finally, a differential algorithm is used to correct the positioning error of each reference station.

[0080] In an optional implementation, in step 1, the coordinates of the base station can be determined by static observation through:

[0081] Multiple reference stations are deployed on the dam slope, and a reference station with existing stable coordinates is selected. The distance between the reference and reference stations should satisfy the effective range of differential correction. Each reference station uses a high-precision GNSS receiver and exchanges data with the reference station wirelessly.

[0082] The base station employs RTK (Real-Time Kinematic) technology, correcting its position by receiving real-time differential correction signals from the reference station. The base station corrects satellite data by receiving differential signals from the reference station in real time. Based on the received correction signals, the base station adjusts pseudorange and carrier phase data in real time, calculates the corrected coordinate data, and updates it in real time.

[0083] The base station uses RTK technology to continuously receive differential correction signals and constantly update its own coordinates.

[0084] In another alternative implementation, in step 1, the static observation to determine the coordinates of the base station can also be achieved by:

[0085] Multiple reference stations were deployed around the dam, and real-time differential correction was performed at selected measurement sites. Sufficient distances were maintained between each reference station to ensure that the differential data covered all monitoring areas of the dam slope.

[0086] Differential GPS technology enhances the accuracy of base station coordinates through real-time differential correction signals. The steps are as follows: The base station adjusts its measurement data by receiving differential correction signals from the reference station. The base station and reference station exchange data via radio frequencies, receiving differential data in real time. The differential signals can instantly correct errors caused by atmospheric conditions, orbital factors, and other factors, ensuring the accuracy of the measurement data.

[0087] The base station and reference station perform coordinate calculations using differential signals, and the base station uses real-time differential data to correct its coordinate position.

[0088] Furthermore, at least three reference stations are deployed in the stable bedrock surrounding the hydropower station dam area, forming a triangular network with sides of 5-15 km. Each reference station synchronously receives signals from BeiDou B1C / B2a, GPS L1 / L5, and Galileo E1 / E5a. Through more than 72 hours of static observation, precise coordinates with a mean square error of <2 mm are obtained using GAMIT / GLOBK software.

[0089] Step 2: Based on the coordinates of the base station, multiple monitoring stations are deployed at different locations on the dam slope to collect displacement data in real time and simultaneously collect environmental factor data, including the following steps B1-B3:

[0090] B1: Set up monitoring stations at the top, middle, and bottom of the slope and near key structures to ensure that the monitoring stations can fully cover the slope area and receive signals well.

[0091] B2: The monitoring station receiver dynamically adjusts the satellite cutoff elevation angle: 10° in open areas, 15° in canyon terrain, and 20° near metal structures;

[0092] B3: The LAMBDA algorithm is used to fix integer ambiguity. In error control, the tropospheric delay is achieved using the Saastamoinen model with random walk filtering, and the multipath effect is suppressed by SNR weighting and multipath hemispherical diagram.

[0093] Step 3: Time synchronization of the collected displacement data and environmental factor data is performed using PPS pulse signals for hard synchronization. The hard-synchronized data is then uniformly converted to the dam's local coordinate system, including the following steps C1-C5:

[0094] C1: Connect the displacement data acquisition device and the environmental factor data acquisition device to the synchronization signal source;

[0095] C2: Use a hardware clock generator to receive PPS pulse signals, adjust the internal clock of the device, and calibrate the device time through a clock comparison algorithm to ensure that the time base of the displacement data acquisition device is completely consistent with that of the environmental factor data acquisition device.

[0096] C3: Align the clock signal of each device with the PPS pulse signal and correct clock drift error through a delay compensation mechanism;

[0097] C4: Merge the time-synchronized displacement data with environmental factor data, and map the data to a unified time reference using a time interpolation algorithm;

[0098] C5: Perform coordinate transformation on the merged synchronous data, and use a coordinate transformation algorithm to uniformly transform the synchronous data to the local coordinate system of the dam.

[0099] In this embodiment of the application, the time interpolation algorithm in step C4 is:

[0100] The synchronized displacement and environmental factor data are preprocessed. First, the synchronized displacement and environmental factor data are arranged in timestamp order. The timestamp of each data point ensures a clear order in the time series, avoiding data timestamp errors or inconsistencies. After the synchronized data is arranged in chronological order, all timestamp data are divided into several pairs of adjacent time points.

[0101] For each pair of adjacent data points, calculate the time difference between them. This process first obtains the timestamp of each pair of data points, and then calculates the time interval based on the timestamps. The time difference is calculated as follows:

[0102] Δt=t i+1 -t i

[0103] Among them, t i+1 For the timestamp of the next point in time, t i Δt represents the timestamp of the previous time point, and Δt represents the time difference of the current time period.

[0104] Calculate the intermediate data points needed for interpolation. Based on the time difference between each pair of adjacent data points, calculate the intermediate value between these two time points using linear interpolation. The formula for linear interpolation is as follows:

[0105]

[0106] Where y represents the interpolation point, y1 and y2 represent the values ​​of two known data points, t1 and t2 are two time points, and t is the intermediate time point to be interpolated. The same interpolation formula is used for both displacement data and environmental factor data to ensure that the interpolated data is continuous and consistent along the time axis.

[0107] All interpolated data are merged to generate a data sequence with a unified time base. Data at each intermediate time point (including displacement data and environmental factor data) is calculated using an interpolation algorithm. Finally, these interpolated data are combined with the original data to generate a time-aligned and continuous dataset.

[0108] In an optional implementation, the time interpolation algorithm in step C4 can be:

[0109] First, the synchronized displacement and environmental factor data are arranged in timestamp order. All data points are processed in chronological order, and missing time points are inserted into the data. Due to differences in acquisition equipment, data may be missing or timestamps may be misaligned; the interpolation algorithm ensures that each time point has valid synchronized data. For each pair of adjacent data points, their time interval is calculated using the following formula:

[0110] Δt=t i+1 -t i

[0111] After calculating the time difference for each pair of data, the interpolation algorithm will select a suitable interpolation method to fill in the missing data based on the time difference.

[0112] Cubic spline interpolation is used to calculate intermediate data. Spline interpolation interpolates the values ​​between each pair of data points by constructing a cubic polynomial, resulting in smoother data during the interpolation process and avoiding the stair-step effect of linear interpolation. Specifically, spline interpolation calculates the interpolation function based on each pair of adjacent data points and its derivative, and then smoothly transitions the data.

[0113] All intermediate data points calculated using spline interpolation will be merged into the original data sequence to ensure that the time base of all data points is consistent. The interpolation process not only considers time differences but also uses polynomials to describe the relationships between data points in detail, thereby improving the accuracy and reliability of the data.

[0114] In another alternative implementation, the time interpolation algorithm in step C4 can also be:

[0115] The synchronized displacement data and environmental factor data are arranged according to timestamps to ensure that each data point has an accurate timestamp and is arranged in chronological order.

[0116] For each pair of adjacent data points, calculate the time difference:

[0117] Δt=t i+1 -t i

[0118] Here, Δt is the time interval between adjacent data points, and the timestamp can be at the millisecond or even microsecond level to ensure the accuracy of the interpolation.

[0119] Lagrange interpolation is used to interpolate between adjacent data points, and it is particularly suitable for situations where data points are sparse or sampling frequencies are inconsistent. This method interpolates the data by constructing a Lagrange polynomial, and the calculation formula is as follows:

[0120]

[0121] Among them, Li (x) is the Lagrange basis function. The Lagrange interpolation method is used to interpolate each pair of data points to generate new interpolated data points.

[0122] After interpolation is complete, all calculated interpolated data will be merged into the time-synchronized dataset. The final data will include displacement data and environmental factor data for each time point, and all data points will have a consistent time reference. Lagrange interpolation can effectively handle irregularly distributed timestamp data, and is particularly suitable for situations where data sampling is uneven or the intervals are large, providing more accurate time-synchronized data for subsequent monitoring and analysis.

[0123] In this embodiment of the application, the coordinate transformation algorithm in step C5 is achieved through:

[0124] Construct a rotation matrix. The rotation matrix is ​​based on three key parameters of the dam: the dam axis azimuth angle α, the dam slope inclination angle β, and the coordinate system twist angle γ. Using these three parameters, the rotation matrix R is calculated, which is used to transform the original coordinate system (x,y,z) into the dam's local coordinate system (X,Y,Z).

[0125] The original measurement data (x, y, z) is multiplied by the rotation matrix R to obtain the transformed coordinate data (X, Y, Z). The transformation formula is:

[0126] [X,Y,Z] T =R(α,β,γ)·[x,y,z] T +[ΔX,ΔY,ΔZ]

[0127] Where, [x,y,z] T Let R be the original measurement coordinates, and R be the rotation matrix, [X,Y,Z]. T The transformed local coordinate data of the dam is shown below, where [ΔX, ΔY, ΔZ] represents the offset from the original coordinate system to the local coordinate system of the dam.

[0128] In an optional implementation, the coordinate transformation algorithm in step C5 can be achieved by:

[0129] In this approach, the synchronized displacement data is first translated. The translation operation adjusts the origin of the original coordinate system to the starting position of the dam's local coordinate system. The translation formula is as follows:

[0130] x = x + ΔX, y = y + ΔY, z = z + ΔZ

[0131] Where ΔX, ΔY, and ΔZ are the translational offsets from the original coordinate system to the local coordinate system of the dam.

[0132] Then, a rotation matrix is ​​applied to rotate the translated data. The rotation matrix is ​​constructed based on the azimuth angle α, slope angle β, and torsion angle γ of the dam, and the calculation formula is as follows:

[0133] [X,Y,Z] T =R(α,β,γ)·[x′,y′,z′] T

[0134] By using a rotation matrix, the translated coordinate points are transformed into coordinate data in the local coordinate system of the dam.

[0135] In another alternative implementation, the coordinate transformation algorithm in step C5 can also be:

[0136] Coordinate transformation is achieved using Euler angles. First, a coordinate rotation matrix is ​​constructed using Euler angles (azimuth α, pitch β, roll γ). Euler angles describe rotational relationships in three-dimensional space, and the formula for calculating the rotation matrix R is:

[0137] R(α,β,γ)=R z (γ)·R y (β)·R x (α)

[0138] Among them, R x ,R y ,R z These are rotation matrices about the x-axis, y-axis, and z-axis, respectively. Combining these rotation matrices yields a comprehensive rotation matrix R, used to transform data in the original coordinate system.

[0139] Then, the rotation matrix is ​​used to transform the synchronized displacement and environmental factor data from the original coordinate system to the dam's local coordinate system. The formula for the transformed coordinate data is:

[0140] [X,Y,Z] T =R(α,β,γ)·[x,y,z] T

[0141] Step 4: Kalman filtering is used to denoise the synchronized displacement data and environmental factor data, and data fusion is performed, including the following steps D1-D3:

[0142] D1: Using the synchronized displacement data and environmental factor data as input, establish a Kalman filter model, and define the state variables as displacement change and environmental factor change;

[0143] D2: Based on the Kalman filter algorithm, predict the system state and calculate the covariance matrix, and use Kalman gain to weight each observation;

[0144] D3: After denoising the synchronized data, the displacement data and environmental factor data are fused into a unified estimate by iteratively calculating the optimal estimate using Kalman filtering.

[0145] In this embodiment of the application, data fusion in step 4 is performed by:

[0146] Using synchronized displacement and environmental factor data as input, state variables are defined as displacement change and environmental factor change. These variables reflect the temporal changes of the dam slope. Displacement change describes the dynamic changes in slope displacement, while environmental factor change describes the changes in environmental factors affecting the dam slope (such as rainfall and temperature).

[0147] The state transition matrix is ​​used to predict the current state. The covariance matrix is ​​calculated using the following formula:

[0148] P k =AP k-1 A T +Q

[0149] Among them, P k Let A be the covariance matrix at the current moment, A be the state transition matrix, and Q be the process noise covariance matrix, representing the influence of process noise on displacement and environmental factor changes.

[0150] Kalman gain calculation and data fusion: The Kalman gain calculation formula is as follows:

[0151] K k =P k H T HP k H T +R) -1

[0152] Kalman gain is used to weight the observed data, thereby fusing displacement and environmental factor data. The optimal estimation result is obtained by combining the weighted observed values ​​and the predicted values.

[0153] In an optional implementation, data fusion in step 4 can be performed by:

[0154] Using synchronized displacement data and environmental factor data as input, weighting coefficients are set. To ensure more accurate data fusion using the weighted sum method, the weight values ​​w1 (weight of displacement data) and w2 (weight of environmental factor data) can be determined as follows:

[0155] Based on the quality and accuracy of historical monitoring data, different weights are assigned to displacement data and environmental factor data. If the sensor from which the displacement data originates has higher accuracy, it can be assigned a higher weight.

[0156] The weights of environmental factors should be adjusted appropriately based on their varying degrees of influence on slope stability. For example, rainfall can be assigned a higher weight when it has a significant impact on dam slopes.

[0157] The fusion result is calculated using a weighted average method, as shown in the formula:

[0158] D fused =w1·D d +w2·D e

[0159] Among them, D fused For the fused displacement and environmental factor data, D d For displacement data, D e The data represents environmental factors, with w1 and w2 representing the assigned weights. Using this method, the fused data reflects the relative importance of displacement data and environmental factors.

[0160] In another alternative implementation, data fusion in step 4 can also be performed by:

[0161] Wavelet transform is used for data decomposition: First, wavelet transform is applied to the synchronized displacement data and environmental factor data to decompose them into sub-signals with different frequency components. Wavelet transform can divide the signal into a low-frequency part (representing trend changes) and a high-frequency part (representing noise and details).

[0162] Wavelet transforms were performed on displacement data and environmental factor data respectively to generate high-frequency and low-frequency signal components.

[0163] Low-frequency signals represent the main trends of displacement and environmental factors, which are then fused using a weighted average method. This weighting factor can be set based on previous analysis results; a common approach is to determine the weights based on the standard deviation of historical data.

[0164] High-frequency components often contain noise, so they need to be suppressed by filtering methods, such as using wavelet denoising techniques to smooth high-frequency signals and remove fluctuations that are meaningless to the result.

[0165] Finally, inverse wavelet transform is used to combine the processed low-frequency signal with the denoised high-frequency signal to reconstruct the fused displacement and environmental factor data. The reconstructed data will retain the main trends of displacement and environmental factors while suppressing noise components.

[0166] Calculating the displacement warning threshold and adjusting it according to changes in environmental factors includes expressing the following:

[0167]

[0168] Where Δh is the daily water level fluctuation, H is the dam height, R is the three-day cumulative rainfall, and R max The highest three-day rainfall in history, Δt is the temperature difference between the inside and outside of the dam, T is the internal stable temperature, and α, β, and γ are the water level coefficient, rainfall coefficient, and temperature coefficient, respectively.

[0169] Step 6: Combining historical displacement data with real-time data and using machine learning algorithms to predict the future displacement trend of the dam slope includes the following steps E1-E6:

[0170] E1: Collect historical displacement data, real-time displacement data, and environmental factor data;

[0171] E2: Detrend-free processing is performed on historical displacement data, and the real-time data is differentially compared with the historical data;

[0172] E3: Construct an LSTM model, using the preprocessed data as input;

[0173] E4: Employ time series forecasting methods, using an LSTM model trained on past displacement data and environmental factors;

[0174] E5: During training, gradient descent is used to optimize network weights, sliding window technique is used to handle the time dependence of data, and backpropagation of error is used to adjust the weights and biases in the network until the loss function converges to the minimum.

[0175] E6: Predict dam slope displacement using an LSTM model and output the predicted displacement change.

[0176] Hard synchronization via PPS pulse signals is represented as follows:

[0177] [X,Y,Z] T =R(α,β,γ)·[x,y,z] T +[ΔX,ΔY,ΔZ]

[0178] Where, [X,Y,Z] T Represents hard-synchronized coordinates, [x, y, z] T Represents the coordinates in the three directions of the original measurement coordinate system, [ΔX,ΔY,ΔZ] represents the coordinate deviation, and R(α,β,γ) represents the rotation matrix containing the azimuth angle α of the dam axis, the slope inclination angle β of the dam, and the coordinate system twist angle γ.

[0179] Example 3 is an embodiment of the present invention, which provides a comprehensive observation system for dam slope monitoring based on BeiDou GNSS, including:

[0180] The base station deployment module is used to deploy multiple BeiDou GNSS base stations in the stable bedrock outside the hydropower station dam area to receive satellite signals and determine the coordinates of the base stations through static observation.

[0181] The monitoring station deployment module is used to deploy multiple monitoring stations at different locations on the dam slope based on the coordinates of the base station, to collect displacement data in real time and collect environmental factor data simultaneously.

[0182] The data conversion module is used to synchronize the collected displacement data and environmental factor data in time. It performs hard synchronization through PPS pulse signals and converts the hard-synchronized data to the local coordinate system of the dam.

[0183] The data fusion module is used to denoise the synchronized displacement data and environmental factor data using Kalman filtering technology, and then perform data fusion.

[0184] The threshold calculation module is used to calculate the displacement warning threshold based on the data after data fusion, and adjust the displacement warning threshold according to changes in environmental factors.

[0185] The trend prediction module combines historical displacement data with real-time data and uses machine learning algorithms to predict the future displacement trend of the dam slope.

[0186] The safety control module is used to trigger an early warning mechanism according to preset rules based on the comparison of real-time monitoring data and prediction results, and to adjust the monitoring frequency and take safety measures for control according to the early warning level.

[0187] The GNSS subsystem extracts three-dimensional position information from the state vector using the position observation matrix. The inclinometer subsystem converts the measured inclination information into horizontal displacement components using a specific observation matrix. In the first stage, each sensor's data undergoes independent Kalman filtering, including standard steps such as state prediction, covariance update, and Kalman gain calculation. In the second stage, the main filter performs optimal weighted fusion of the local filtering results based on preset information allocation coefficients.

[0188] Furthermore, a displacement early warning threshold is established based on historical displacement data, collecting monitoring data for at least one year, including complete hydrological cycle changes. Statistical analysis is performed on the historical displacement sequence to calculate the daily average displacement μ and standard deviation σ. The initial threshold Th0 is determined according to the following formula:

[0189] Th0=μ+2.5σ

[0190] During the real-time monitoring phase, reservoir water level, rainfall, and dam temperature are collected simultaneously, and the dynamic threshold Th is calculated using the following formula. dynamic The threshold is recalculated every hour and instantaneous fluctuations are eliminated by filtering with a 6-hour moving average.

[0191]

[0192] Where Δh is the daily water level fluctuation, H is the dam height, R is the three-day cumulative rainfall, and R max The highest three-day rainfall in history, Δt is the temperature difference between the inside and outside of the dam, T is the internal stable temperature, and α, β, and γ are the water level coefficient, rainfall coefficient, and temperature coefficient, respectively.

[0193] A two-layer long short-term memory (LSTM) network prediction model was constructed to predict the displacement of the dam slope in the next 3 hours.

[0194] The BeiDou / GNSS displacement data (X / Y / Z directions) is detrended and normalized with five-dimensional features, including reservoir water level difference, 3-hour sliding rainfall, and 24-hour average temperature. Then, it is used to form an input matrix and input into a two-layer long short-term memory network with 64 neurons in the first layer and 32 neurons in the second layer. Finally, a 16-dimensional ReLU fully connected layer is used to output the three-dimensional displacement prediction value. The Sigmoid weighting method is used to smooth the anomalous jump data of the output prediction results to obtain the final prediction data.

[0195] The input features are represented as follows:

[0196]

[0197] Where, d x ,d y ,d z Indicates the displacement in the XYZ directions (mm); w l Indicates the reservoir water level elevation (m); r f t represents the cumulative rainfall over 3 hours (mm). g This represents the temperature gradient of the dam body (°C / m).

[0198] Time encoding is performed, mapping a 24-hour day to a unit circle to capture periodic patterns. This process includes the following steps:

[0199] Data standardization:

[0200]

[0201] Where, μ x σ represents the historical data mean of the feature. x The standard deviation of historical data representing the feature;

[0202] To construct a time series window, enter: Output:

[0203] Performing LSTM forward propagation calculations specifically includes:

[0204] Forget Gate Control:

[0205] f t =σ(W f ·[h t-1 ,X t ]+b f )

[0206] in, Represents the forget gate weight matrix; σ represents the forgetting gate bias vector; σ represents the sigmoid function, which outputs [0,1] and controls the degree of information forgetting.

[0207] Input gate control:

[0208]

[0209] Among them, W i W C Represents the input gate and candidate state weight matrix; i t This indicates the proportion of new information input; Indicates the state of candidate cells, containing potentially new information.

[0210] Output gate control: o t =σ(W o ·[h t-1 ,X t ]+b o )h t =o t ⊙tanh(C t )

[0211] Among them, o t Indicates the proportion of information output to the hidden state; h t This indicates the current hidden state and is passed to the next time step.

[0212] Multilayer network output calculation:

[0213] First layer output:

[0214]

[0215] Second layer output:

[0216]

[0217] The fully connected layer outputs the predicted three-dimensional displacement rate of change:

[0218]

[0219] in, This represents the output layer weight matrix; This represents the output layer bias vector. The loss function calculation includes:

[0220] Huber loss calculation:

[0221]

[0222] Where δ = 1.0 represents the threshold parameter, balancing the advantages of L1 and L2 losses; Y represents the actual displacement change and the Adam optimizer update:

[0223]

[0224] Where η = 0.001 represents the learning rate; θ represents all trainable parameters (W f W i W C W o W y ,b f ,b i ,b C ,b o ,b y ); The bias correction term represents the first-order and second-order moment estimates.

[0225] Sigmoid weighted smoothing is represented as:

[0226]

[0227] The rule-triggered early warning mechanism includes: Primary early warning: triggered by real-time monitoring data, a Level 1 early warning is immediately initiated when the measured value exceeds the dynamic threshold; Secondary early warning: triggered based on prediction results, the early warning level is automatically upgraded to Level 2 when the output result of the prediction model exceeds 80% of the dynamic threshold, and the BeiDou / GNSS sampling frequency is increased; Ultimate early warning: a Level 3 early warning is directly triggered when both the measured and predicted values ​​exceed the limits.

[0228] This embodiment also provides an electronic device applicable to the integrated observation method for dam slope monitoring based on BeiDou GNSS, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the integrated observation method for dam slope monitoring based on BeiDou GNSS proposed in the above embodiment.

[0229] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the comprehensive observation method for dam slope monitoring based on BeiDou GNSS as proposed in the above embodiment.

[0230] The storage medium proposed in this embodiment and the method for comprehensive monitoring of dam slope based on BeiDou GNSS proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0231] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0232] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not 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, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A comprehensive observation method for dam slope monitoring based on Beidou GNSS, characterized in that: Comprising, Deploying multiple Beidou GNSS reference stations at the stable bedrock outside the dam area of the hydropower station, receiving satellite signals, and determining the reference station coordinates through static observation; According to the reference station coordinates, multiple monitoring stations are arranged at different positions of the dam slope, real-time displacement data and synchronous environmental factor data are collected; Time synchronization is performed on the collected displacement data and environmental factor data, hard synchronization is performed through PPS pulse signal, and the hard-synchronized data is uniformly converted to the local coordinate system of the dam; The synchronized displacement data and environmental factor data are denoised and fused using Kalman filtering technology; Based on the data fusion data, the displacement warning threshold is calculated, and the displacement warning threshold is adjusted according to the change of the environmental factor; Combine historical displacement data with real-time data, use machine learning algorithm to predict future displacement trend of dam slope; Based on the comparison of real-time monitoring data and prediction results, the warning mechanism is triggered according to the preset rule, the monitoring frequency is adjusted according to the warning level, and safety measures are taken for regulation and control.

2. The Beidou GNSS-based comprehensive observation method for dam slope monitoring according to claim 1, characterized in that: The receiving satellite signals and determining the reference station coordinates through static observation includes, Recording carrier phase, pseudo-range and signal arrival time; Pretreatment is performed on the collected satellite signal data to remove noise signals and extract carrier phase, pseudo-range and timestamp information of each satellite; Least square method is used for coordinate solution, based on the received carrier phase and pseudo-range data, the observation equation is constructed and iterative calculation is performed, and the first reference station coordinates are solved and determined; Using difference technology, the carrier phase of multiple reference stations is differentiated to eliminate errors and obtain the second reference station coordinates. 3.The comprehensive observation method for dam slope monitoring based on Beidou GNSS according to claim 2, characterized in that: The time synchronization of the collected displacement data and environmental factor data, the hard synchronization through PPS pulse signal, and the conversion of the hard-synchronized data to the local coordinate system of the dam include connecting the displacement data collection equipment and the environmental factor data collection equipment to the synchronous signal source; The PPS pulse signal is received by the hardware clock generator, the internal clock of the equipment is adjusted, and the device time is calibrated through the clock comparison algorithm, so that the time reference of the displacement data collection equipment and the environmental factor data collection equipment is completely consistent; Align the clock signal of each device with the PPS pulse signal, and correct the clock drift error through the delay compensation mechanism; Merge the time-synchronized displacement data and environmental factor data, and map the data to a unified time reference through time interpolation algorithm; Coordinate conversion is performed on the merged synchronized data, and the synchronized data is uniformly converted to the local coordinate system of the dam through coordinate conversion algorithm.

4. The Beidou GNSS-based comprehensive observation method for dam slope monitoring according to claim 3, characterized in that: The synchronized displacement data and environmental factor data are denoised and fused using Kalman filtering technology, including, The synchronized displacement data and environmental factor data are used as input to establish a Kalman filtering model, and the state variables are defined as displacement change and environmental factor change; According to the Kalman filtering algorithm, the system state is predicted and the covariance matrix is calculated, and each observation value is weighted using Kalman gain; The denoising processing is performed on the synchronized data, and after the optimal estimation value is calculated through Kalman filter iteration, the displacement data and the environmental factor data are fused into a unified estimation value.

5. The Beidou GNSS-based comprehensive observation method for dam slope monitoring according to claim 4, characterized in that: The calculation of the displacement early warning threshold value includes adjusting the displacement early warning threshold value according to the change of the environmental factor, which is represented as, Where Δh is the water level amplitude of the day, H is the dam height, R is the three-day cumulative rainfall, R max is the historical maximum three-day rainfall, Δt is the internal / external temperature difference of the dam body, T is the internal stable temperature, and α, β, and γ are the water level coefficient, rainfall coefficient, and temperature coefficient, respectively. 6.The comprehensive observation method for dam slope monitoring based on Beidou GNSS according to claim 5, characterized in that: The prediction of the future displacement change trend of the dam slope using the machine learning algorithm includes, collecting historical displacement data, real-time displacement data and environmental factor data; The historical displacement data is de-trended, and the real-time data is differentiated from the historical data; An LSTM model is constructed, and the preprocessed data is used as input; The time series prediction method is adopted, and the past displacement data and environmental factors are trained through the LSTM model; During the training process, the gradient descent method is used to optimize the network weight, the sliding window technology is used to process the time dependence of the data, and the weight and bias in the network are adjusted through error back propagation until the loss function converges to a minimum value; The LSTM model is used to predict the displacement of the dam slope, and the predicted displacement change is output.

7. The Beidou GNSS-based comprehensive observation method for dam slope monitoring according to claim 6, characterized in that: The hard synchronization through the PPS pulse signal is represented as, [X, Y, Z] T = R(a, b, g) - [x, y, z] T + [AX, AY, AZ] where [X, Y, Z] T represents the hard synchronization coordinates, [x, y, z] T represents the coordinates of the three directions in the original measurement coordinate system, [ΔX, ΔY, ΔZ] represents the coordinate deviation, and R(α, β, γ) represents a rotation matrix containing the dam axis azimuth angle α, the dam slope inclination angle β, and the coordinate system torsion angle γ.​​ 8. A dam slope monitoring comprehensive observation system based on Beidou GNSS, applying the dam slope monitoring comprehensive observation method based on Beidou GNSS according to any one of claims 1-7, characterized in that, It comprises: The reference station deployment module is used to deploy a plurality of Beidou GNSS reference stations on the stable bedrock outside the dam area of the hydropower station, receive satellite signals, and determine the reference station coordinates through static observation; The monitoring station layout module is used to layout a plurality of monitoring stations at different positions of the dam slope according to the reference station coordinates, and real-time collection of displacement data and synchronous collection of environmental factor data; The data conversion module is used to time synchronize the collected displacement data and environmental factor data, hard synchronize through the PPS pulse signal, and uniformly convert the hard synchronized data to the local coordinate system of the dam; The data fusion module is used to denoise the synchronized displacement data and environmental factor data using Kalman filter technology, and perform data fusion; The threshold value calculation module is used to calculate the displacement early warning threshold value based on the data after data fusion, and adjust the displacement early warning threshold value according to the change of the environmental factor; The trend prediction module is used to combine the historical displacement data with the real-time data, and predict the future displacement change trend of the dam slope using the machine learning algorithm; The safety control module is used to trigger the early warning mechanism according to the preset rules based on the comparison of the real-time monitoring data and the prediction results, adjust the monitoring frequency according to the early warning level, and take safety measures for regulation and control. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. The processor executes the computer program to realize the steps of the Beidou GNSS-based comprehensive observation method for dam slope monitoring according to any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to realize the steps of the Beidou GNSS-based comprehensive observation method for dam slope monitoring according to any one of claims 1 to 7.

Images