Deformation monitoring method and system

Abstract

The invention provides a deformation monitoring method and system. The deformation monitoring method comprises the following steps: carrying out preliminary calculation on a three-dimensional displacement variable quantity of a monitored object at a monitoring station end; according to the preliminary resolving result of each time point, deformation trend analysis is carried out, and deformation trend information is obtained; judging whether an alarm mechanism needs to be triggered or not according to the preliminary calculation result and the deformation trend information; if an alarm mechanism is triggered, the data center starts high-precision calculation, high-precision calculation is carried out through a variational Bayesian method, a high-precision calculation result is obtained, and satellite integrity monitoring and interference detection are carried out through RAIM; and according to the value of the high-precision calculation result and the deformation trend information, whether an alarm mechanism needs to be triggered continuously is judged. According to the invention, through the real-time preliminary calculation and deformation trend analysis of the monitoring station end, the rapid response to the deformation condition of the monitored object is realized; and in combination with high-precision calculation of a data center, the calculation precision is improved by using a variational Bayesian method, and the real-time performance, accuracy and reliability of deformation monitoring are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of deformation monitoring, and in particular to a deformation monitoring method and system. Background Art

[0002] Deformation monitoring technology is widely used in various fields, such as building structure safety monitoring, geological disaster warning, bridge health monitoring, etc. The core of deformation monitoring technology is to accurately and real-time obtain the displacement change information of the monitored object, so as to timely discover and warn potential safety risks.

[0003] Traditional deformation monitoring methods, such as using professional monitoring equipment or manual inspections, can meet the needs of deformation monitoring to a certain extent, but they have many limitations. In particular, traditional technologies face many challenges in high-precision solution.

[0004] First, traditional deformation monitoring methods often rely on ground monitoring equipment. Although these devices can collect deformation data, they are insufficient in high-precision calculation. The data collected by ground monitoring equipment may be affected by factors such as environmental interference and equipment accuracy limitations, resulting in errors in the calculation results.

[0005] Secondly, traditional deformation monitoring methods often transmit all data to data centers or servers for unified processing, which will lead to large data transmission volume and high processing delay in large-scale monitoring scenarios. Especially when there are a large number of monitoring stations, the processing pressure of the data center will increase significantly, thus affecting the real-time and accuracy of monitoring.

[0006] In addition, traditional deformation monitoring methods usually only focus on a single deformation result and lack in-depth analysis of deformation trends. This results in the system being unable to warn of potential deformation risks in advance and can only respond after deformation occurs, missing the best time to intervene. Summary of the invention

[0007] Based on this, the purpose of the present invention is to provide a deformation monitoring method and system to solve the above-mentioned problems.

[0008] A deformation monitoring method proposed in the present invention comprises: Based on the original observation data set X collected by the reference station within the time interval T and the preliminary observation data set Y collected by the monitoring station, the three-dimensional displacement change of the monitored object is preliminarily solved at the monitoring station to obtain a preliminary solution result ΔY; At the monitoring station, deformation trend analysis is performed based on the preliminary solution results at each time point to obtain deformation trend information; At the monitoring station, determine whether the alarm mechanism needs to be triggered based on the preliminary solution result ΔY and deformation trend information; If the alarm mechanism is triggered, high-precision solution is started through the data center. The original observation data set X and the preliminary observation data set Y are combined to perform high-precision solution through the variational Bayesian method to obtain the high-precision solution result ΔY*. During high-precision solution, the satellite health status is judged through RAIM satellite integrity monitoring, and interference detection is used to determine whether the satellite signal is interfered with, so as to eliminate positioning accuracy abnormalities caused by satellite failures or signal interference. Based on the value of the high-precision solution result ΔY* and the deformation trend information, it is determined whether the alarm mechanism needs to be triggered again.

[0009] Furthermore, the step of performing a preliminary solution on the three-dimensional displacement variation of the monitored object at the monitoring station based on the original observation data set X collected by the reference station within the time interval T and the preliminary observation data set Y collected by the monitoring station to obtain the preliminary solution result ΔY also includes: The original observation data is collected and acquired through the base station receiver at a preset frequency; The original observation data of the reference station is sent to the data center to form the original observation data set X, and then forwarded to the monitoring station receiver through the data center; The original observation data set X is received by the monitoring station receiver, and preliminary observation data is collected and obtained at a preset frequency, and saved in the memory as the original observation data set Y.

[0010] Furthermore, the step of performing a preliminary calculation on the three-dimensional displacement variation of the monitored object at the monitoring station to obtain a preliminary calculation result ΔY includes: The observation equation is established as: , Among them, x i and i are the original observation data and preliminary observation data of the base station and monitoring station at time i, respectively; f is the observation model, is the observation noise, ΔY=[ΔY x , ΔY y , ΔY z ] is the preliminary solution result of the three-dimensional displacement change of the monitored object; Linearize the nonlinear model and obtain the linear observation equation: , Among them, A i is the design matrix, b i is a constant term; The least squares objective function J(ΔY) is constructed as: , Find the preliminary solution ΔY that minimizes J(ΔY): , where A is the stack of design matrices, A T is the transpose of the design matrix A, B is the stack of constant terms, and Y is the stack of preliminary observations.

[0011] Furthermore, the step of performing deformation trend analysis based on the preliminary solution result at each time point to obtain deformation trend information includes: The preliminary solution result ΔY of the three-dimensional displacement change at each time point k k The difference calculation is performed over time to obtain the displacement change rate ΔY' at each time point k , the calculation formula is: , where T is the time interval; Calculate the mean of the displacement change rate and standard deviation ; According to the displacement change rate ΔY' at each time point k , the mean of the displacement change rate and standard deviation , judge the degree of abnormal deformation trend; like , then the deformation trend is judged to be normal; like , it is determined that there is a slight abnormal deformation trend; like , then it is determined that there is a significant abnormal deformation trend; like , it is determined that there is a serious abnormal deformation trend; Among them, θ1, θ2, and θ3 represent the first abnormal deformation trend threshold, the second abnormal deformation trend threshold, and the third abnormal deformation trend threshold, respectively, and θ1<θ2<θ3.

[0012] Furthermore, the step of judging whether to trigger an alarm mechanism according to the preliminary solution result ΔY and the deformation trend information includes: If the preliminary solution results ΔY within the time interval T are all within the first threshold range, and the deformation trend information does not indicate an abnormal deformation trend, the preliminary solution results ΔY are sent to the data center and stored; If any preliminary solution result ΔY within the time interval T exceeds the first threshold range, or the deformation trend information indicates an abnormal deformation trend, the monitoring station will no longer perform the preliminary solution of the three-dimensional displacement, upload the preliminary observation data set Y stored in the memory to the data center, and upload the warning / alarm information according to the threshold, and upload the subsequent preliminary observation data to the data center in real time; After the step of judging whether to trigger the alarm mechanism according to the preliminary solution result ΔY and the deformation trend information, the following steps are further included: If the alarm mechanism is not triggered, the preliminary solution result ΔY is received through the data center, and the preliminary solution result ΔY is stored and the status is recorded as normal. As the time interval accumulates, a preliminary solution result set is formed.

[0013] Furthermore, the step of combining the cached original observation data set X and the received preliminary observation data set Y to perform high-precision solution by using the variational Bayesian method to obtain a high-precision solution result ΔY* includes: A hierarchical probability model is constructed to represent the relationship between the original observation data X, the preliminary observation data Y and the three-dimensional displacement change, so as to decompose the three-dimensional displacement change into a combination of different components; Specify prior distributions for each component separately and perform joint inference; Compute a lower bound on the marginal likelihood based on the variational distribution and the observed data; The parameters of the variational distribution are adjusted through the optimization algorithm to maximize the lower bound of the marginal likelihood. During the optimization process, the variational distribution is continuously updated to gradually approach the true posterior distribution. Repeatedly calculate the lower bound of the marginal likelihood and optimize the variational distribution until the lower bound of the marginal likelihood converges, and extract the parameters of the variational distribution as the optimal high-precision solution result ΔY*.

[0014] Furthermore, the step of combining the original observation data set X and the preliminary observation data set Y and performing high-precision solution by using the variational Bayesian method to obtain a high-precision solution result ΔY* also includes: In the high-precision solution process, the DRL optimization solution strategy is introduced to dynamically adjust the original observation data upload time interval ΔT and the solution time interval T_calc according to the current observation data, solution results, deformation trend and alarm level through the DRL model. At the same time, through the continuous learning and optimization of the DRL model, it can adapt to the dynamic changes of deformation monitoring.

[0015] Furthermore, the step of dynamically adjusting the original observation data uploading time interval ΔT and the calculation time interval T_calc includes: Define the state space, including the current raw observation data, preliminary observation data, high-precision solution results obtained by the variational Bayesian method, deformation trends, and alarm levels; Define the action space, including adjusting the original observation data upload time interval ΔT and the solution time interval T_calc; The reinforcement learning model selects the best action based on the current state, obtains rewards based on the results after executing the action, and updates the model parameters based on the rewards. The results include solution accuracy and real-time performance. Through the continuous interaction with the deformation monitoring system, the reinforcement learning model learns to select the optimal action according to the current state and gradually optimizes its strategy to adapt to the dynamic changes of deformation monitoring. According to the output of the reinforcement learning model, the original observation data upload time interval ΔT and the solution time interval T_calc are dynamically adjusted so that the deformation monitoring system can appropriately extend ΔT and T_calc when the deformation trend is stable to reduce the data transmission and processing burden, or appropriately shorten ΔT and T_calc when the deformation trend is abnormal or the alarm level is high to improve the monitoring accuracy and real-time performance.

[0016] Furthermore, the step of judging whether to continue to trigger the alarm mechanism according to the value of the high-precision solution result ΔY* and the deformation trend information includes: If the high-precision solution result ΔY* is within the first threshold range and the deformation trend information does not indicate an abnormal deformation trend, the high-precision solution result ΔY* is stored and the state is recorded as normal, and the normal state is fed back to the monitoring station end, so that the monitoring station end stops uploading the early warning / alarm signal and continues to obtain preliminary observation data from the monitoring station; If the high-precision solution result ΔY* exceeds the first threshold range but does not reach the second threshold range, or the deformation trend information indicates a slight abnormal deformation trend, a low-risk warning reminder is sent to the security personnel, and high-precision solution is continued to track the deformation situation; If the high-precision solution result ΔY* exceeds the second threshold range but does not reach the third threshold range, or the deformation trend information indicates a significant abnormal deformation trend, a medium-risk alarm reminder is sent to the security personnel, and the high-precision solution continues; If the high-precision solution result ΔY* exceeds the third threshold range, or the deformation trend information indicates a serious abnormal deformation trend, a high-risk alarm reminder is sent to the security personnel, and high-precision solution is continued to obtain more detailed deformation information.

[0017] The present invention further provides a deformation monitoring system for implementing the above deformation monitoring method, the system comprising: Preliminary solution module: used to perform preliminary solution on the three-dimensional displacement change of the monitored object at the monitoring station end based on the original observation data set X collected by the reference station within the time interval T and the preliminary observation data set Y collected by the monitoring station, and obtain the preliminary solution result ΔY; Deformation trend analysis module: used to perform deformation trend analysis based on the preliminary solution results at each time point at the monitoring station to obtain deformation trend information; The first alarm mechanism module is used to determine whether the alarm mechanism needs to be triggered based on the preliminary solution result ΔY and deformation trend information at the monitoring station; High-precision solution module: If the alarm mechanism is triggered, the high-precision solution is started through the data center. The original observation data set X and the preliminary observation data set Y are combined to perform high-precision solution through the variational Bayesian method to obtain the high-precision solution result ΔY*. During the high-precision solution, the satellite health status is judged through RAIM satellite integrity monitoring, and the satellite signal is judged through interference detection to eliminate the positioning accuracy abnormality caused by satellite failure or signal interference. The second alarm mechanism module is used to determine whether the alarm mechanism needs to be triggered further based on the value of the high-precision solution result ΔY* and the deformation trend information.

[0018] In summary, the deformation monitoring method of the present invention first performs a preliminary solution of the three-dimensional displacement change of the collected preliminary observation data on the monitoring station side, quickly obtains the preliminary solution result, and provides a real-time preliminary judgment for deformation monitoring. Then, based on the preliminary solution result at each time point, a deformation trend analysis is performed, which can timely discover the deformation trend of the monitored object and provide a basis for subsequent alarm judgment. At the monitoring station side, it is also determined whether the alarm mechanism needs to be triggered based on the preliminary solution result and deformation trend information, thereby realizing real-time monitoring and early warning of the deformation situation.

[0019] When the alarm mechanism is triggered, high-precision solution is initiated through the data center, and in-depth analysis is performed on the original observation data and preliminary observation data, and high-precision solution results are obtained using the variational Bayesian method. This not only improves the accuracy of deformation monitoring, but also effectively eliminates positioning accuracy anomalies caused by satellite failures or signal interference through RAIM satellite integrity monitoring and interference detection, ensuring the reliability of high-precision solution results.

[0020] Finally, based on the high-precision solution results and deformation trend information, it is determined again whether the alarm mechanism needs to be triggered again, thus achieving continuous monitoring and accurate early warning of the deformation situation.

[0021] The deformation monitoring method proposed in the present invention realizes a rapid response to the deformation of the monitored object through real-time preliminary solution and deformation trend analysis at the monitoring station end; combined with the high-precision solution of the data center, the variational Bayesian method is used to improve the solution accuracy, while eliminating satellite failures or signal interference, which can ensure the real-time, accuracy and reliability of the deformation monitoring results, and provide strong technical support for related engineering safety monitoring, geological disaster early warning and other fields.

[0022] Additional aspects and advantages of the present invention will be given in part in the following description and in part will be obvious from the following description or will be learned through embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The above and / or additional aspects and advantages of the present invention will become apparent and easily understood from the description of the embodiments in conjunction with the following drawings, in which: Figure 1 This is a flow chart of a deformation monitoring method according to Embodiment 1 of the present invention; Figure 2 This is a system block diagram of a deformation monitoring system according to Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0024] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0025] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which the present invention belongs. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0027] Example 1: Please refer to Figure 1 The present invention provides a deformation monitoring method, which includes steps S101 to S105: S101 , based on the original observation data set X collected by the reference station within the time interval T and the preliminary observation data set Y collected by the monitoring station, a preliminary solution is performed on the three-dimensional displacement variation of the monitored object at the monitoring station to obtain a preliminary solution result ΔY.

[0028] It should be noted that the preliminary solution of this embodiment is performed at the monitoring station end. At the monitoring station end, based on the original observation data set X collected by the reference station within the time interval T and the preliminary observation data set Y collected by the monitoring station, the three-dimensional displacement change of the monitored object is preliminarily solved to obtain the preliminary solution result ΔY. The preliminary solution result ΔY reflects the three-dimensional displacement change of the monitoring station within the time interval T, but the accuracy of this result is affected by many factors, such as the accuracy of the observation data, the accuracy of the observation model, the noise level, etc., so the accuracy may not be enough, but it can be used as daily data monitoring in the absence of deformation to avoid a large waste of computing power and resources. And the preliminary solution result is the basis for subsequent deformation trend analysis and alarm mechanism triggering. It provides a preliminary estimate of the deformation state of the monitoring station and provides important input data for subsequent steps.

[0029] Before performing the preliminary solution, the following steps can be performed to achieve timely data transmission and sharing among the reference station, monitoring station and data center, thereby providing a real-time data basis for deformation monitoring and analysis: The base station receiver collects and obtains the original observation data x1, x2, ..., x at a preset frequency. n ,The preset frequency is set according to the requirements of deformation ,monitoring and the performance of the equipment to ensure that sufficient deformation ,information can be captured. The collected raw observation data can include ,various types such as pseudo-range and carrier phase.

[0030] The base station will send the original observation data to the data center, forming the original observation data set X in the data center. At the same time, the data center forwards the original observation data set X to the monitoring station receiver, ensuring the timely transmission and sharing of data.

[0031] The monitoring station receiver receives the original observation data set X and collects preliminary observation data y1, y2, ..., y at the same preset frequency as the base station receiver. n , and saved in the memory as the original observation data set Y, which reflects the deformation state of the monitoring station at the current moment.

[0032] Further optionally, the step of performing a preliminary solution on the three-dimensional displacement variation of the monitored object at the monitoring station to obtain a preliminary solution result ΔY includes: The observation equation is established as: , Among them, x i and i are the original observation data and preliminary observation data of the base station and monitoring station at time i, respectively; f is the observation model, is the observation noise, ΔY=[ΔY x , ΔY y , ΔYz ] is the preliminary solution result of the three-dimensional displacement change of the monitored object; Linearize the nonlinear model and obtain the linear observation equation: , Among them, A i is the design matrix, b i is a constant term; The least squares objective function J(ΔY) is constructed as: , Find the preliminary solution ΔY that minimizes J(ΔY): , where A is the stack of design matrices, A T is the transpose of the design matrix A, B is the stack of constant terms, and Y is the stack of preliminary observations.

[0033] It can be understood that by establishing the observation equation, linearizing the nonlinear model, constructing the least squares objective function and solving it, a preliminary solution to the three-dimensional displacement change of the monitoring station is achieved, so that the monitoring station can use the preliminary observation data of the base station and the monitoring station to accurately estimate the three-dimensional displacement change of the monitored object, which is convenient for daily monitoring when no abnormal deformation occurs, and also provides a basis for subsequent deformation trend analysis and alarm mechanism triggering.

[0034] Among them, ΔY=[ΔY x , ΔY y , ΔY z ] x is the displacement change of the monitoring station in the X-axis direction, ΔY y is the displacement change of the monitoring station in the Y-axis direction, ΔY z is the displacement change of the monitoring station in the Z-axis direction.

[0035] S102: At the monitoring station, deformation trend analysis is performed based on the preliminary solution results at each time point to obtain deformation trend information.

[0036] It should be noted that the deformation trend can be identified by analyzing the change in the three-dimensional displacement of the monitored object over time, and then it can be determined whether the monitored object has an abnormal deformation trend.

[0037] Further optionally, the step of performing deformation trend analysis based on the preliminary solution result at each time point to obtain deformation trend information includes: The preliminary solution result ΔY of the three-dimensional displacement change at each time point k k The difference calculation is performed over time to obtain the displacement change rate ΔY' at each time pointk , the calculation formula is: , where T is the time interval; Calculate the mean of the displacement change rate and standard deviation ; According to the displacement change rate ΔY' at each time point k , the mean of the displacement change rate and standard deviation , judge the degree of abnormal deformation trend; like , then the deformation trend is judged to be normal; like , it is determined that there is a slight abnormal deformation trend; like , then it is determined that there is a significant abnormal deformation trend; like , it is determined that there is a serious abnormal deformation trend; Among them, θ1, θ2, and θ3 represent the first abnormal deformation trend threshold, the second abnormal deformation trend threshold, and the third abnormal deformation trend threshold, respectively, and θ1<θ2<θ3.

[0038] It is understandable that by performing differential calculation on the three-dimensional displacement change at each time point, the displacement change rate is obtained, thereby identifying the deformation trend of the monitored object over time, which helps to understand the deformation law of the monitored object in space and time. By calculating the difference between the displacement change rate and the mean at each time point and comparing it with the standard deviation, it is determined whether the deformation trend is abnormal and the degree of abnormality based on the size of the difference. Through deformation trend analysis and abnormal deformation judgment, a scientific basis is provided for subsequent decisions such as whether to trigger an alarm mechanism and whether high-precision solution is required, which helps to make timely and accurate judgments during deformation monitoring, and ultimately ensure the safety and stability of the monitored object.

[0039] The mean value of the displacement change rate μ ΔY' and standard deviation σ ΔY' The calculation formula is as follows: , , Where n is the number of time points in the time interval T.

[0040] S103: At the monitoring station, it is determined whether an alarm mechanism needs to be triggered based on the preliminary solution result ΔY and the deformation trend information.

[0041] It should be noted that based on the preliminary solution result ΔY and deformation trend information, the monitoring station determines whether it is necessary to trigger the alarm mechanism. If no abnormal deformation or abnormal deformation trend is found, the preliminary solution result can be used as the final solution result of the three-dimensional displacement change of the monitoring station, and there is no need to trigger the alarm mechanism, thereby avoiding the subsequent high-precision solution process. This can not only ensure the efficiency of the solution under normal circumstances, but also reduce the dependence on the data center.

[0042] If abnormal deformation or abnormal deformation trend is found, the monitoring station will trigger an alarm mechanism, which will notify the data center to start high-precision solution. This process realizes the joint analysis and solution of data between the monitoring station and the data center. Real-time preliminary solution, deformation trend analysis and early warning judgment are carried out at the monitoring station, while the data center is responsible for high-precision solution to track the deformation in real time after receiving the alarm to ensure the accuracy of the solution results. This mechanism improves the sensitivity and accuracy of deformation monitoring, while optimizing resource utilization, ensuring that high-precision solution of the data center is only performed when necessary, thereby improving the overall solution and deformation monitoring efficiency.

[0043] Further optionally, the step of judging whether to trigger an alarm mechanism according to the preliminary solution result ΔY and the deformation trend information includes: If the preliminary solution results ΔY within the time interval T are all within the first threshold range, and the deformation trend information does not indicate an abnormal deformation trend, the preliminary solution results ΔY are sent to the data center and stored; If any preliminary solution result ΔY within the time interval T exceeds the first threshold range, or the deformation trend information indicates an abnormal deformation trend, the monitoring station will no longer perform the preliminary solution of the three-dimensional displacement, upload the preliminary observation data set Y stored in the memory to the data center, and upload the warning / alarm information according to the threshold, and upload the subsequent preliminary observation data to the data center in real time.

[0044] It is understandable that if the preliminary solution results ΔY within the time interval T are all within the first threshold range, and the deformation trend information does not indicate an abnormal deformation trend, that is, when the deformation of the monitored object is normal or the deformation trend is normal, the preliminary solution results ΔY are sent to the data center and stored as a record of the three-dimensional displacement change of the monitored object in the time period. In this case, there is no need to trigger the alarm mechanism, thereby avoiding the subsequent high-precision solution process, which can not only ensure the efficiency of the solution under normal circumstances, but also reduce the dependence on the data center.

[0045] If any preliminary solution result ΔY within the time interval T exceeds the first threshold range, or the deformation trend information indicates an abnormal deformation trend, that is, the monitored object has an abnormal deformation or an abnormal deformation trend (that is, there is a potential deformation risk), the alarm mechanism is triggered in time, the monitoring station stops executing the preliminary solution of the three-dimensional displacement, and uploads the preliminary observation data set Y stored in the memory to the data center to provide data support for subsequent high-precision solutions, and uploads early warning / alarm information to the data center according to the threshold to remind the data center that the monitored object has an abnormal deformation, and the subsequent preliminary observation data Y of the monitoring station is uploaded to the data center. a+1 , Y a+2 , …, Y a+n It is also uploaded to the data center in real time so that the data center can continue to perform high-precision solutions to track deformation in real time and ensure the sensitivity, accuracy and continuity of deformation monitoring.

[0046] Further optionally, after the step of judging whether to trigger an alarm mechanism according to the preliminary solution result ΔY and the deformation trend information, the step further includes: If the alarm mechanism is not triggered, the preliminary solution result ΔY is received through the data center, and the preliminary solution result ΔY is stored and the status is recorded as normal. As the time interval accumulates, a preliminary solution result set is formed.

[0047] Understandably, at the data center end, if the data center receives a preliminary solution result ΔY, it means that the alarm mechanism is not triggered, that is, there is no abnormal deformation or abnormal deformation trend. The data center will store the preliminary solution result ΔY and record the status as normal. With the accumulation of time intervals, a preliminary solution result set is formed, such as ΔY T1 , ΔY T2 ,…,ΔY Tn The preliminary solution results at this time can be used as daily monitoring data for the three-dimensional displacement change of the monitored object, without the need for high-precision solution. This helps optimize the monitoring process, improve monitoring efficiency, and reduce unnecessary resource and computing power consumption.

[0048] S104: If the alarm mechanism is triggered, high-precision solution is started through the data center. The original observation data set X and the preliminary observation data set Y are combined to perform high-precision solution through the variational Bayesian method to obtain a high-precision solution result ΔY*. During high-precision solution, the satellite health status is judged through RAIM satellite integrity monitoring, and interference detection is used to judge whether the satellite signal is interfered with, so as to eliminate positioning accuracy abnormalities caused by satellite failures or signal interference.

[0049] It should be noted that the preliminary solution result ΔY and the deformation trend information jointly determine whether the alarm mechanism needs to be triggered. If the preliminary solution result shows that the deformation may exceed the first threshold range, or the deformation trend information indicates that there is an abnormal deformation trend, the alarm mechanism is triggered. Once the alarm mechanism is triggered, the early warning / alarm information sent by the monitoring station will remind the data center to start the high-precision solution algorithm, then the cached original observation data set X and the received preliminary observation data set are combined, and the variational Bayesian method is used to perform more refined data processing and higher-precision solution to obtain a high-precision solution result ΔY*, that is, the precise three-dimensional displacement change is, and deformation monitoring provides more reliable results. The variational Bayesian method of this embodiment can approximate complex posterior distributions by finding the optimal approximate probability distribution, which not only ensures computational efficiency, but also maximizes the accuracy of the solution results. Because this method can handle the uncertainty of parameters and latent variables, and by modeling uncertainty, it provides a more robust and reliable solution result.

[0050] Before high-precision solution, the cached original observation data set X and the received preliminary observation data set Y can be subjected to a Kalman filter algorithm to reduce data noise.

[0051] At the same time, during the high-precision solution process, RAIM technology is used to determine the health status of the satellite. RAIM is a technology used to detect anomalies in satellite navigation systems. It can identify and eliminate signals from faulty satellites, thereby improving the accuracy and reliability of positioning. Interference detection is also performed to determine whether the satellite signal is subject to external interference. When solving with high precision, combining RAIM and interference detection technology can eliminate positioning accuracy anomalies caused by satellite failures or signal interference, thereby improving the accuracy and reliability of data processing.

[0052] Further optionally, the step of combining the cached original observation data set X and the received preliminary observation data set Y to perform high-precision solution by using a variational Bayesian method to obtain a high-precision solution result ΔY* includes: A hierarchical probability model is constructed to represent the relationship between the original observation data X, the preliminary observation data Y and the three-dimensional displacement change, so as to decompose the three-dimensional displacement change into a combination of different components; Specify prior distributions for each component separately and perform joint inference; Compute a lower bound on the marginal likelihood based on the variational distribution and the observed data; The parameters of the variational distribution are adjusted through the optimization algorithm to maximize the lower bound of the marginal likelihood. During the optimization process, the variational distribution is continuously updated to gradually approach the true posterior distribution. Repeatedly calculate the lower bound of the marginal likelihood and optimize the variational distribution until the lower bound of the marginal likelihood converges, and extract the parameters of the variational distribution as the optimal high-precision solution result ΔY*.

[0053] It can be understood that by constructing a hierarchical probability model to represent the relationship between the original observation data X, the preliminary observation data Y and the three-dimensional displacement change, the complex three-dimensional displacement change can be decomposed into a combination of different components (such as trend terms, periodic terms, and noise terms). This decomposition helps to more clearly understand the source and characteristics of the displacement change.

[0054] By specifying a prior distribution for each component and performing joint inference, it is possible to simultaneously consider the impact of all components on the observed data, that is, to comprehensively consider all factors that may affect the observed data, thereby improving the accuracy and reliability of the solution.

[0055] According to the variational distribution and the observed data, the lower bound of the marginal likelihood is calculated as the objective function of the optimization. The parameters of the variational distribution are adjusted through optimization algorithms (such as gradient descent, coordinate ascent, etc.) to maximize the lower bound of the marginal likelihood. This process continuously optimizes the variational distribution, making it gradually close to the true posterior distribution, thereby improving the accuracy of the solution. Repeat the calculation of the lower bound of the marginal likelihood and the optimization of the variational distribution until the lower bound of the marginal likelihood converges, and then extract the parameters of the variational distribution as the optimal high-precision solution result ΔY*, thereby achieving a high-precision and high-efficiency solution process, ensuring the accuracy and reliability of the solution result.

[0056] Further optionally, the step of combining the original observation data set X and the preliminary observation data set Y to perform high-precision solution by using a variational Bayesian method to obtain a high-precision solution result ΔY* also includes: In the high-precision solution process, the DRL optimization solution strategy is introduced to dynamically adjust the original observation data upload time interval ΔT and the solution time interval T_calc according to the current observation data, solution results, deformation trend and alarm level through the DRL model. At the same time, through the continuous learning and optimization of the DRL model, it can adapt to the dynamic changes of deformation monitoring.

[0057] It is understandable that once the alarm mechanism is triggered, the data center will start the high-precision solution process, perform high-precision solution through the variational Bayesian method, and obtain the high-precision solution result ΔY*, making full use of the existing observation data, improving the accuracy of the solution, and providing more reliable results for deformation monitoring.

[0058] At the same time, in the high-precision solution process, the DRL (deep reinforcement learning) optimization solution strategy is introduced. The DRL model can dynamically adjust the original observation data upload time interval ΔT and the solution time interval T_calc according to the current observation data, the high-precision solution results solved by the variational Bayesian method, the deformation trend and the alarm level. This dynamic adjustment mechanism enables the system to flexibly adjust the time interval for data transmission and processing according to the real-time needs of deformation monitoring, thereby ensuring the accuracy of the solution while improving the real-time performance of the system.

[0059] Further optionally, the step of dynamically adjusting the original observation data uploading time interval ΔT and the calculation time interval T_calc includes: Define the state space, including the current raw observation data, preliminary observation data, high-precision solution results obtained by the variational Bayesian method, deformation trends, and alarm levels; Define the action space, including adjusting the original observation data upload time interval ΔT and the solution time interval T_calc; The reinforcement learning model selects the best action based on the current state, obtains rewards based on the results after executing the action, and updates the model parameters based on the rewards. The results include solution accuracy and real-time performance. Through the continuous interaction with the deformation monitoring system, the reinforcement learning model learns to select the optimal action according to the current state and gradually optimizes its strategy to adapt to the dynamic changes of deformation monitoring. According to the output of the reinforcement learning model, the original observation data upload time interval ΔT and the solution time interval T_calc are dynamically adjusted so that the deformation monitoring system can appropriately extend ΔT and T_calc when the deformation trend is stable to reduce the data transmission and processing burden, or appropriately shorten ΔT and T_calc when the deformation trend is abnormal or the alarm level is high to improve the monitoring accuracy and real-time performance.

[0060] Understandably, in the high-precision solution process, the DRL optimization solution strategy is introduced. The DRL model dynamically adjusts the original observation data upload time interval ΔT and the solution time interval T_calc according to the current observation data, the high-precision solution results solved by the variational Bayesian method, the deformation trend and the alarm level. This dynamic adjustment mechanism enables the system to flexibly adjust the time interval for data transmission and processing according to the real-time needs of deformation monitoring, thereby ensuring the accuracy of the solution while improving the real-time performance of the system.

[0061] Through the dynamic adjustment of the DRL model, the system can appropriately extend the time interval for uploading the original observation data and the time interval for solving the problem when the deformation trend is stable, reduce the burden of data transmission and processing, and improve the efficiency of the system. And when the deformation trend is abnormal or the alarm level is high, the system can shorten the time interval, obtain and process the observation data in time, improve the accuracy and real-time performance of the solution, and ensure the timeliness and reliability of deformation monitoring.

[0062] Since the DRL model has the ability to continuously learn and optimize, through continuous interaction with the deformation monitoring system, the DRL model can gradually optimize its decision-making strategy to adapt to the dynamic changes of deformation monitoring. This learning and optimization mechanism enables the system to better cope with various complex and changing deformation monitoring scenarios, improve the robustness and adaptability of the system, and ensure that the system can work stably and accurately in different environments.

[0063] S105, judging whether it is necessary to continue to trigger the alarm mechanism according to the value of the high-precision solution result ΔY* and the deformation trend information.

[0064] It should be noted that after high-precision solution is performed in the data center and the high-precision solution result is obtained, a more accurate deformation trend analysis is performed based on the high-precision solution result ΔY* to obtain deformation trend information, and based on the value of the high-precision solution result ΔY* and the deformation trend information, it is determined whether the alarm mechanism needs to be triggered. Specifically, according to different deformation degrees or different degrees of abnormal deformation trends, early warnings or alarm reminders can be automatically sent to ensure that security personnel can take timely measures. Regardless of the degree of deformation, the system continues to perform high-precision solution to obtain more accurate deformation information and trends.

[0065] Further optionally, the step of judging whether it is necessary to continue to trigger the alarm mechanism according to the value of the high-precision solution result ΔY* and the deformation trend information includes: If the high-precision solution result ΔY* is within the first threshold range and the deformation trend information does not indicate an abnormal deformation trend, the high-precision solution result ΔY* is stored and the state is recorded as normal, and the normal state is fed back to the monitoring station end, so that the monitoring station end stops uploading the early warning / alarm signal and continues to obtain preliminary observation data from the monitoring station; If the high-precision solution result ΔY* exceeds the first threshold range but does not reach the second threshold range, or the deformation trend information indicates a slight abnormal deformation trend, a low-risk warning reminder is sent to the security personnel, and high-precision solution is continued to track the deformation situation; If the high-precision solution result ΔY* exceeds the second threshold range but does not reach the third threshold range, or the deformation trend information indicates a significant abnormal deformation trend, a medium-risk alarm reminder is sent to the security personnel, and the high-precision solution continues; If the high-precision solution result ΔY* exceeds the third threshold range, or the deformation trend information indicates a serious abnormal deformation trend, a high-risk alarm reminder is sent to the security personnel, and high-precision solution is continued to obtain more detailed deformation information.

[0066] The following is a specific implementation of a deformation monitoring method of the present invention, specifically as follows: The base station receiver collects and obtains the original observation data x1, x2, ..., x at a preset frequency. n ; The original observation data of the reference station is sent to the data center to form the original observation data set X = {x1, x2, ..., x n} and forwarded to the monitoring station receiver through the data center; The original observation data set X is received by the monitoring station receiver, and preliminary observation data y1, y2, ..., y are collected at a preset frequency. n , and saved in memory as the original observation data set Y={y1, y2, …, y n}; Based on the original observation data set X collected by the reference station within the time interval T and the preliminary observation data set Y collected by the monitoring station, the three-dimensional displacement change of the monitored object is preliminarily solved at the monitoring station to obtain a preliminary solution result ΔY; Based on the preliminary solution results at each time point, deformation trend analysis is performed to obtain deformation trend information; Combine the deformation trend information and the preset threshold range to determine whether to trigger the alarm mechanism: If the preliminary solution results ΔY within the time interval T are all within the first threshold range, and the deformation trend information does not indicate an abnormal deformation trend, the preliminary solution results ΔY are sent to the data center and stored; If any preliminary solution result ΔY within the time interval T exceeds the first threshold range, or the deformation trend information indicates an abnormal deformation trend, the monitoring station will no longer perform the preliminary solution of the three-dimensional displacement, upload the preliminary observation data set Y stored in the memory to the data center, and upload the warning / alarm information according to the threshold, and at the same time upload the subsequent preliminary observation data Y a+1 , Y a+2 , …, Y a+n It is also uploaded to the data center in real time; If the data center receives a preliminary solution result ΔY, it stores the preliminary solution result ΔY and records the status as normal. As the time interval accumulates, a preliminary solution result set is formed, including ΔY. T1 , ΔY T2 ,…,ΔY Tn ; If the data center receives early warning / alarm information and a preliminary observation data set Y, a high-precision solution is initiated through the data center, and a variational Bayesian algorithm is used to combine the original observation data set X and the preliminary observation data set Y for high-precision solution to obtain a high-precision solution result ΔY*, which is the precise three-dimensional displacement change; In the high-precision solution process, deep reinforcement learning (DRL) is introduced to optimize the solution strategy. The DRL model dynamically adjusts the original observation data upload time interval ΔT and the solution time interval T_calc according to the current observation data, solution results, deformation trend and alarm level to improve the accuracy and real-time performance of the solution. Through continuous learning and optimization, it adapts to the dynamic changes of deformation monitoring and improves the robustness and adaptability of the system. In addition, when solving high-precision problems, it is necessary to determine the health status of the satellite through RAIM satellite integrity monitoring, and determine whether the satellite signal is interfered with through interference detection, so as to eliminate positioning accuracy abnormalities caused by satellite failures or signal interference; After judging and eliminating interference and other situations, a more accurate deformation trend analysis is performed based on the high-precision solution result ΔY* to obtain deformation trend information, and based on the value of the high-precision solution result ΔY* and the deformation trend information, it is determined whether the alarm mechanism needs to be triggered again; If the high-precision solution result ΔY* is within the first threshold range and the deformation trend information does not indicate an abnormal deformation trend, the high-precision solution result ΔY* is stored in the data center and the state is recorded as normal. At the same time, the normal state is fed back to the monitoring station end, so that the monitoring station end stops uploading the early warning / alarm signal and continues to obtain preliminary observation data from the monitoring station; If the high-precision solution result ΔY* exceeds the first threshold range but does not reach the second threshold range, or the deformation trend information indicates a slight abnormal deformation trend, a low-risk warning reminder is sent to the security personnel, and high-precision solution is continued to track the deformation situation; If the high-precision solution result ΔY* exceeds the second threshold range but does not reach the third threshold range, or the deformation trend information indicates a significant abnormal deformation trend, a medium-risk alarm reminder is sent to the security personnel, and the high-precision solution continues; If the high-precision solution result ΔY* exceeds the third threshold range, or the deformation trend information indicates a serious abnormal deformation trend, a high-risk alarm reminder is sent to the security personnel, and high-precision solution is continued to obtain more detailed deformation information.

[0067] In summary, the deformation monitoring method of the present invention first performs a preliminary solution of the three-dimensional displacement change of the collected preliminary observation data on the monitoring station side, quickly obtains the preliminary solution result, and provides a real-time preliminary judgment for deformation monitoring. Then, based on the preliminary solution result at each time point, a deformation trend analysis is performed, which can timely discover the deformation trend of the monitored object and provide a basis for subsequent alarm judgment. At the monitoring station side, it is also determined whether the alarm mechanism needs to be triggered based on the preliminary solution result and deformation trend information, thereby realizing real-time monitoring and early warning of the deformation situation.

[0068] When the alarm mechanism is triggered, high-precision solution is initiated through the data center, and in-depth analysis is performed on the original observation data and preliminary observation data, and high-precision solution results are obtained using the variational Bayesian method. This not only improves the accuracy of deformation monitoring, but also effectively eliminates positioning accuracy anomalies caused by satellite failures or signal interference through RAIM satellite integrity monitoring and interference detection, ensuring the reliability of high-precision solution results.

[0069] Finally, based on the high-precision solution results and deformation trend information, it is determined again whether the alarm mechanism needs to be triggered again, thus achieving continuous monitoring and accurate early warning of the deformation situation.

[0070] The deformation monitoring method proposed in the present invention realizes a rapid response to the deformation of the monitored object through real-time preliminary solution and deformation trend analysis at the monitoring station end; combined with the high-precision solution of the data center, the variational Bayesian method is used to improve the solution accuracy, while eliminating satellite failures or signal interference, which can ensure the real-time, accuracy and reliability of the deformation monitoring results, and provide strong technical support for related engineering safety monitoring, geological disaster early warning and other fields.

[0071] Example 2: Please refer to Figure 2 The present invention provides a deformation monitoring system, which comprises: Preliminary solution module: used to perform preliminary solution on the three-dimensional displacement change of the monitored object at the monitoring station end based on the original observation data set X collected by the reference station within the time interval T and the preliminary observation data set Y collected by the monitoring station, and obtain the preliminary solution result ΔY; Deformation trend analysis module: used to perform deformation trend analysis based on the preliminary solution results at each time point to obtain deformation trend information; The first alarm mechanism module is used to determine whether the alarm mechanism needs to be triggered based on the preliminary solution result ΔY and deformation trend information; High-precision solution module: If the alarm mechanism is triggered, the high-precision solution is started through the data center. The original observation data set X and the preliminary observation data set Y are combined to perform high-precision solution through the variational Bayesian method to obtain the high-precision solution result ΔY*. During the high-precision solution, the satellite health status is judged through RAIM satellite integrity monitoring, and the satellite signal is judged through interference detection to eliminate the positioning accuracy abnormality caused by satellite failure or signal interference. The second alarm mechanism module is used to determine whether the alarm mechanism needs to be triggered further based on the value of the high-precision solution result ΔY* and the deformation trend information.

[0072] Further optionally, the preliminary solution module is also used for: The original observation data is collected and acquired through the base station receiver at a preset frequency; The original observation data of the reference station is sent to the data center to form the original observation data set X, and then forwarded to the monitoring station receiver through the data center; The original observation data set X is received by the monitoring station receiver, and preliminary observation data is collected and obtained at a preset frequency, and saved in the memory as the original observation data set Y.

[0073] Further optionally, the preliminary solution module is also used for: The observation equation is established as: , Among them, x i and i are the original observation data and preliminary observation data of the base station and monitoring station at time i, respectively; f is the observation model, is the observation noise, ΔY=[ΔY x , ΔY y , ΔY z ] is the preliminary solution result of the three-dimensional displacement change of the monitored object; Linearize the nonlinear model and obtain the linear observation equation: , Among them, A i is the design matrix, b i is a constant term; The least squares objective function J(ΔY) is constructed as: , Find the preliminary solution ΔY that minimizes J(ΔY): , where A is the stack of design matrices, A Tis the transpose of the design matrix A, B is the stack of constant terms, and Y is the stack of preliminary observations.

[0074] Further optionally, the deformation trend analysis module is also used for: The preliminary solution result ΔY of the three-dimensional displacement change at each time point k k The difference calculation is performed over time to obtain the displacement change rate ΔY' at each time point k , the calculation formula is: , where T is the time interval; Calculate the mean of the displacement change rate and standard deviation ; According to the displacement change rate ΔY' at each time point k , the mean of the displacement change rate and standard deviation , judge the degree of abnormal deformation trend; like , then the deformation trend is judged to be normal; like , it is determined that there is a slight abnormal deformation trend; like , then it is determined that there is a significant abnormal deformation trend; like , it is determined that there is a serious abnormal deformation trend; Among them, θ1, θ2, and θ3 represent the first abnormal deformation trend threshold, the second abnormal deformation trend threshold, and the third abnormal deformation trend threshold, respectively, and θ1<θ2<θ3.

[0075] Further optionally, the first alarm mechanism module is also used for: If the preliminary solution results ΔY within the time interval T are all within the first threshold range, and the deformation trend information does not indicate an abnormal deformation trend, the preliminary solution results ΔY are sent to the data center and stored; If any preliminary solution result ΔY within the time interval T exceeds the first threshold range, or the deformation trend information indicates an abnormal deformation trend, the monitoring station will no longer perform the preliminary solution of the three-dimensional displacement, upload the preliminary observation data set Y stored in the memory to the data center, and upload the warning / alarm information according to the threshold, and upload the subsequent preliminary observation data to the data center in real time.

[0076] Further optionally, the first alarm mechanism module is also used for: If the alarm mechanism is not triggered, the preliminary solution result ΔY is received through the data center, and the preliminary solution result ΔY is stored and the status is recorded as normal. As the time interval accumulates, a preliminary solution result set is formed.

[0077] Further optionally, the high-precision solution module is also used for: A hierarchical probability model is constructed to represent the relationship between the original observation data X, the preliminary observation data Y and the three-dimensional displacement change, so as to decompose the three-dimensional displacement change into a combination of different components; Specify prior distributions for each component and perform joint inference; Compute a lower bound on the marginal likelihood based on the variational distribution and the observed data; The parameters of the variational distribution are adjusted through the optimization algorithm to maximize the lower bound of the marginal likelihood. During the optimization process, the variational distribution is continuously updated to gradually approach the true posterior distribution. Repeatedly calculate the lower bound of the marginal likelihood and optimize the variational distribution until the lower bound of the marginal likelihood converges, and extract the parameters of the variational distribution as the optimal high-precision solution result ΔY*.

[0078] Further optionally, the high-precision solution module is also used for: In the high-precision solution process, the DRL optimization solution strategy is introduced to dynamically adjust the original observation data upload time interval ΔT and the solution time interval T_calc according to the current observation data, solution results, deformation trend and alarm level through the DRL model. At the same time, through the continuous learning and optimization of the DRL model, it can adapt to the dynamic changes of deformation monitoring.

[0079] Further optionally, the high-precision solution module is also used for: Define the state space, including the current raw observation data, preliminary observation data, high-precision solution results obtained by the variational Bayesian method, deformation trends, and alarm levels; Define the action space, including adjusting the original observation data upload time interval ΔT and the solution time interval T_calc; The reinforcement learning model selects the best action based on the current state, obtains rewards based on the results after executing the action, and updates the model parameters based on the rewards. The results include solution accuracy and real-time performance. Through the continuous interaction with the deformation monitoring system, the reinforcement learning model learns to select the optimal action according to the current state and gradually optimizes its strategy to adapt to the dynamic changes of deformation monitoring. According to the output of the reinforcement learning model, the original observation data upload time interval ΔT and the solution time interval T_calc are dynamically adjusted so that the deformation monitoring system can appropriately extend ΔT and T_calc when the deformation trend is stable to reduce the data transmission and processing burden, or appropriately shorten ΔT and T_calc when the deformation trend is abnormal or the alarm level is high to improve the monitoring accuracy and real-time performance.

[0080] Further optionally, the second alarm mechanism module is further used for: If the high-precision solution result ΔY* is within the first threshold range and the deformation trend information does not indicate an abnormal deformation trend, the high-precision solution result ΔY* is stored and the state is recorded as normal, and the normal state is fed back to the monitoring station end, so that the monitoring station end stops uploading the early warning / alarm signal and continues to obtain preliminary observation data from the monitoring station; If the high-precision solution result ΔY* exceeds the first threshold range but does not reach the second threshold range, or the deformation trend information indicates a slight abnormal deformation trend, a low-risk warning reminder is sent to the security personnel, and high-precision solution is continued to track the deformation situation; If the high-precision solution result ΔY* exceeds the second threshold range but does not reach the third threshold range, or the deformation trend information indicates a significant abnormal deformation trend, a medium-risk alarm reminder is sent to the security personnel, and the high-precision solution continues; If the high-precision solution result ΔY* exceeds the third threshold range, or the deformation trend information indicates a serious abnormal deformation trend, a high-risk alarm reminder is sent to the security personnel, and high-precision solution is continued to obtain more detailed deformation information.

[0081] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A deformation monitoring method, characterized in that: The method comprises: Based on the original observation data set X collected by the reference station within the time interval T and the preliminary observation data set Y collected by the monitoring station, the three-dimensional displacement change of the monitored object is preliminarily solved at the monitoring station to obtain a preliminary solution result ΔY; At the monitoring station, deformation trend analysis is performed based on the preliminary solution results at each time point to obtain deformation trend information; At the monitoring station, determine whether the alarm mechanism needs to be triggered based on the preliminary solution result ΔY and deformation trend information; If the alarm mechanism is triggered, high-precision solution is started through the data center. The original observation data set X and the preliminary observation data set Y are combined to perform high-precision solution through the variational Bayesian method to obtain the high-precision solution result ΔY*. During high-precision solution, the satellite health status is judged through RAIM satellite integrity monitoring, and interference detection is used to determine whether the satellite signal is interfered with, so as to eliminate positioning accuracy abnormalities caused by satellite failures or signal interference. Based on the value of the high-precision solution result ΔY* and the deformation trend information, it is determined whether the alarm mechanism needs to be triggered again.

2. The deformation monitoring method according to claim 1, characterized in that: The step of performing a preliminary calculation of the three-dimensional displacement variation of the monitored object at the monitoring station based on the original observation data set X collected by the reference station within the time interval T and the preliminary observation data set Y collected by the monitoring station to obtain the preliminary calculation result ΔY also includes: The original observation data is collected and acquired through the base station receiver at a preset frequency; The original observation data of the reference station is sent to the data center to form the original observation data set X, and then forwarded to the monitoring station receiver through the data center; The original observation data set X is received by the monitoring station receiver, and preliminary observation data is collected and obtained at a preset frequency, and saved in the memory as the original observation data set Y.

3. The deformation monitoring method according to claim 1, characterized in that: The step of performing a preliminary calculation on the three-dimensional displacement variation of the monitored object at the monitoring station to obtain a preliminary calculation result ΔY comprises: The observation equation is established as: ， Among them, x i and i are the original observation data and preliminary observation data of the base station and monitoring station at time i, respectively; f is the observation model, is the observation noise, ΔY=[ΔY x , ΔY y , ΔY z ] is the preliminary solution result of the three-dimensional displacement change of the monitored object; Linearize the nonlinear model and obtain the linear observation equation: ， Among them, A i is the design matrix, b i is a constant term; The least squares objective function J(ΔY) is constructed as: ， Find the preliminary solution ΔY that minimizes J(ΔY): ， where A is the stack of design matrices, A T is the transpose of the design matrix A, B is the stack of constant terms, and Y is the stack of preliminary observations.

4. The deformation monitoring method according to claim 1, characterized in that: The step of performing deformation trend analysis based on the preliminary solution results at each time point to obtain deformation trend information includes: The preliminary solution result ΔY of the three-dimensional displacement change at each time point k k The difference calculation is performed over time to obtain the displacement change rate ΔY' at each time point k , the calculation formula is: , where T is the time interval; Calculate the mean of the displacement change rate and standard deviation ; According to the displacement change rate ΔY' at each time point k , the mean of the displacement change rate and standard deviation , judge the degree of abnormal deformation trend; like , then the deformation trend is judged to be normal; like , it is determined that there is a slight abnormal deformation trend; like , then it is determined that there is a significant abnormal deformation trend; like , it is determined that there is a serious abnormal deformation trend; Among them, θ1, θ2, and θ3 represent the first abnormal deformation trend threshold, the second abnormal deformation trend threshold, and the third abnormal deformation trend threshold, respectively, and θ1<θ2<θ3.

5. The deformation monitoring method according to claim 1, characterized in that: The step of judging whether to trigger the alarm mechanism according to the preliminary solution result ΔY and the deformation trend information includes: If the preliminary solution results ΔY within the time interval T are all within the first threshold range, and the deformation trend information does not indicate an abnormal deformation trend, the preliminary solution results ΔY are sent to the data center and stored; If any preliminary solution result ΔY within the time interval T exceeds the first threshold range, or the deformation trend information indicates an abnormal deformation trend, the monitoring station will no longer perform the preliminary solution of the three-dimensional displacement, upload the preliminary observation data set Y stored in the memory to the data center, and upload the warning / alarm information according to the threshold, and upload the subsequent preliminary observation data to the data center in real time; After the step of judging whether to trigger the alarm mechanism according to the preliminary solution result ΔY and the deformation trend information, the following steps are further included: If the alarm mechanism is not triggered, the preliminary solution result ΔY is received through the data center, and the preliminary solution result ΔY is stored and the status is recorded as normal. As the time interval accumulates, a preliminary solution result set is formed.

6. The deformation monitoring method according to claim 1, characterized in that: The step of combining the original observation data set X and the preliminary observation data set Y and performing high-precision calculation by using the variational Bayesian method to obtain a high-precision calculation result ΔY* comprises: A hierarchical probability model is constructed to represent the relationship between the original observation data X, the preliminary observation data Y and the three-dimensional displacement change, so as to decompose the three-dimensional displacement change into a combination of different components; Specify prior distributions for each component and perform joint inference; Compute a lower bound on the marginal likelihood based on the variational distribution and the observed data; The parameters of the variational distribution are adjusted through the optimization algorithm to maximize the lower bound of the marginal likelihood. During the optimization process, the variational distribution is continuously updated to gradually approach the true posterior distribution. Repeatedly calculate the lower bound of the marginal likelihood and optimize the variational distribution until the lower bound of the marginal likelihood converges, and extract the parameters of the variational distribution as the optimal high-precision solution result ΔY*.

7. The deformation monitoring method according to claim 1, characterized in that: The step of combining the original observation data set X and the preliminary observation data set Y and performing high-precision calculation by using the variational Bayesian method to obtain a high-precision calculation result ΔY* also includes: In the high-precision solution process, the DRL optimization solution strategy is introduced to dynamically adjust the original observation data upload time interval ΔT and the solution time interval T_calc according to the current observation data, solution results, deformation trend and alarm level through the DRL model. At the same time, through the continuous learning and optimization of the DRL model, it can adapt to the dynamic changes of deformation monitoring.

8. The deformation monitoring method according to claim 7, characterized in that: The step of dynamically adjusting the original observation data uploading time interval ΔT and the calculation time interval T_calc comprises: Define the state space, including the current raw observation data, preliminary observation data, high-precision solution results obtained by the variational Bayesian method, deformation trends, and alarm levels; Define the action space, including adjusting the original observation data upload time interval ΔT and the solution time interval T_calc; The reinforcement learning model selects the best action based on the current state, obtains rewards based on the results after executing the action, and updates the model parameters based on the rewards. The results include solution accuracy and real-time performance. Through the continuous interaction with the deformation monitoring system, the reinforcement learning model learns to select the optimal action according to the current state and gradually optimizes its strategy to adapt to the dynamic changes of deformation monitoring. According to the output of the reinforcement learning model, the original observation data upload time interval ΔT and the solution time interval T_calc are dynamically adjusted so that the deformation monitoring system can appropriately extend ΔT and T_calc when the deformation trend is stable to reduce the data transmission and processing burden, or appropriately shorten ΔT and T_calc when the deformation trend is abnormal or the alarm level is high to improve the monitoring accuracy and real-time performance.

9. The deformation monitoring method according to claim 1, characterized in that: The step of judging whether to continue to trigger the alarm mechanism according to the value of the high-precision solution result ΔY* and the deformation trend information comprises: If the high-precision solution result ΔY* is within the first threshold range and the deformation trend information does not indicate an abnormal deformation trend, the high-precision solution result ΔY* is stored and the state is recorded as normal, and the normal state is fed back to the monitoring station end, so that the monitoring station end stops uploading the early warning / alarm signal and continues to obtain preliminary observation data from the monitoring station; If the high-precision solution result ΔY* exceeds the first threshold range but does not reach the second threshold range, or the deformation trend information indicates a slight abnormal deformation trend, a low-risk warning reminder is sent to the security personnel, and high-precision solution is continued to track the deformation situation; If the high-precision solution result ΔY* exceeds the second threshold range but does not reach the third threshold range, or the deformation trend information indicates a significant abnormal deformation trend, a medium-risk alarm reminder is sent to the security personnel, and the high-precision solution continues; If the high-precision solution result ΔY* exceeds the third threshold range, or the deformation trend information indicates a serious abnormal deformation trend, a high-risk alarm reminder is sent to the security personnel, and high-precision solution is continued to obtain more detailed deformation information.

10. A deformation monitoring system, used to implement the deformation monitoring method according to any one of claims 1 to 9, characterized in that: The system comprises: Preliminary solution module: used to perform preliminary solution on the three-dimensional displacement change of the monitored object at the monitoring station end based on the original observation data set X collected by the reference station within the time interval T and the preliminary observation data set Y collected by the monitoring station, and obtain the preliminary solution result ΔY; Deformation trend analysis module: used to perform deformation trend analysis based on the preliminary solution results at each time point at the monitoring station to obtain deformation trend information; The first alarm mechanism module is used to determine whether the alarm mechanism needs to be triggered based on the preliminary solution result ΔY and deformation trend information at the monitoring station; High-precision solution module: If the alarm mechanism is triggered, the high-precision solution is started through the data center. The original observation data set X and the preliminary observation data set Y are combined to perform high-precision solution through the variational Bayesian method to obtain the high-precision solution result ΔY*. During the high-precision solution, the satellite health status is judged through RAIM satellite integrity monitoring, and the satellite signal is judged through interference detection to eliminate the positioning accuracy abnormality caused by satellite failure or signal interference. The second alarm mechanism module is used to determine whether the alarm mechanism needs to be triggered further based on the value of the high-precision solution result ΔY* and the deformation trend information.

Images