Dam deformation trend prediction system based on Beidou and GNSS data

Abstract

The invention relates to the field of hydraulic engineering safety, and discloses a big dipper and GNSS data-based dam deformation trend prediction system, which comprises a data acquisition and transmission module, a data preprocessing module, a feature engineering module, a model training and prediction module, a database management module, a result display and early warning module and a system management module, the data acquisition and transmission module is used for acquiring data from the GNSS receiver and the environment sensor and transmitting the data to the processing center; the data preprocessing module is used for carrying out preprocessing operation such as cleaning, denoising and coordinate conversion on the collected data. According to the dam deformation trend prediction system based on Beidou and GNSS data, temperature changes of all parts of a dam are monitored in real time, expansion or contraction caused by temperature fluctuation is captured, it is ensured that potential deformation is found in time, rainfall data are integrated, and the influence of rainfall on the reservoir level and the dam load is predicted.

Description

Technical Field

[0001] The present invention relates to the field of water conservancy project safety, and particularly to a dam deformation trend prediction system based on Beidou and GNSS data. Background Technique

[0002] The safety of water conservancy projects refers to a comprehensive measure to ensure the normal and safe operation of project structures and equipment during the construction, operation, and management of water conservancy projects, prevent accidents, ensure that the projects can effectively manage and utilize water resources, and protect the surrounding ecological environment and the safety of residents. Dams play a very important role in water conservancy projects. They are not only used for water storage and power generation, but also related to flood control, irrigation, and other aspects.

[0003] The safety of dams and water conservancy projects is a comprehensive issue involving multiple aspects such as technology and environment. Through scientific management and effective monitoring, the safety of water conservancy projects can be guaranteed to the greatest extent, promoting sustainable development. With the continuous progress of technology and the increasing awareness of environmental protection in society, the safety management of water conservancy projects needs to be more scientific and systematic. Among them, the dam deformation trend prediction system is a comprehensive monitoring and prediction method, which mainly uses satellite positioning technology to monitor the minute deformation of dams and predict their future deformation trends.

[0004] Existing prediction systems ignore many problems. For example, rainfall has a very significant impact on the dam deformation prediction system. First of all, rainfall will affect the water pressure of the dam and increase the water level of the reservoir, which may cause the deformation of the dam body. In addition, rainfall may also cause seepage or infiltration of the dam body and foundation, affecting the stability of the structure. Moreover, the accompanying temperature and humidity changes during rainfall may also affect the performance of the dam body materials. Dams are large structures, and building materials such as concrete have the property of thermal expansion and contraction. Temperature changes will cause the expansion or contraction of the materials, thus causing the deformation of the dam, especially in different seasons or at different times of the day, the temperature changes may be relatively large. GNSS signals may be interfered with during transmission, such as multipath effects, satellite signals being blocked, or electromagnetic interference, etc. These interferences may affect the accuracy of the data and even cause data loss. Existing prediction systems do not consider these potential problems and there is great room for improvement. Summary of the Invention

[0005] (1) Technical Problems to be Solved

[0006] In view of the deficiencies of the prior art, the present invention provides a dam deformation trend prediction system based on Beidou and GNSS data to solve the above problems.

[0007] (2) Technical Solutions

[0008] To achieve the above object, the present invention provides the following technical solutions: A dam deformation trend prediction system based on Beidou and GNSS data, including a data acquisition and transmission module, a data preprocessing module, a feature engineering module, a model training and prediction module, a database management module, a result display and warning module, and a system management module. The data acquisition and transmission module is used to obtain data from GNSS receivers and environmental sensors and transmit the data to the processing center; the data preprocessing module is used to perform preprocessing operations such as cleaning, denoising, and coordinate transformation on the acquired data; the feature engineering module is used to extract time features and environmental features from the preprocessed data to provide meaningful inputs for model training; the model training and prediction module is used to train a deformation prediction model using historical data and predict future deformation trends; the database management module is used to store and manage all data, including raw data, processed data, and prediction results; the result display and warning module is used to visualize the prediction results and set warning thresholds to issue alarms when the predicted deformation exceeds the thresholds; the system management module is used to manage user permissions, system parameter settings, and log records.

[0009] According to another aspect of the embodiments of the present invention, there is also provided a method for predicting the dam deformation trend based on Beidou and GNSS data, including the following steps:

[0010] S1. Data acquisition: Install GNSS receivers at key structural points of the dam and adopt a continuous monitoring mode to capture rapid changes;

[0011] S2. Data preprocessing: Remove noise and outliers, use filters for correction, ensure the time synchronization of all receivers, and use GPS time as a reference;

[0012] S3. Feature extraction: Extract the displacement data of each monitoring point of the dam in the x, y, and z directions, calculate the time derivative of the displacement, obtain the deformation rate and acceleration, and solve the influence of seasonal patterns and environmental data (rainfall, temperature);

[0013] S4. Model training and verification: Extract features from the preprocessed GNSS data, including the three-dimensional displacement data and their timestamps of each monitoring point of the dam, divide the data set into a training set and a verification set. Usually, 80% of the data is used for training and 20% for verification. Normalize the data to ensure that the input data is on a similar scale and improve the model training efficiency. Select LSTM (Long Short-Term Memory Network) as the prediction model; select the mean square error (MSE) as the loss function and select the Adam optimizer; increase the number of layers or neurons of the LSTM layer to enhance the model's expression ability, find the best hyperparameter combination through grid search or random search, and combine physical models for training. Integrate the trained model into the dam deformation prediction system for real-time prediction;

[0014] S5. Deformation Trend Prediction: Establish a real-time data processing pipeline to ensure that the latest GNSS data enters the model, analyze the predicted values and their uncertainties, such as confidence intervals, to judge risks;

[0015] S6. Early Warning and Fault Tolerance Mechanism: Set thresholds according to historical data or safety specifications. At the same time, use multiple satellite systems to improve the reliability and redundancy of signals, reduce the impact of single-system failures, use multiple data transmission channels to ensure the reliability of data transmission, adopt data verification and retransmission mechanisms to ensure data integrity, adopt distributed storage or cloud storage methods to ensure data backup and security, prevent data loss caused by storage device failures, and conduct system tests and simulated fault drills regularly to test the fault tolerance and emergency response capabilities of the system

[0016] It should be noted that the key structural points are the center line and both sides of the dam crest, the central area and both sides of the dam foundation, the side close to the reservoir (the upstream face) and the back side (the downstream face), the middle part of the dam body, expansion joints, settlement joints, the top of the gate, both sides of the spillway, the crown cantilever of the arch dam, and the core wall of the earth-rock dam.

[0017] It should be noted that the specific implementation of feature extraction is as follows: convert the timestamp into date, month, and season, calculate the number of days or hours since the reference time to represent the time trend, directly use the displacement values: the displacement values in the x, y, and z directions, use the difference method or fitting method to calculate the change rate of displacement, collect data such as temperature, rainfall, and water level, and align them according to the timestamp, consider the lag effect of environmental factors such as rainfall on the dam deformation, calculate the rolling mean and rolling variance of the displacement data, calculate the maximum value, minimum value, and average value within a certain period of time, perform Fourier transform on the displacement time series, extract the frequency spectrum, extract the amplitude and phase of the main frequency components, and use a moving window to calculate the slope of the displacement curve to reflect the degree of curvature of the curve.

[0018] It should be noted that the specific implementation of combining with the physical model is as follows: establish a simplified linear elastic model, describe the relationship between displacement and load according to the structural parameters and material properties of the dam, determine the input variables of the model, ensure that these variables are aligned with the actual monitoring data, collect and organize GNSS monitoring data, environmental data, and the input data of the physical model, perform data interpolation or resampling to ensure the consistency of all data in time and space, design the LSTM network structure, the input includes historical displacement data, the predicted values of the physical model, and environmental factors, set appropriate sequence lengths, batch sizes, and training parameters, divide the training set and validation set, train the LSTM model, ensure the consistency of the physical model and the LSTM model in spatial and temporal scales through data interpolation or resampling, and adopt dimensionality reduction or feature selection methods to simplify the calculation.

[0019] It should be noted that the training parameters are set as follows: the number of epochs is 50; the batch size is 64; during the training process, monitor the loss of the validation set. If there is no improvement within a certain number of epochs, stop training early to prevent overfitting;

[0020] The model architecture is as follows:

[0021] Input layer: Accept time series data, and the dimension is determined according to the number of features;

[0022] LSTM layer: Add one or more LSTM layers to capture time dependencies;

[0023] Dropout layer: Add a Dropout layer between LSTM layers to prevent overfitting;

[0024] Output layer: A fully connected layer that outputs the prediction results.

[0025] It should be noted that the specific implementation to address the impact of temperature factors is to synchronize the time of temperature data with GNSS data, match the timestamps through interpolation processing, perform filtering to remove noise, detect and process outliers in the temperature data, combine the temperature data with other features, perform feature selection, install temperature sensors at the dam crest, dam foundation, and different depths to ensure data representativeness, set a temperature change threshold to trigger the model to re-evaluate or update, and under extreme temperature conditions, increase the prediction frequency and use an online learning algorithm to enable the model to be updated in real time to adapt to temperature changes.

[0026] It should be noted that the specific implementation to address the impact of rainfall factors is to incorporate rainfall as an important feature into the machine learning model, divide different levels according to rainfall intensity, analyze its impact on deformation, introduce a time lag effect in the model to reflect the delayed impact of rising water levels and seepage after rainfall, consider the spatial distribution of rainfall, evaluate its uneven impact on the pressure of different parts of the dam body, analyze the correlation between rainfall and deformation through historical data, optimize the model parameters, combine physical models and statistical models, integrate the physical impact of rainfall and data-driven statistical laws, avoid over-reliance on rainfall data during model training to ensure the generalization ability of the model, design a special processing module for extreme rainfall events to enhance the robustness of the model, and consider the interaction between rainfall and other environmental factors such as temperature and humidity to improve the prediction accuracy.

[0027] According to another aspect of the embodiments of the present invention, there is also provided a computer-readable medium, in which computer-executable instructions are stored. When the computer-executable instructions are loaded and executed by a processor, the steps of the method according to any one of claims 2 to 8 are implemented.

[0028] According to another aspect of the embodiments of the present invention, there is also provided a computer program product, which is tangibly stored on a non-transitory computer-readable medium and includes machine-executable instructions that, when executed, cause the machine to perform the steps of the method according to any one of claims 2 to 8.

[0029] (III) Beneficial effects

[0030] Compared with the prior art, the present invention provides a dam deformation trend prediction system based on Beidou and GNSS data, which has the following beneficial effects:

[0031] The dam deformation trend prediction system based on Beidou and GNSS data can monitor the temperature changes of each part of the dam in real time, capture the expansion or contraction caused by temperature fluctuations, ensure the timely discovery of potential deformations, integrate rainfall data, predict the impact of rainfall on the reservoir water level and the dam load. Especially under strong rainfall conditions, the system can give early warnings of possible deformations to ensure the safety of the dam. By combining the dam structure design and material characteristics, using mechanical principles to simulate the deformation of the dam under different loads, it improves the accuracy and reliability of the prediction. It can automatically learn the complex relationships between temperature, rainfall, etc. and the dam deformation, enhancing the intelligence level of the prediction. It has high precision, real-time performance, reliability and scalability, is applicable to different types of large structures, and provides strong technical support for practical applications. Detailed implementation manners

[0032] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0033] It should be understood that in various embodiments of the present invention, the order numbers of the processes do not mean the order of execution. The execution order of each process should be determined according to its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0034] It should be understood that in the present invention, "including" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device that includes a series of steps or units does not necessarily have to be limited to those steps or units clearly listed, but may include other steps or units not clearly listed or inherent to these processes, methods, products or devices.

[0035] A dam deformation trend prediction system based on Beidou and GNSS data, including a data acquisition and transmission module, a data preprocessing module, a feature engineering module, a model training and prediction module, a database management module, a result display and warning module, and a system management module. The data acquisition and transmission module is used to obtain data from GNSS receivers and environmental sensors and transmit the data to the processing center; the data preprocessing module is used to perform preprocessing operations such as data cleaning, denoising, and coordinate transformation on the acquired data; the feature engineering module is used to extract time features and environmental features from the preprocessed data to provide meaningful inputs for model training; the model training and prediction module is used to train a deformation prediction model using historical data and predict future deformation trends; the database management module is used to store and manage all data, including raw data, processed data, and prediction results; the result display and warning module is used to visualize the prediction results and set warning thresholds to issue alarms when the predicted deformation exceeds the thresholds; the system management module is used to manage user permissions, system parameter settings, and log records.

[0036] According to another aspect of the embodiments of the present invention, there is also provided a method for predicting the dam deformation trend based on Beidou and GNSS data, including the following steps:

[0037] S1. Data acquisition: Install GNSS receivers at key structural points of the dam and capture rapid changes in continuous monitoring mode;

[0038] S2. Data preprocessing: Remove noise and outliers, use filters for correction, ensure time synchronization of all receivers, and use GPS time as a reference;

[0039] S3. Feature extraction: Extract displacement data of each monitoring point of the dam in the x, y, and z directions, calculate the time derivative of the displacement, obtain the deformation rate and acceleration, and solve the influence of seasonal patterns and environmental data (rainfall, temperature);

[0040] S4. Model training and verification: Extract features from the preprocessed GNSS data, including the three-dimensional displacement data and their timestamps of each monitoring point of the dam, divide the data set into a training set and a verification set, usually using 80% of the data for training and 20% for verification, normalize the data to ensure that the input data is on a similar scale and improve the model training efficiency, select LSTM (Long Short-Term Memory Network) as the prediction model; select the mean square error (MSE) as the loss function, select the Adam optimizer; increase the number of layers or neurons of the LSTM layer to enhance the model's expression ability, find the best hyperparameter combination through grid search or random search, and combine with a physical model for training, integrate the trained model into the dam deformation prediction system for real-time prediction;

[0041] S5. Deformation Trend Prediction: Establish a real-time data processing pipeline to ensure that the latest GNSS data enters the model, analyze the predicted values and their uncertainties, such as confidence intervals, to judge risks;

[0042] S6. Early Warning and Fault Tolerance Mechanism: Set thresholds according to historical data or safety specifications. At the same time, use multiple satellite systems to improve the reliability and redundancy of signals, reduce the impact of single-system failures, use multiple data transmission channels to ensure the reliability of data transmission, adopt data verification and retransmission mechanisms to ensure data integrity, adopt distributed storage or cloud storage methods to ensure data backup and security, prevent data loss caused by storage device failures, and conduct system tests and simulated fault drills regularly to test the fault tolerance and emergency response capabilities of the system

[0043] It should be noted that the continuous monitoring mode is specifically implemented as follows: Set a high-frequency data acquisition interval, collect data once per second, use a high-precision and multi-system compatible GNSS receiver, use a stable communication network to transmit data in real time, and be equipped with wireless transmission equipment.

[0044] It should be noted that the key structural points are the center line and both sides of the dam crest, the central area and both sides of the dam foundation, the side close to the reservoir (the upstream face) and the back side (the downstream face), the middle part of the dam body, expansion joints, settlement joints, the top of the gate, both sides of the spillway, the crown beam of the arch dam, and the core wall of the earth-rock dam.

[0045] It should be noted that the feature extraction is specifically implemented as follows: Convert the timestamp into date, month, and season, calculate the number of days or hours since the reference time to represent the time trend, directly use the displacement values: the displacement values in the x, y, and z directions, use the difference method or fitting method to calculate the change rate of displacement, collect data such as temperature, rainfall, and water level, and align them according to the timestamp, consider the lag effect of environmental factors such as rainfall on the dam deformation, calculate the rolling mean and rolling variance of the displacement data, calculate the maximum value, minimum value, and average value within a certain period of time, perform Fourier transform on the displacement time series to extract the frequency spectrum, extract the amplitude and phase of the main frequency components, and use a moving window to calculate the slope of the displacement curve to reflect the degree of curvature of the curve.

[0046] It should be noted that the specific implementation in combination with the physical model is as follows: establish a simplified linear elastic model, describe the relationship between displacement and load according to the structural parameters and material properties of the dam, determine the input variables of the model, ensure that these variables are aligned with the actual monitoring data, collect and organize GNSS monitoring data, environmental data, and the input data of the physical model, perform data interpolation or resampling to ensure the consistency of all data in time and space, design the LSTM network structure, the input includes historical displacement data, the predicted values of the physical model, and environmental factors, set appropriate sequence lengths, batch sizes, and training parameters, divide the training set and the validation set, train the LSTM model, ensure the consistency of the physical model and the LSTM model in the spatial and temporal scales through data interpolation or resampling, and adopt dimensionality reduction or feature selection methods to simplify the calculation.

[0047] It should be noted that the training parameters are set as follows: the number of epochs is 50 epochs; the batch size is 64; monitor the loss of the validation set during the training process, and if there is no improvement within a certain number of epochs, stop training in advance to prevent overfitting.

[0048] The model architecture is as follows:

[0049] Input layer: Accept time series data, and the dimension is determined according to the number of features.

[0050] LSTM layer: Add one or more LSTM layers to capture time dependence.

[0051] Dropout layer: Add a Dropout layer between the LSTM layers to prevent overfitting.

[0052] Output layer: A fully connected layer that outputs the prediction results.

[0053] It should be noted that the specific implementation to solve the influence of temperature factors is as follows: synchronize the time of temperature data and GNSS data, match the timestamps through interpolation processing, perform filtering processing to remove noise, detect and process outliers in the temperature data, combine the temperature data with other features, perform feature selection, install temperature sensors at the dam crest, dam foundation, and different depths to ensure data representativeness, set the temperature change threshold to trigger the model to re-evaluate or update, under extreme temperature conditions, increase the prediction frequency, and use an online learning algorithm to enable the model to be updated in real time to adapt to temperature changes.

[0054] Among them, historical temperature data and corresponding dam deformation data are collected and analyzed to obtain temperature data and dam deformation monitoring data in the past few years. Noise and outliers are removed to ensure data quality. The Pearson correlation coefficient is used to analyze the relationship between temperature and deformation. The significance threshold of temperature change is determined by standard deviation or percentile. According to the results of historical data analysis, a temperature change threshold is set. Statistical method to determine the threshold: Calculate the standard deviation of the temperature data, and set the threshold as the mean plus or minus twice the standard deviation. Seasonal adjustment: Adjust the threshold range according to the temperature characteristics of different seasons. Summer temperature threshold: average temperature ± 2σ, winter temperature threshold: average temperature ± 1.5σ. Stochastic Gradient Descent is used as the online learning algorithm to establish an initial prediction model. A data pipeline is established to ensure that newly collected temperature and deformation data are input into the model in real time. Each time new data is received, the SGD algorithm is used to update the model parameters. The update frequency of SGD is adjusted according to the severity of temperature change. The system load is monitored in real time to ensure sufficient computing resources. A sliding window is used to regularly evaluate the performance of the model on recent data, and the learning rate or other hyperparameters of SGD are adjusted according to the evaluation results.

[0055] It should be noted that the specific implementation to address the impact of rainfall factors is to incorporate rainfall as an important feature into the machine learning model. Different levels are divided according to rainfall intensity to analyze its impact on deformation. The time lag effect is introduced into the model to reflect the delayed impact of rising water level and seepage after rainfall. The spatial distribution of rainfall is considered to evaluate its uneven impact on the pressure of different parts of the dam. The correlation between rainfall and deformation is analyzed through historical data to optimize the model parameters. The physical model and the statistical model are combined to integrate the physical impact of rainfall and the data-driven statistical law. Over-reliance on rainfall data is avoided during model training to ensure the generalization ability of the model. A special processing module is designed for extreme rainfall events to enhance the robustness of the model. The interaction between rainfall and other environmental factors such as temperature and humidity is considered to improve the prediction accuracy.

[0056] Among them, collect the rainfall, temperature, and humidity data in the area where the dam is located, as well as the GNSS monitoring data of the dam, ensure the accuracy and integrity of the data, process missing values and outliers, create interaction features such as rainfall × temperature, rainfall × humidity, and temperature × humidity, consider the triple product feature of rainfall × temperature × humidity, add the lag feature of rainfall, the rainfall of the previous day, and interact with temperature and humidity, use Lasso regression for feature selection, filter out important interaction features, select linear regression for training, evaluate the model performance on the validation set to ensure the generalization ability, conduct an F-test on the selected interaction features to confirm their significance, use Lasso regression to handle the problem of multicollinearity, use SHAP values to explain the contributions of each feature in the model, especially the role of interaction features, and further optimize the model according to the above analysis, which may involve adding more features or adjusting model parameters, and evaluate the performance of the final model on an independent test set to ensure the prediction accuracy and stability.

[0057] According to another aspect of the embodiments of the present invention, there is also provided a computer-readable medium storing computer-executable instructions that, when loaded and executed by a processor, implement the steps of the method according to any one of claims 2 to 8.

[0058] According to another aspect of the embodiments of the present invention, there is also provided a computer program product tangibly stored on a non-transitory computer-readable medium and including machine-executable instructions that, when executed, cause the machine to perform the steps of the method according to any one of claims 2 to 8.

[0059] In summary, the present invention provides a dam deformation trend prediction system based on Beidou and GNSS data, which can monitor the temperature changes of various parts of the dam in real time, capture the expansion or contraction caused by temperature fluctuations, ensure the timely discovery of potential deformations, integrate rainfall data, and predict the impact of rainfall on the reservoir water level and dam load. Especially under strong rainfall conditions, the system can give early warnings of possible deformations to ensure the safety of the dam. By combining the dam structure design and material properties, using mechanical principles to simulate the deformation of the dam under different loads, the accuracy and reliability of the prediction are improved. The system can automatically learn the complex relationships between temperature, rainfall, etc. and dam deformation, enhance the intelligent level of prediction, and has high precision, real-time performance, reliability, and scalability, and is applicable to different types of large structures, providing strong technical support for practical applications.

[0060] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A dam deformation trend prediction system based on Beidou and GNSS data, including data acquisition and transmission module, data preprocessing module, feature engineering module, model training and prediction module, database management module, result display and early warning module, system management module, characterized in that: The data acquisition and transmission module is used to obtain data from the GNSS receiver and environmental sensors and transmit the data to the processing center; The data preprocessing module is used to perform preprocessing operations such as cleaning, denoising, and coordinate conversion on the collected data; the feature engineering module is used to extract time features and environmental features from the preprocessed data to provide meaningful input for model training; the model training and prediction module is used to train the deformation prediction model using historical data and predict future deformation trends; The database management module is used to store and manage all data, including raw data, processed data and prediction results; The result display and warning module is used to visualize the prediction results and set the warning threshold. When the predicted deformation exceeds the threshold, an alarm is issued. The system management module is used to manage user permissions, system parameter settings, and log records.

2. A method for predicting dam deformation trend based on Beidou and GNSS data, characterized in that: The following steps are involved: S1. Data collection: GNSS receivers are installed at key structural points of the dam to capture rapid changes in continuous monitoring mode; S2. Data preprocessing: remove noise and outliers, use filters for correction, ensure time synchronization of all receivers, use GPS time as reference; S3. Feature extraction: Extract the displacement data of each monitoring point of the dam in the x, y, and z directions, calculate the time derivative of the displacement, obtain the deformation rate and acceleration, and solve the influence of seasonal patterns and environmental data (rainfall, temperature); S4. Model training and verification: Extract features from the preprocessed GNSS data, including the 3D displacement data of each monitoring point of the dam and its timestamp, divide the data set into a training set and a verification set, usually using 80% of the data for training and 20% for verification, normalize the data to ensure that the input data is on a similar scale, improve the model training efficiency, select LSTM (Long Short-Term Memory Network) as the prediction model; select mean square error (MSE) as the loss function, and select Adam optimizer; increase the number of layers or neurons of the LSTM layer to enhance the model's expression ability, find the best hyperparameter combination through grid search or random search, and combine the physical model for training, integrate the trained model into the dam deformation prediction system for real-time prediction; S5. Deformation trend prediction: Establish a real-time data processing pipeline to ensure that the latest GNSS data enters the model, analyze the predicted value and its uncertainty, such as confidence interval, to determine the risk; S6. Early warning and fault-tolerance mechanism: Set thresholds based on historical data or safety specifications, use multiple satellite systems at the same time to improve signal reliability and redundancy, reduce the impact of single system failures, use multiple data transmission channels to ensure the reliability of data transmission, adopt data verification and retransmission mechanisms to ensure data integrity, use distributed storage or cloud storage to ensure data backup and security, prevent data loss due to storage device failures, conduct regular system tests and simulated failure drills to verify the system's fault tolerance and emergency response capabilities.

3. The method for predicting dam deformation trend based on Beidou and GNSS data according to claim 2 is characterized in that: The key structural points are the center line of the dam crest and its two side edges, the central area and two side edges of the dam foundation, the side close to the reservoir (the water-facing side) and the side facing away from the water (the back water side), the middle of the dam body, expansion joints, settlement joints, the top of the gate, both sides of the spillway, the crown beam of the arch dam, and the core wall of the earth-rock dam.

4. The method for predicting dam deformation trend based on Beidou and GNSS data according to claim 2 is characterized in that: The specific implementation of feature extraction is to convert the timestamp into date, month, and season, calculate the number of days or hours since the base time to represent the time trend, directly use the displacement value: the displacement value in the three directions of x, y, and z, use the difference method or fitting method to calculate the rate of change of displacement, collect temperature, rainfall, water level and other data, and align them by timestamp, consider the lagging effect of environmental factors such as rainfall on the deformation of the dam, calculate the rolling mean and rolling variance of the displacement data, calculate the maximum, minimum, and average values ​​over a period of time, perform Fourier transform on the displacement time series, extract the frequency spectrum, extract the amplitude and phase of the main frequency components, and use a moving window to calculate the slope of the displacement curve to reflect the curvature of the curve.

5. The method for predicting dam deformation trend based on Beidou and GNSS data according to claim 2 is characterized in that: The specific implementation of the physical model is to establish a simplified linear elastic model, describe the relationship between displacement and load according to the structural parameters and material properties of the dam, determine the input variables of the model, ensure that these variables are aligned with the actual monitoring data, collect and organize GNSS monitoring data, environmental data and input data of the physical model, perform data interpolation or resampling to ensure the consistency of all data in time and space, design the LSTM network structure, input including historical displacement data, predicted values ​​of the physical model and environmental factors, set appropriate sequence length, batch size and training parameters, divide the training set and validation set, train the LSTM model, ensure the consistency of the physical model and the LSTM model in space and time scales through data interpolation or resampling, and use dimensionality reduction or feature selection methods to simplify calculations.

6. The method for predicting dam deformation trend based on Beidou and GNSS data according to claim 2 is characterized in that: The training parameters are set as follows: the number of epochs is 50 epochs; the batch size is 64; the loss of the validation set is monitored during training, and if there is no improvement within a certain number of epochs, the training is stopped early to prevent overfitting; The model architecture is as follows: Input layer: accepts time series data, and the dimension is determined by the number of features; LSTM layer: Add one or more LSTM layers to capture temporal dependencies; Dropout layer: Add a Dropout layer between LSTM layers to prevent overfitting; Output layer: fully connected layer, outputs the prediction results.

7. The method for predicting dam deformation trend based on Beidou and GNSS data according to claim 2 is characterized in that: The specific implementation to address the impact of temperature factors is to synchronize the temperature data with the GNSS data, match the timestamps through interpolation, perform filtering to remove noise, detect and process outliers in the temperature data, combine the temperature data with other features, perform feature selection, install temperature sensors on the dam top, dam foundation and different depths to ensure data representativeness, set temperature change thresholds to trigger model re-evaluation or update, increase the prediction frequency under extreme temperature conditions, and use online learning algorithms to update the model in real time to adapt to temperature changes.

8. The method for predicting dam deformation trend based on Beidou and GNSS data according to claim 2 is characterized in that: The specific implementation of solving the influence of rainfall factors is to incorporate rainfall as an important feature into the machine learning model, divide it into different levels according to rainfall intensity, analyze its influence on deformation, introduce time lag effect into the model to reflect the delayed influence of water level rise and seepage after rainfall, consider the spatial distribution of rainfall, evaluate its uneven influence on the pressure of different parts of the dam body, analyze the correlation between rainfall and deformation through historical data, optimize model parameters, combine physical models with statistical models, integrate the physical influence of rainfall with data-driven statistical laws, avoid over-reliance on rainfall data in model training, ensure the generalization ability of the model, design special processing modules for extreme rainfall events, enhance the robustness of the model, consider the interaction between rainfall and other environmental factors such as temperature and humidity, and improve prediction accuracy.

9. A computer-readable medium, characterized in that The storage medium stores computer executable instructions, and when the computer executable instructions are loaded and executed by the processor, the steps of the method as claimed in any one of claims 2 to 8 are implemented.

10. A computer program product, characterized in that The computer program product is tangibly stored on a non-transitory computer readable medium and comprises machine executable instructions which, when executed, cause a machine to perform the steps of the method according to any one of claims 2 to 8.