A dam earthquake safety monitoring method based on beidou and accelerometer fusion
By deploying BeiDou receivers and accelerometers at the same location at the dam monitoring point and constructing a joint state vector model, high-precision earthquake-time safety monitoring and post-earthquake safety assessment were achieved. This solved the problems of insufficient monitoring accuracy and stability in existing technologies, reduced the false alarm rate, and improved monitoring efficiency.
Patent Information
- Application Number
- CN202611321783.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-28
- Publication Date
- 2026-11-13
AI Technical Summary
The existing dam earthquake monitoring system has poor monitoring accuracy and stability, making it difficult to acquire displacement, velocity and acceleration synchronously and with high precision. Furthermore, the alarm mechanism is prone to false alarms and missed alarms, and it cannot effectively distinguish between environmental noise and the actual earthquake response.
Beidou receivers and accelerometers are deployed at the same site as the dam monitoring point to collect displacement and acceleration data with a unified time reference. A recursive model is constructed with displacement, velocity, and acceleration as the joint state vector. The state prediction is driven by high sampling rate accelerometers and corrected by low sampling rate Beidou displacement data. Forward filtering estimation and smoothing are performed to extract peak displacement, velocity, and acceleration. Safety assessment is carried out by combining response spectrum analysis and frequency domain dynamic characteristic analysis.
It achieves high precision and stability in earthquake-time safety monitoring, reduces false alarm rate, provides automation and quantification for post-earthquake safety assessment, simplifies on-site deployment procedures, and improves monitoring efficiency.
Smart Images

Figure CN123088239A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method for monitoring the earthquake safety of dams, belonging to the field of water conservancy project safety monitoring technology, and particularly to a method for monitoring the earthquake safety of dams based on the fusion of Beidou and accelerometer. Background Technology
[0002] Dams are crucial water conservancy infrastructures. Earthquakes can lead to serious consequences such as dam cracking, abutment rock loosening, and even dam failure. Therefore, seismic safety monitoring of dams is essential. Currently, dam seismic monitoring mainly uses GPS / BeiDou satellite positioning or accelerometers. While GPS / BeiDou can acquire the absolute displacement of the dam, its low sampling rate (1-20Hz) makes it difficult to capture high-frequency dynamic deformation during earthquakes, and it cannot capture the complete vibration process in a timely manner during an earthquake. Accelerometers have a high sampling rate (up to 200Hz), but drift problems exist when integrating displacement, making long-term monitoring results unreliable, and they cannot provide accurate permanent displacement for damage assessment after an earthquake. In addition, the alarm mechanisms of existing seismic safety monitoring systems are mostly based on single physical quantities, which are prone to false alarms and missed alarms, and cannot effectively distinguish between environmental noise and the actual earthquake response. Therefore, the monitoring accuracy and stability are poor.
[0003] Chinese patent application number 202511852836.9, filed on December 10, 2025, discloses a smart sensing system and operation method for reservoir dam operation safety based on air and ground hydraulic engineering. This system consists of an integrated air-ground hydraulic engineering monitoring and sensing system, an adaptive sampling module, a hierarchical distributed architecture, and an integrated hardware and software module. It collaboratively collects multi-source data such as deformation, seepage, stress and strain, vibration, and environmental parameters from five dimensions: air, ground, water, and engineering. It dynamically adjusts the monitoring frequency using an adaptive sampling strategy and completes data cleaning, standardization, spatiotemporal registration, and fusion processing through a distributed architecture to generate a high-quality comprehensive dataset. Although this patent can achieve full-element monitoring, it still has the following shortcomings:
[0004] The optimization objective of this design is only to minimize the displacement residual, which cannot simultaneously acquire velocity and acceleration with high precision, and its alarm mechanism is still based on a single indicator; thus, the monitoring accuracy and stability are still poor.
[0005] The information disclosed in this background section is intended only to enhance the understanding of the overall background of this patent application and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention
[0006] The purpose of this invention is to overcome the defects and problems of poor monitoring accuracy and stability in the existing technology, and to provide a dam earthquake safety monitoring method based on the fusion of Beidou and accelerometer with better monitoring accuracy and stability.
[0007] To achieve the above objectives, the technical solution of the present invention is:
[0008] This invention provides a method for monitoring the seismic safety of dams based on the fusion of BeiDou navigation satellite system and accelerometer data. The method includes the following steps:
[0009] Step 1: First, deploy a Beidou receiver for collecting dam displacement data and an accelerometer for collecting dam acceleration data at the same location as the dam monitoring point. The sampling frequency of the accelerometer is an integer multiple of the sampling frequency of the Beidou receiver. Then, perform time reference unification and preprocessing on the displacement data and acceleration data collected in the same time period to obtain the preprocessed displacement sequence and the preprocessed acceleration sequence, respectively.
[0010] Step 2: First, based on the preprocessed displacement and acceleration sequences, a recursive model is constructed. The recursive model uses displacement, velocity, and acceleration as a joint state vector, and the state equation of the recursive model uses the preprocessed acceleration sequence as the driving input. Then, the initial filtering state and initial covariance matrix are set, and the process noise covariance matrix and observation noise covariance matrix are determined. Then, based on the recursive model, state prediction is performed according to the accelerometer sampling frequency to obtain the state prediction value and covariance matrix. Then, based on the preprocessed displacement sequence, the state prediction value is corrected according to the BeiDou receiver sampling frequency, and the covariance matrix is updated. The state prediction and correction are repeated until all data epochs are processed to obtain the forward filtering estimation sequence.
[0011] Step 3: Perform RTS fixed-interval smoothing on the above forward-filtered estimation sequence to obtain a smoothed state estimation sequence; and extract the displacement time history, velocity time history and acceleration time history estimates with the same sampling frequency as the accelerometer from the smoothed state estimation sequence.
[0012] Step 4: Extract the peak displacement, peak velocity, and peak acceleration from the above displacement time history, velocity time history, and acceleration time history estimates respectively; determine that an earthquake has occurred in the dam and trigger an alarm if and only if the peak displacement, peak velocity, and peak acceleration all exceed their respective alarm thresholds, thereby achieving earthquake-time safety monitoring.
[0013] The method further includes:
[0014] Step 5: First, based on the estimated values of displacement time history, velocity time history, and acceleration time history, calculate the acceleration response spectrum, velocity response spectrum, displacement response spectrum, and cumulative absolute velocity in the time domain to obtain the response spectrum analysis results; then, perform a Fourier transform on the acceleration time history in the frequency domain to extract the natural frequency, damping ratio, and mode shape parameters of the dam body; finally, compare the natural frequency and damping ratio with the historical benchmark values under the pre-earthquake normal state to obtain the rate of change of natural frequency and the amplitude of change of damping ratio, which serve as the basis for judging structural stiffness degradation; thus, obtain the frequency domain dynamic characteristic analysis results.
[0015] Then, based on the results of the response spectrum analysis, the frequency domain dynamic characteristics analysis, and the cumulative absolute velocity, the degree of post-earthquake damage to the dam is graded and assessed, and a post-earthquake safety assessment report is generated to achieve post-earthquake safety monitoring.
[0016] The fifth step, namely, the graded assessment of the post-earthquake damage to the dam, includes:
[0017] First, we use three types of physical quantities—displacement, velocity, and acceleration—as well as the rate of change of natural frequency and the amplitude of change of damping ratio as evaluation indicators.
[0018] Then, the measured values of each evaluation index are extracted from the response spectrum analysis results and the frequency domain dynamic characteristic analysis results. Among them, the measured values of displacement index include peak displacement and maximum displacement change at the measuring point; the measured values of velocity index include peak velocity and cumulative absolute velocity; and the measured values of acceleration index include peak acceleration and maximum spectral value of acceleration response spectrum. Each measured value is compared with the seismic design parameters of the dam and the pre-earthquake benchmark values to obtain the degree of displacement damage, the degree of velocity damage, the degree of acceleration damage, and the degree of stiffness degradation.
[0019] Then, the comprehensive post-earthquake damage score of the dam body is obtained by multiplying the degree of displacement damage, the degree of velocity damage, the degree of acceleration damage, and the degree of stiffness degradation by their respective weights and summing them.
[0020] Finally, based on the comprehensive damage score and in accordance with the damage level classification standard, the post-earthquake damage level of the dam body is determined.
[0021] The post-earthquake safety assessment report includes: peak response parameters, response spectrum characteristic parameters, rate of change of dynamic characteristic parameters, and damage level.
[0022] In the first step, the co-location deployment includes: installing the Beidou receiver and the accelerometer on the same instrument pier at the dam monitoring point.
[0023] In the first step, the time base unification and preprocessing include:
[0024] First, both the accelerometer and the Beidou receiver are connected to the Beidou satellite timing signal to unify the acquisition clock of the accelerometer and the Beidou receiver with the Beidou time reference; then, based on this, the displacement data is linearly interpolated according to the time ratio to obtain complete displacement data.
[0025] The collected acceleration data is then subjected to zero-bias correction and noise reduction to obtain a preprocessed acceleration sequence; at the same time, the complete displacement data is subjected to gross error removal and multipath effect reduction preprocessing to obtain a preprocessed displacement sequence.
[0026] In the second step, the state vector includes:
[0027] ;
[0028] In the formula: For a continuous-time system state vector, The filtered displacement state components to be estimated are... The filtered velocity state component to be estimated. The filtered acceleration state component to be estimated;
[0029] In the second step, the recursive model includes state equations and observation equations;
[0030] The state equations include:
[0031] ;
[0032] ;
[0033] ;
[0034] In the formula: This is the current accelerometer sampling time. For the next accelerometer sampling time, This is the discrete state vector at the next accelerometer sampling time. Here is the state transition matrix. This is the discrete state vector at the current accelerometer sampling time. As the acceleration input matrix, This refers to the acceleration data collected by the accelerometer at the current sampling time. For discretized process noise, The accelerometer sampling interval;
[0035] The observation equations include:
[0036] ;
[0037] In the formula: The sampling time for the BeiDou receiver. for Displacement observations collected by the BeiDou receiver at all times. for The discrete state vector at time t. To design the matrix, for Displacement observation noise at any given time.
[0038] In the second step, setting the initial filtering state and the initial covariance matrix includes:
[0039] ;
[0040] ;
[0041] In the formula: Let be the initial state vector. The initial state covariance matrix;
[0042] In the second step, the state prediction based on the recursive model, performed according to the accelerometer sampling frequency, to obtain the state prediction value and covariance matrix, includes:
[0043] At each accelerometer sampling time, state prediction is performed using the following formula:
[0044] ;
[0045] In the formula: This is the predicted state value at the next accelerometer sampling time. This is the forward filtered estimate at the current accelerometer sampling time. This refers to the acceleration data collected by the accelerometer at the current sampling moment.
[0046] The state prediction values include displacement prediction values, velocity prediction values, and acceleration prediction values;
[0047] The state covariance matrix is updated synchronously using the following formula:
[0048] ;
[0049] In the formula: The state covariance matrix at the next accelerometer sampling time. Let be the state covariance matrix at the current accelerometer sampling time. Let be the process noise covariance matrix.
[0050] In the second step, the step of correcting the state prediction value and updating the covariance matrix based on the preprocessed displacement sequence according to the sampling frequency of the Beidou receiver includes:
[0051] When the displacement observation value collected by the Beidou receiver arrives, the Kalman filter gain is first calculated using the following formula:
[0052] ;
[0053] In the formula: for Kalman filter gain at time t. The sampling time for the BeiDou receiver. for The predicted state covariance matrix at time t. To design the matrix, The displacement observation noise covariance matrix;
[0054] The predicted state value is then corrected using the following formula:
[0055] ;
[0056] In the formula: for The state estimate after time correction, for The predicted state value at time 10:00. for Displacement observations collected by the BeiDou receiver at all times;
[0057] At the same time, the covariance matrix is updated using the following formula:
[0058] ;
[0059] In the formula: for The state covariance matrix after time correction, It is an identity matrix.
[0060] In the third step, extracting displacement time history, velocity time history, and acceleration time history estimates from the smoothed state estimation sequence at the same sampling frequency as the accelerometer includes:
[0061] First, extract the displacement component, velocity component, and acceleration component at the current moment from the smoothed state vector at each moment of the smoothed state estimation sequence.
[0062] The displacement components extracted at each time point are then arranged in chronological order to obtain displacement time history estimates; the velocity components extracted at each time point are arranged in chronological order to obtain velocity time history estimates; the acceleration components extracted at each time point are arranged in chronological order to obtain acceleration time history estimates; the sampling frequencies of the displacement time history, velocity time history, and acceleration time history estimates are all the same as the sampling frequency of the accelerometer.
[0063] In the fourth step, extracting the peak displacement, peak velocity, and peak acceleration from the estimated values of the displacement time history, velocity time history, and acceleration time history includes:
[0064] For the displacement time history estimate, the absolute value of the displacement value at each moment is taken, and the maximum absolute value is taken as the peak displacement; for the velocity time history estimate, the absolute value of the velocity value at each moment is taken, and the maximum absolute value is taken as the peak velocity; for the acceleration time history estimate, the absolute value of the acceleration value at each moment is taken, and the maximum absolute value is taken as the peak acceleration.
[0065] Compared with the prior art, the beneficial effects of the present invention are as follows:
[0066] 1. This invention discloses a method for earthquake safety monitoring of dams based on the fusion of BeiDou and accelerometers. First, a BeiDou receiver and an accelerometer are deployed at the same location at the dam monitoring point to collect displacement and acceleration data. Time reference unification and preprocessing are performed to obtain preprocessed displacement and acceleration sequences. Then, a recursive model with displacement, velocity, and acceleration as a joint state vector is constructed. High-sampling-rate accelerometer data drives state prediction, while low-sampling-rate BeiDou displacement data is used for periodic correction to obtain a forward-filtered estimation sequence. Next, the forward-filtered estimation sequence undergoes RTS fixed-interval smoothing processing to extract displacement, velocity, and acceleration time histories with the same sampling frequency as the accelerometers. Finally, peak displacement, peak velocity, and peak acceleration are extracted. An alarm is triggered when all three exceed their respective preset alarm thresholds, achieving earthquake-time safety monitoring. The advantages of this invention also include:
[0067] Firstly, only two types of equipment, namely Beidou receivers and accelerometers, need to be deployed. By constructing a three-parameter state space model with displacement, velocity, and acceleration as the joint state vector, the three physical quantities of displacement, velocity, and acceleration can be obtained synchronously and with high precision. This replaces the traditional solution that requires the deployment of three types of equipment, namely accelerometers, velocity meters, and Beidou, and provides a complete and accurate data foundation for earthquake safety monitoring.
[0068] Secondly, high-frequency prediction is performed using the accelerometer sampling interval as the basic step size, and measurement correction is performed using the BeiDou displacement sampling interval as the period. Between two BeiDou observations, the state prediction is maintained by accelerometer integration recursion. Whenever BeiDou displacement data arrives, correction is performed using absolute displacement as the anchor point. This ensures that the final output displacement, velocity, and acceleration time history sampling rate is consistent with the high sampling rate accelerometer input (generally 100Hz or 200Hz). This improves the problem that the low sampling rate (1-20Hz) of BeiDou monitoring makes it difficult to capture high-frequency dynamic deformation of earthquakes, ensuring that no key vibration information is missed during the earthquake response and improving the stability of monitoring.
[0069] Thirdly, the alarm is triggered only when the peak displacement, peak velocity, and peak acceleration all exceed their respective preset alarm thresholds. Compared with single-index alarms, it has stronger anti-interference capabilities, effectively reducing the false alarm rate caused by single sensor anomalies or environmental interference, and avoiding unnecessary emergency response and economic losses due to false alarms.
[0070] Therefore, this invention can not only achieve safety monitoring during earthquakes, but also has good monitoring accuracy and stability.
[0071] 2. This invention provides a method for monitoring the earthquake safety of dams based on the fusion of BeiDou navigation and accelerometers. The method further includes: calculating the displacement response spectrum, velocity response spectrum, acceleration response spectrum, and cumulative absolute velocity in the time domain based on the displacement time history, velocity time history, and acceleration time history, obtaining response spectrum analysis results; performing a Fourier transform on the acceleration time history in the frequency domain to extract the dam's natural frequency, damping ratio, and mode shape parameters, obtaining frequency domain dynamic characteristic analysis results; and then comprehensively assessing the degree of post-earthquake damage to the dam based on the response spectrum analysis results and the frequency domain dynamic characteristic analysis results, generating a post-earthquake safety assessment report, and achieving post-earthquake safety monitoring. In application, the time domain response spectrum is used to quantify the dynamic amplification effect of vibrations of different periods on the dam structure, the frequency domain dynamic characteristic parameters are used to determine whether the dam structure stiffness has degraded due to the earthquake, and the comprehensive assessment compares the measured parameters with the dam's seismic design parameters, integrating displacement, velocity, and acceleration information to determine the post-earthquake damage level of the dam body, providing a scientific and rapid quantitative basis for post-earthquake repair decisions. Therefore, this invention not only has good monitoring accuracy and stability, but also realizes the automation and quantification of post-earthquake safety assessment.
[0072] 3. This invention discloses a dam earthquake safety monitoring method based on the fusion of BeiDou and accelerometers. The BeiDou receiver and accelerometer are installed on the same instrument pier at the dam monitoring point. In application, this co-location ensures that the BeiDou receiver and accelerometer have a consistent spatial reference, guaranteeing that the displacement and acceleration data collected by both sensors are synchronous observations from the same physical measurement point. This provides a spatially consistent foundation for subsequent data fusion, effectively avoiding spatial errors introduced by inconsistent measurement points and ensuring the accuracy and reliability of the data basis for algorithm fusion. Simultaneously, it simplifies the on-site deployment process, reduces construction difficulty and installation time, and achieves efficient monitoring of three physical quantities produced by two devices. Therefore, this invention not only automates and quantifies post-earthquake safety assessment but also improves monitoring efficiency. Attached Figure Description
[0073] Figure 1 This is a schematic diagram of the overall process of the method described in this invention.
[0074] Figure 2 This is a schematic diagram of the installation of the Beidou receiver and accelerometer described in this invention.
[0075] In the diagram: Beidou receiver 1, pole 11, Beidou antenna 12, solar panel 13, waterproof box 14, strong earthquake data recorder 15, accelerometer 2, stainless steel protective shell 21, instrument pier 3. Detailed Implementation
[0076] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0077] See Figures 1-2 A method for monitoring the seismic safety of dams based on the fusion of BeiDou navigation satellite system and accelerometer, the method comprising the following steps:
[0078] Step 1: First, deploy a Beidou receiver 1 for collecting dam displacement data and an accelerometer 2 for collecting dam acceleration data at the same location as the dam monitoring point. The sampling frequency of the accelerometer is an integer multiple of the sampling frequency of the Beidou receiver. Then, perform time reference unification and preprocessing on the displacement data and acceleration data collected in the same time period to obtain the preprocessed displacement sequence and the preprocessed acceleration sequence, respectively.
[0079] Step 2: First, based on the preprocessed displacement and acceleration sequences, a recursive model is constructed. The recursive model uses displacement, velocity, and acceleration as a joint state vector, and the state equation of the recursive model uses the preprocessed acceleration sequence as the driving input. Then, the initial filtering state and initial covariance matrix are set, and the process noise covariance matrix and observation noise covariance matrix are determined. Then, based on the recursive model, state prediction is performed according to the accelerometer sampling frequency to obtain the state prediction value and covariance matrix. Then, based on the preprocessed displacement sequence, the state prediction value is corrected according to the BeiDou receiver sampling frequency, and the covariance matrix is updated. The state prediction and correction are repeated until all data epochs are processed to obtain the forward filtering estimation sequence.
[0080] Step 3: Perform RTS fixed-interval smoothing on the above forward-filtered estimation sequence to obtain a smoothed state estimation sequence; and extract the displacement time history, velocity time history and acceleration time history estimates with the same sampling frequency as the accelerometer from the smoothed state estimation sequence.
[0081] Step 4: Extract the peak displacement, peak velocity, and peak acceleration from the above displacement time history, velocity time history, and acceleration time history estimates respectively; determine that an earthquake has occurred in the dam and trigger an alarm if and only if the peak displacement, peak velocity, and peak acceleration all exceed their respective alarm thresholds, thereby achieving earthquake-time safety monitoring.
[0082] The method further includes:
[0083] Step 5: First, based on the estimated values of displacement time history, velocity time history, and acceleration time history, calculate the acceleration response spectrum, velocity response spectrum, displacement response spectrum, and cumulative absolute velocity in the time domain to obtain the response spectrum analysis results; then, perform a Fourier transform on the acceleration time history in the frequency domain to extract the natural frequency, damping ratio, and mode shape parameters of the dam body; finally, compare the natural frequency and damping ratio with the historical benchmark values under the pre-earthquake normal state to obtain the rate of change of natural frequency and the amplitude of change of damping ratio, which serve as the basis for judging structural stiffness degradation; thus, obtain the frequency domain dynamic characteristic analysis results.
[0084] Then, based on the results of the response spectrum analysis, the frequency domain dynamic characteristics analysis, and the cumulative absolute velocity, the degree of post-earthquake damage to the dam is graded and assessed, and a post-earthquake safety assessment report is generated to achieve post-earthquake safety monitoring.
[0085] The fifth step, namely, the graded assessment of the post-earthquake damage to the dam, includes:
[0086] First, we use three types of physical quantities—displacement, velocity, and acceleration—as well as the rate of change of natural frequency and the amplitude of change of damping ratio as evaluation indicators.
[0087] Then, the measured values of each evaluation index are extracted from the response spectrum analysis results and the frequency domain dynamic characteristic analysis results. Among them, the measured values of displacement index include peak displacement and maximum displacement change at the measuring point; the measured values of velocity index include peak velocity and cumulative absolute velocity; and the measured values of acceleration index include peak acceleration and maximum spectral value of acceleration response spectrum. Each measured value is compared with the seismic design parameters of the dam and the pre-earthquake benchmark values to obtain the degree of displacement damage, the degree of velocity damage, the degree of acceleration damage, and the degree of stiffness degradation.
[0088] Then, the comprehensive post-earthquake damage score of the dam body is obtained by multiplying the degree of displacement damage, the degree of velocity damage, the degree of acceleration damage, and the degree of stiffness degradation by their respective weights and summing them.
[0089] Finally, based on the comprehensive damage score and in accordance with the damage level classification standard, the post-earthquake damage level of the dam body is determined.
[0090] The post-earthquake safety assessment report includes: peak response parameters, response spectrum characteristic parameters, rate of change of dynamic characteristic parameters, and damage level.
[0091] In the first step, the co-location deployment includes: installing the Beidou receiver 1 and the accelerometer 2 on the same instrument pier 3 at the dam monitoring point.
[0092] In the first step, the time base unification and preprocessing include:
[0093] First, both accelerometer 2 and Beidou receiver 1 are connected to the Beidou satellite timing signal to unify the acquisition clock of accelerometer 2 and Beidou receiver 1 with the Beidou time reference; and on this basis, the displacement data is linearly interpolated according to the time ratio to obtain complete displacement data.
[0094] The collected acceleration data is then subjected to zero-bias correction and noise reduction to obtain a preprocessed acceleration sequence; at the same time, the complete displacement data is subjected to gross error removal and multipath effect reduction preprocessing to obtain a preprocessed displacement sequence.
[0095] In the second step, the state vector includes:
[0096] ;
[0097] In the formula: For a continuous-time system state vector, The filtered displacement state components to be estimated are... The filtered velocity state component to be estimated. The filtered acceleration state component to be estimated;
[0098] In the second step, the recursive model includes state equations and observation equations;
[0099] The state equations include:
[0100] ;
[0101] ;
[0102] ;
[0103] In the formula: This is the current accelerometer sampling time. For the next accelerometer sampling time, This is the discrete state vector at the next accelerometer sampling time. Here is the state transition matrix. This is the discrete state vector at the current accelerometer sampling time. As the acceleration input matrix, The acceleration data collected by accelerometer 2 at the current accelerometer sampling time. For discretized process noise, The accelerometer sampling interval;
[0104] The observation equations include:
[0105] ;
[0106] In the formula: The sampling time for the BeiDou receiver. for Displacement observation values collected by Beidou receiver 1 at all times. for The discrete state vector at time t. To design the matrix, for Displacement observation noise at any given time.
[0107] In the second step, setting the initial filtering state and the initial covariance matrix includes:
[0108] ;
[0109] ;
[0110] In the formula: Let be the initial state vector. The initial state covariance matrix;
[0111] In the second step, the state prediction based on the recursive model, performed according to the accelerometer sampling frequency, to obtain the state prediction value and covariance matrix, includes:
[0112] At each accelerometer sampling time, state prediction is performed using the following formula:
[0113] ;
[0114] In the formula: This is the predicted state value at the next accelerometer sampling time. This is the forward filtered estimate at the current accelerometer sampling time. This refers to the acceleration data collected by accelerometer 2 at the current accelerometer sampling moment;
[0115] The state prediction values include displacement prediction values, velocity prediction values, and acceleration prediction values;
[0116] The state covariance matrix is updated synchronously using the following formula:
[0117] ;
[0118] In the formula: The state covariance matrix at the next accelerometer sampling time. Let be the state covariance matrix at the current accelerometer sampling time. Let be the process noise covariance matrix.
[0119] In the second step, the step of correcting the state prediction value and updating the covariance matrix based on the preprocessed displacement sequence according to the sampling frequency of the Beidou receiver includes:
[0120] When the displacement observation value collected by Beidou receiver 1 arrives, the Kalman filter gain is first calculated using the following formula:
[0121] ;
[0122] In the formula: for Kalman filter gain at time t. The sampling time for the BeiDou receiver. for The predicted state covariance matrix at time t. To design the matrix, The displacement observation noise covariance matrix;
[0123] The predicted state value is then corrected using the following formula:
[0124] ;
[0125] In the formula: for The state estimate after time correction, for The predicted state value at time 10:00. for Displacement observation values collected by Beidou receiver 1 at any given moment;
[0126] At the same time, the covariance matrix is updated using the following formula:
[0127] ;
[0128] In the formula: for The state covariance matrix after time correction, It is an identity matrix.
[0129] In the third step, extracting displacement time history, velocity time history, and acceleration time history estimates from the smoothed state estimation sequence at the same sampling frequency as the accelerometer includes:
[0130] First, extract the displacement component, velocity component, and acceleration component at the current moment from the smoothed state vector at each moment of the smoothed state estimation sequence.
[0131] The displacement components extracted at each time point are then arranged in chronological order to obtain displacement time history estimates; the velocity components extracted at each time point are arranged in chronological order to obtain velocity time history estimates; the acceleration components extracted at each time point are arranged in chronological order to obtain acceleration time history estimates; the sampling frequencies of the displacement time history, velocity time history, and acceleration time history estimates are all the same as the sampling frequency of the accelerometer.
[0132] In the fourth step, extracting the peak displacement, peak velocity, and peak acceleration from the estimated values of the displacement time history, velocity time history, and acceleration time history includes:
[0133] For the displacement time history estimate, the absolute value of the displacement value at each moment is taken, and the maximum absolute value is taken as the peak displacement; for the velocity time history estimate, the absolute value of the velocity value at each moment is taken, and the maximum absolute value is taken as the peak velocity; for the acceleration time history estimate, the absolute value of the acceleration value at each moment is taken, and the maximum absolute value is taken as the peak acceleration.
[0134] The following are supplementary descriptions of the present invention:
[0135] In the first step of this invention, the locations of the dam monitoring points include the dam crest, the dam slope change location, the dam shoulder, and the dam foundation.
[0136] Preferably, in the first step of this invention, the accelerometer sampling frequency is 100Hz and the Beidou receiver sampling frequency is 1Hz.
[0137] Example 1:
[0138] See Figure 1 A method for monitoring the seismic safety of dams based on the fusion of BeiDou navigation satellite system and accelerometer, the method comprising the following steps:
[0139] Step 1: First, deploy a Beidou receiver 1 for collecting dam displacement data and an accelerometer 2 for collecting dam acceleration data at the same location as the dam monitoring point. The sampling frequency of the accelerometer is an integer multiple of the sampling frequency of the Beidou receiver. Then, perform time reference unification and preprocessing on the displacement data and acceleration data collected in the same time period to obtain the preprocessed displacement sequence and the preprocessed acceleration sequence, respectively.
[0140] Step 2: First, based on the preprocessed displacement and acceleration sequences, a recursive model is constructed. The recursive model uses displacement, velocity, and acceleration as a joint state vector, and the state equation of the recursive model uses the preprocessed acceleration sequence as the driving input. Then, the initial filtering state and initial covariance matrix are set, and the process noise covariance matrix and observation noise covariance matrix are determined. Then, based on the recursive model, state prediction is performed according to the accelerometer sampling frequency to obtain the state prediction value and covariance matrix. Then, based on the preprocessed displacement sequence, the state prediction value is corrected according to the BeiDou receiver sampling frequency, and the covariance matrix is updated. The state prediction and correction are repeated until all data epochs are processed to obtain the forward filtering estimation sequence.
[0141] Step 3: Perform RTS fixed-interval smoothing on the above forward-filtered estimation sequence to obtain a smoothed state estimation sequence; and extract the displacement time history, velocity time history and acceleration time history estimates with the same sampling frequency as the accelerometer from the smoothed state estimation sequence.
[0142] Step 4: Extract the peak displacement, peak velocity, and peak acceleration from the above displacement time history, velocity time history, and acceleration time history estimates respectively; determine that an earthquake has occurred in the dam and trigger an alarm if and only if the peak displacement, peak velocity, and peak acceleration all exceed their respective alarm thresholds, thereby achieving earthquake-time safety monitoring.
[0143] Example 2:
[0144] The basic content is the same as in Example 1, except that the method further includes:
[0145] Step 5: First, based on the estimated values of displacement time history, velocity time history, and acceleration time history, calculate the acceleration response spectrum, velocity response spectrum, displacement response spectrum, and cumulative absolute velocity in the time domain to obtain the response spectrum analysis results; then, perform a Fourier transform on the acceleration time history in the frequency domain to extract the dam's natural frequency, damping ratio, and mode shape parameters; next, compare the natural frequency and damping ratio with historical benchmark values under the pre-earthquake normal state to obtain the rate of change of natural frequency and the amplitude of change of damping ratio, which serve as the basis for judging structural stiffness degradation; thus obtaining the frequency domain dynamic characteristic analysis results; then, based on the response spectrum analysis results, frequency domain dynamic characteristic analysis results, and cumulative absolute velocity, classify and assess the degree of post-earthquake damage to the dam, generate a post-earthquake safety assessment report, and realize post-earthquake safety monitoring;
[0146] The fifth step, the graded assessment of the post-earthquake damage to the dam, includes: first, using three physical quantities—displacement, velocity, and acceleration—as well as the rate of change of natural frequency and the amplitude of change of damping ratio as assessment indicators; then, extracting the measured values of each assessment indicator from the response spectrum analysis results and the frequency domain dynamic characteristic analysis results. Specifically, the measured values of the displacement indicator include peak displacement and the maximum displacement change at the measuring point; the measured values of the velocity indicator include peak velocity and cumulative absolute velocity; and the measured values of the acceleration indicator include peak acceleration and the maximum spectral value of the acceleration response spectrum. The measured values are then compared with the... The seismic design parameters of the dam were compared one by one with the pre-earthquake baseline values to obtain the degree of displacement damage, velocity damage, acceleration damage, and stiffness degradation. Then, the comprehensive post-earthquake damage score of the dam body was obtained by multiplying each of the degree of displacement damage, velocity damage, acceleration damage, and stiffness degradation by their respective weights and summing them. Finally, based on the comprehensive damage score and in accordance with the damage level classification standard, the post-earthquake damage level of the dam body was determined. The post-earthquake safety assessment report includes: peak response parameters, response spectrum characteristic parameters, rate of change of dynamic characteristic parameters, and damage level.
[0147] In application, the response spectrum analysis results are used to quantify the dynamic amplification effect of earthquakes of different periods on the dam structure. The displacement response spectrum, velocity response spectrum, and acceleration response spectrum represent the maximum displacement, maximum velocity, and maximum acceleration response of a single-mass system with different natural periods under the same earthquake action, respectively. The cumulative absolute velocity (CAV), the integral of the absolute value of the acceleration time history over time, is an important indicator for assessing cumulative earthquake damage. Frequency domain dynamic characteristic analysis is used to determine whether the dam structure stiffness has degraded due to the earthquake. The natural frequency is determined by the peak position of the power spectral density curve, the damping ratio is calculated using the half-power bandwidth method, and the mode shape is determined by the amplitude ratio at each measuring point. The frequency domain dynamic characteristic analysis results include the natural frequency, damping ratio, mode shape parameters, as well as the rate of change of the natural frequency, the amplitude of the damping ratio change, and the criteria for determining structural stiffness degradation.
[0148] Meanwhile, based on long-term monitoring data, the normal fluctuation range and evolution trend line of various dynamic characteristic parameters such as natural frequency and damping ratio are established. Combining theoretical calculation and statistical methods, a three-level early warning threshold of "attention-warning-alarm" is determined. Structural damage such as dam body cracking and dam shoulder rock relaxation will lead to a decrease in stiffness and a drop in natural frequency. When the monitoring parameters deviate significantly from the normal evolution trend, the system will automatically trigger the corresponding level of early warning.
[0149] In the post-earthquake comprehensive assessment, displacement, velocity, and acceleration, along with the rate of change of natural frequency and the amplitude of change of damping ratio, were used as assessment indicators. The rate of change of natural frequency and the amplitude of change of damping ratio were obtained by comparing the natural frequency and damping ratio identified during the earthquake with historical benchmark values under normal pre-earthquake conditions. Measured values of each assessment indicator were extracted from the response spectrum analysis and frequency domain dynamic characteristic analysis results. Measured values for displacement included peak displacement (PGD) and the maximum displacement change at the measuring point; measured values for velocity included peak velocity (PGV) and cumulative absolute velocity (CAV); and measured values for acceleration included peak acceleration (PGA) and the maximum spectral value of the acceleration response spectrum. Each measured value was compared with the dam's seismic design parameters and pre-earthquake benchmark values to obtain the degree of displacement damage, velocity damage, acceleration damage, and stiffness degradation. Then, each degree of damage was multiplied by its corresponding weight and summed to obtain the comprehensive post-earthquake damage score of the dam body. Finally, based on the comprehensive damage score and in accordance with the damage level classification standards, the post-earthquake damage level of the dam body was determined.
[0150] In the post-earthquake safety assessment report, the peak response parameters include measured values of PGA, PGV, and PGD; the response spectrum characteristic parameters include the peak value and characteristic period of the response spectrum; the rate of change of the dynamic characteristic parameters includes the rate of decrease of natural frequency and the increase of damping ratio; and the damage level assessment conclusion includes normal, caution, warning, and alarm.
[0151] Example 3:
[0152] The basic content is the same as in Embodiment 1, except that: in the first step, the co-location deployment includes: installing the Beidou receiver 1 and the accelerometer 2 on the same instrument pier 3 at the dam monitoring point;
[0153] In the first step, the time reference unification and preprocessing includes: firstly, connecting both accelerometer 2 and Beidou receiver 1 to the Beidou satellite timing signal to unify the acquisition clocks of accelerometer 2 and Beidou receiver 1 with the Beidou time reference; then, linearly interpolating the displacement data according to the time ratio to obtain complete displacement data; next, performing zero-bias correction and noise reduction on the acquired acceleration data to obtain a preprocessed acceleration sequence; simultaneously, performing gross error removal and multipath effect reduction preprocessing on the complete displacement data to obtain a preprocessed displacement sequence.
[0154] In application, by installing the Beidou receiver 1 and the accelerometer 2 on the same instrument pier 3 at the dam monitoring point, the Beidou receiver 1 and the accelerometer 2 have a consistent spatial reference. This ensures that the displacement data collected by the Beidou receiver 1 and the acceleration data collected by the accelerometer 2 are synchronous observations from the same physical measurement point, providing a spatial consistency basis for subsequent data fusion. The co-location effectively avoids spatial errors caused by inconsistent measurement points, ensuring that the data basis for algorithm fusion is accurate and reliable.
[0155] Unified time reference and preprocessing further ensure the quality and time consistency of the input data. By using BeiDou timing signals and linear interpolation, high-frequency acceleration and low-frequency displacement are precisely aligned on the time axis. Through zero bias correction, noise reduction, gross error elimination and multipath attenuation, the influence of sensor errors and environmental interference on the original data is eliminated, ensuring the accuracy and reliability of the input data for subsequent recursive models.
[0156] See Figure 2 Specifically, the instrument pier 3 is made of concrete; the Beidou receiver 1 includes a pole 11, the bottom of which is fixedly connected to the top of the instrument pier 3, a Beidou antenna 12 is installed on the top of the pole 11, a solar panel 13 is installed in the middle of the pole 11, and a waterproof box 14 is installed in the lower middle part of the pole 11; the accelerometer 2 is fixedly installed on the top of the instrument pier 3 through a stainless steel protective shell 21; the solar panel 13 is used to power the accelerometer 2 and the Beidou receiver 1; the waterproof box 14 contains a solar controller, a lightning protection module, a battery, a 4G router, and a strong earthquake data recorder 15; the strong earthquake data recorder 15 is connected to the accelerometer 2 and is used to collect data from the accelerometer 2; the battery is used to power the equipment in poor sunlight conditions or at night; the 4G router is used for equipment communication; the lightning protection module is connected to the ground network to protect the equipment in thunderstorms.
[0157] Example 4:
[0158] The basic content is the same as in Example 1, except that in the second step, the state vector includes:
[0159] ;
[0160] In the formula: For a continuous-time system state vector, The filtered displacement state components to be estimated are... The filtered velocity state component to be estimated. The filtered acceleration state component to be estimated;
[0161] In the second step, the recursive model includes state equations and observation equations;
[0162] The state equations include:
[0163] ;
[0164] ;
[0165] ;
[0166] In the formula: This is the current accelerometer sampling time. For the next accelerometer sampling time, This is the discrete state vector at the next accelerometer sampling time. Here is the state transition matrix. This is the discrete state vector at the current accelerometer sampling time. As the acceleration input matrix, The acceleration data collected by accelerometer 2 at the current accelerometer sampling time. For discretized process noise, The accelerometer sampling interval;
[0167] The observation equations include:
[0168] ;
[0169] In the formula: The sampling time for the BeiDou receiver. for Displacement observation values collected by Beidou receiver 1 at all times. for The discrete state vector at time t. To design the matrix, for Displacement observation noise at any given time.
[0170] In application, accelerometer observations are used as control variables for the state equations to establish continuous-time state equations:
[0171] ;
[0172] In the formula: The state vector of a continuous-time system; The continuous-time acceleration data (i.e., control quantity) collected by accelerometer 2; The system state noise follows a set pattern. exist distributed, Let Variance be the acceleration variance, and its covariance matrix be... ,in Determined based on the static noise characteristics of the accelerometer;
[0173] Discretizing the above continuous-time state equations yields the discrete-time state equations:
[0174] ;
[0175] Wherein, the state transition matrix With acceleration input matrix All based on accelerometer sampling intervals The only certainty is that its derivation is based on Newton's kinematics formulas (Taylor series expansion).
[0176] Establish an observation equation using BeiDou displacement as the observed quantity:
[0177] ;
[0178] in, obey exist distributed, Let V be the displacement variance, and its covariance matrix be... , Based on the accuracy of BeiDou positioning; design matrix. .
[0179] Example 5:
[0180] The basic content is the same as in Example 1, except that in the second step, setting the initial filtering state and the initial covariance matrix includes:
[0181] ;
[0182] ;
[0183] In the formula: Let be the initial state vector. The initial state covariance matrix;
[0184] In the second step, the state prediction based on the recursive model, performed according to the accelerometer sampling frequency, to obtain the state prediction value and covariance matrix, includes:
[0185] At each accelerometer sampling time, state prediction is performed using the following formula:
[0186] ;
[0187] In the formula: This is the predicted state value at the next accelerometer sampling time. This is the forward filtered estimate at the current accelerometer sampling time. This refers to the acceleration data collected by accelerometer 2 at the current accelerometer sampling moment;
[0188] The state prediction values include displacement prediction values, velocity prediction values, and acceleration prediction values;
[0189] The state covariance matrix is updated synchronously using the following formula:
[0190] ;
[0191] In the formula: The state covariance matrix at the next accelerometer sampling time. Let be the state covariance matrix at the current accelerometer sampling time. The process noise covariance matrix;
[0192] In the second step, the step of correcting the state prediction value and updating the covariance matrix based on the preprocessed displacement sequence according to the sampling frequency of the Beidou receiver includes:
[0193] When the displacement observation value collected by Beidou receiver 1 arrives, the Kalman filter gain is first calculated using the following formula:
[0194] ;
[0195] In the formula: for Kalman filter gain at time t. The sampling time for the BeiDou receiver. for The predicted state covariance matrix at time t. To design the matrix, The displacement observation noise covariance matrix;
[0196] The predicted state value is then corrected using the following formula:
[0197] ;
[0198] In the formula: for The state estimate after time correction, for The predicted state value at time 10:00. for Displacement observation values collected by Beidou receiver 1 at any given moment;
[0199] At the same time, the covariance matrix is updated using the following formula:
[0200] ;
[0201] In the formula: for The state covariance matrix after time correction, It is an identity matrix.
[0202] In application, initialization is performed before filtering begins, setting the initial state vector to zero and the initial state covariance matrix to the identity matrix. The process noise covariance matrix is then determined based on the static noise characteristics of the accelerometer. The observation noise covariance matrix is determined based on the BeiDou positioning accuracy. ;
[0203] At each accelerometer sampling time, perform a time update based on the accelerometer data (prediction step): using the acceleration data collected by accelerometer 2 at the current accelerometer sampling time. Perform state prediction in one step:
[0204] ;
[0205] The state covariance matrix is updated synchronously using the following formula:
[0206] ;
[0207] The uncertainty of the state estimation is propagated forward. This prediction step is performed with a high sampling rate of the accelerometer (usually 100Hz or 200Hz) as the basic step size. The state prediction is maintained by accelerometer integration recursion between two BeiDou observations to ensure that no high-frequency response is missed.
[0208] When the displacement observation value collected by the Beidou receiver arrives, the correction step is executed;
[0209] The Kalman filter gain is determined by the predicted state covariance matrix at the current time. and displacement observation noise covariance matrix Together, we determine that the greater the prediction uncertainty or the lower the BeiDou observation noise, the greater the gain and the more the filtering depends on the current observation value; conversely, the smaller the gain, the more the filtering depends on the model prediction.
[0210] The displacement observations collected by the BeiDou receiver are then used to correct the state prediction values.
[0211] in, The residual represents the deviation between the measured displacement and the predicted displacement in BeiDou. As a three-dimensional gain vector, the displacement residual is simultaneously transformed into a correction amount for the three state variables of displacement, velocity, and acceleration, so that the correction amount of the BeiDou displacement observation value can be synchronously distributed to the three state components.
[0212] Then, the state covariance matrix is updated so that the corrected state covariance matrix is smaller than that at the time of prediction, indicating an increase in confidence in the current estimate.
[0213] Between two BeiDou observations, the state prediction is maintained by accelerometer integral recursion. Whenever BeiDou displacement data arrives, Kalman gain weighting correction is implemented with absolute displacement as the anchor point to achieve closed-loop correction of low sampling rate absolute observation to high sampling rate relative observation, providing real-time, continuous and optimal state estimation results for subsequent RTS smoothing.
[0214] By alternately executing the above-mentioned high-frequency prediction and low-frequency correction asynchronously until all data epochs have been processed, the forward filter estimation sequence is obtained.
[0215] Example 6:
[0216] The basic content is the same as in Embodiment 1, except that in the third step, extracting displacement time history, velocity time history, and acceleration time history estimates from the smoothed state estimation sequence with the same sampling frequency as the accelerometer includes: firstly, extracting the displacement component, velocity component, and acceleration component from the smoothed state vector at each moment of the smoothed state estimation sequence; then, arranging the displacement components extracted at each moment in chronological order to obtain displacement time history estimates; arranging the velocity components extracted at each moment in chronological order to obtain velocity time history estimates; and arranging the acceleration components extracted at each moment in chronological order to obtain acceleration time history estimates; the sampling frequency of the displacement time history, velocity time history, and acceleration time history estimates is the same as the sampling frequency of the accelerometer.
[0217] In application, after obtaining all forward filtering estimation results within the entire time interval [1,N], the RTS fixed interval smoothing algorithm is used for backward refinement; the process is then recursively applied from the final time N, with the initial smoothing condition set to... ;
[0218] For each time k (k=N-1, N-2, ..., 1), the following formula is used to recursively calculate the formula in reverse:
[0219] ;
[0220] Synchronously update the smoothed state covariance matrix:
[0221] ;
[0222] In the formula: for The state estimate after time-smoothing. for Forward filter estimate at time t, for The forward filter state covariance matrix at time t. for The predicted state covariance matrix at time t. for The state estimate after time-smoothing. for A one-step prediction value at time 1. for The state covariance matrix after time-smoothing; for The state covariance matrix after time-smoothing;
[0223] RTS smoothing utilizes all observation information within the entire time interval to globally refine the forward filtering results in a "correcting the past with the future" manner, thereby further improving the estimation accuracy of displacement and velocity compared to using forward filtering alone. At the same time, it effectively suppresses BeiDou high-frequency noise and overcomes the initial value error of filtering, ensuring that the time histories of the three physical quantities in the fused output reach the global optimal estimate within the entire time interval.
[0224] From the smoothed state vector The displacement estimates at each time point are extracted separately. Speed estimates and acceleration estimates The displacement time histories were then arranged in chronological order to obtain the estimated values. Speed and time history Estimates and acceleration time history estimates The sampling frequencies of the three time histories are the same as those of the accelerometer (generally set to 100Hz or 200Hz), realizing the complete reconstruction of the time histories of the three physical quantities from the low sampling rate BeiDou displacement (1-20Hz) to the high sampling rate.
[0225] Example 7:
[0226] The basic content is the same as in Example 1, except that in the fourth step, extracting the peak displacement, peak velocity, and peak acceleration from the above-mentioned displacement time history, velocity time history, and acceleration time history estimates respectively includes: taking the absolute value of the displacement value at each moment in the displacement time history estimate, and using the maximum absolute value as the peak displacement; taking the absolute value of the velocity value at each moment in the velocity time history estimate, and using the maximum absolute value as the peak velocity; taking the absolute value of the acceleration value at each moment in the acceleration time history estimate, and using the maximum absolute value as the peak acceleration.
[0227] In application, the peak displacement (PGD) characterizes the maximum permanent deformation and overall stability change of the dam body during an earthquake; the peak velocity (PGV) characterizes the maximum structural stress and damage potential; and the peak ground acceleration (PGA) characterizes the maximum seismic input intensity. These three are the three core quantitative indicators for dam safety monitoring during earthquakes.
[0228] In terms of alarm logic, a three-indicator joint alarm strategy is adopted: an earthquake is determined to have occurred in the dam and an alarm is triggered only when the peak displacement, peak velocity and peak acceleration all exceed their respective preset alarm thresholds; if any one of the indicators does not exceed the threshold, no alarm is triggered; this mechanism eliminates false alarms caused by single sensor anomalies or environmental interference from a logical level, avoiding unnecessary emergency response and economic losses due to false alarms.
[0229] The alarm thresholds for peak displacement (PGD), peak velocity (PGV), and peak acceleration (PGA) are predetermined based on the dam type, dam height, seismic intensity, and seismic design parameters. Specifically, the PGA alarm threshold is determined based on an earthquake intensity of 5 degrees, typically ranging from 0.025g to 0.05g; the PGV alarm threshold is also determined based on an earthquake intensity of 5 degrees, typically ranging from 1.77 to 3.8 cm / s; the PGD alarm threshold is determined based on a long-term monitoring baseline according to statistical control limits, with specific values varying depending on the dam type. This is because dams are generally not damaged when the earthquake intensity is 5 degrees or below, but may be affected when the earthquake intensity exceeds 5 degrees. Furthermore, since different dam types have different seismic resistance capabilities, the alarm thresholds can be adjusted within the above ranges. For example, earth-rock dams have relatively weaker seismic resistance capabilities, so lower thresholds can be used; concrete dams have relatively stronger seismic resistance capabilities, so higher thresholds can be used.
[0230] The extracted peak displacement, peak velocity, and peak acceleration are used directly for real-time earthquake alarm judgment and provide peak response parameters for post-earthquake safety assessment reports. Together with response spectrum characteristic parameters and damage level assessment conclusions, they constitute a complete post-earthquake safety assessment report.
[0231] The above description is only a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. Any equivalent modifications or changes made by those skilled in the art based on the content disclosed in the present invention should be included within the scope of protection set forth in the claims.
Claims
1. A method for monitoring the seismic safety of dams based on the fusion of BeiDou navigation satellite system and accelerometer, characterized in that: The method includes the following steps: Step 1: First, deploy a Beidou receiver (1) for collecting dam displacement data and an accelerometer (2) for collecting dam acceleration data at the same location as the dam monitoring point. The sampling frequency of the accelerometer is an integer multiple of the sampling frequency of the Beidou receiver. Then, perform time reference unification and preprocessing on the displacement data and acceleration data collected in the same time period to obtain the preprocessed displacement sequence and the preprocessed acceleration sequence, respectively. Step 2: First, based on the preprocessed displacement and acceleration sequences, a recursive model is constructed. The recursive model uses displacement, velocity, and acceleration as a joint state vector, and the state equation of the recursive model uses the preprocessed acceleration sequence as the driving input. Then, the initial filtering state and initial covariance matrix are set, and the process noise covariance matrix and observation noise covariance matrix are determined. Then, based on the recursive model, state prediction is performed according to the accelerometer sampling frequency to obtain the state prediction value and covariance matrix. Then, based on the preprocessed displacement sequence, the state prediction value is corrected according to the BeiDou receiver sampling frequency, and the covariance matrix is updated. The state prediction and correction are repeated until all data epochs are processed to obtain the forward filtering estimation sequence. Step 3: Perform RTS fixed-interval smoothing on the above forward-filtered estimation sequence to obtain a smoothed state estimation sequence; and extract the displacement time history, velocity time history and acceleration time history estimates with the same sampling frequency as the accelerometer from the smoothed state estimation sequence. Step 4: Extract the peak displacement, peak velocity, and peak acceleration from the above displacement time history, velocity time history, and acceleration time history estimates, respectively; determine that an earthquake has occurred in the dam and trigger an alarm if and only if the peak displacement, peak velocity, and peak acceleration all exceed their respective alarm thresholds, thereby achieving earthquake-time safety monitoring.
2. The method for monitoring the seismic safety of dams based on the fusion of BeiDou and accelerometers as described in claim 1, characterized in that: The method further includes: Step 5: First, based on the estimated values of displacement time history, velocity time history, and acceleration time history, calculate the acceleration response spectrum, velocity response spectrum, displacement response spectrum, and cumulative absolute velocity in the time domain to obtain the response spectrum analysis results; then, perform a Fourier transform on the acceleration time history in the frequency domain to extract the natural frequency, damping ratio, and mode shape parameters of the dam body; finally, compare the natural frequency and damping ratio with the historical benchmark values under the pre-earthquake normal state to obtain the rate of change of natural frequency and the amplitude of change of damping ratio, which serve as the basis for judging structural stiffness degradation; thus, obtain the frequency domain dynamic characteristic analysis results. Then, based on the results of the response spectrum analysis, the frequency domain dynamic characteristics analysis, and the cumulative absolute velocity, the degree of post-earthquake damage to the dam is graded and assessed, and a post-earthquake safety assessment report is generated to achieve post-earthquake safety monitoring.
3. The method for monitoring the seismic safety of dams based on the fusion of BeiDou and accelerometers according to claim 2, characterized in that: The fifth step, namely, the graded assessment of the post-earthquake damage to the dam, includes: First, we use three types of physical quantities—displacement, velocity, and acceleration—as well as the rate of change of natural frequency and the amplitude of change of damping ratio as evaluation indicators. Then, the measured values of each evaluation index are extracted from the response spectrum analysis results and the frequency domain dynamic characteristic analysis results. Among them, the measured values of displacement index include peak displacement and maximum displacement change at the measuring point; the measured values of velocity index include peak velocity and cumulative absolute velocity; and the measured values of acceleration index include peak acceleration and maximum spectral value of acceleration response spectrum. Each measured value is compared with the seismic design parameters of the dam and the pre-earthquake benchmark values to obtain the degree of displacement damage, the degree of velocity damage, the degree of acceleration damage, and the degree of stiffness degradation. Then, the comprehensive post-earthquake damage score of the dam body is obtained by multiplying the degree of displacement damage, the degree of velocity damage, the degree of acceleration damage, and the degree of stiffness degradation by their respective weights and summing them. Finally, based on the comprehensive damage score and in accordance with the damage level classification standard, the post-earthquake damage level of the dam body is determined. The post-earthquake safety assessment report includes: peak response parameters, response spectrum characteristic parameters, rate of change of dynamic characteristic parameters, and damage level.
4. A method for monitoring the seismic safety of dams based on the fusion of BeiDou and accelerometers as described in claim 1 or 2, characterized in that: In the first step, the co-location deployment includes: installing the Beidou receiver (1) and the accelerometer (2) on the same instrument pier (3) at the dam monitoring point.
5. A method for monitoring the seismic safety of dams based on the fusion of BeiDou and accelerometers as described in claim 1 or 2, characterized in that: In the first step, the time base unification and preprocessing include: First, connect both the accelerometer (2) and the Beidou receiver (1) to the Beidou satellite timing signal, and unify the acquisition clock of the accelerometer (2) and the Beidou receiver (1) with the Beidou time reference; and on this basis, perform linear interpolation on the displacement data according to the time ratio to obtain complete displacement data; The collected acceleration data is then subjected to zero-bias correction and noise reduction to obtain a preprocessed acceleration sequence; at the same time, the complete displacement data is subjected to gross error removal and multipath effect reduction preprocessing to obtain a preprocessed displacement sequence.
6. A method for monitoring the seismic safety of dams based on the fusion of BeiDou and accelerometers as described in claim 1 or 2, characterized in that: In the second step, the state vector includes: ; In the formula: For a continuous-time system state vector, The filtered displacement state components to be estimated are: The filtered velocity state component to be estimated. The filtered acceleration state component to be estimated; In the second step, the recursive model includes state equations and observation equations; The state equations include: ; ; ; In the formula: This is the current accelerometer sampling time. For the next accelerometer sampling time, This is the discrete state vector at the next accelerometer sampling time. Here is the state transition matrix. This is the discrete state vector at the current accelerometer sampling time. As the acceleration input matrix, The acceleration data collected by accelerometer (2) at the current accelerometer sampling time. For discretized process noise, The accelerometer sampling interval; The observation equations include: ; In the formula: The sampling time for the BeiDou receiver. for Displacement observations collected by the Beidou receiver (1) at any given time. for The discrete state vector at time t. To design the matrix, for Displacement observation noise at any given time.
7. A method for monitoring the seismic safety of dams based on the fusion of BeiDou and accelerometers as described in claim 6, characterized in that: In the second step, setting the initial filtering state and the initial covariance matrix includes: ; ; In the formula: Let be the initial state vector. The initial state covariance matrix; In the second step, the state prediction based on the recursive model, performed according to the accelerometer sampling frequency, to obtain the state prediction value and covariance matrix, includes: At each accelerometer sampling time, state prediction is performed using the following formula: ; In the formula: This is the predicted state value at the next accelerometer sampling time. This is the forward filtered estimate at the current accelerometer sampling time. The acceleration data collected by accelerometer (2) at the current accelerometer sampling time; The state prediction values include displacement prediction values, velocity prediction values, and acceleration prediction values; The state covariance matrix is updated synchronously using the following formula: ; In the formula: The state covariance matrix at the next accelerometer sampling time. Let be the state covariance matrix at the current accelerometer sampling time. Let be the process noise covariance matrix.
8. A method for monitoring the seismic safety of dams based on the fusion of BeiDou and accelerometers as described in claim 6, characterized in that: In the second step, the step of correcting the state prediction value and updating the covariance matrix based on the preprocessed displacement sequence according to the sampling frequency of the Beidou receiver includes: When the displacement observation value collected by the Beidou receiver (1) arrives, the Kalman filter gain is first calculated using the following formula: ; In the formula: for Kalman filter gain at time t. The sampling time for the BeiDou receiver. for The predicted state covariance matrix at time t. To design the matrix, The displacement observation noise covariance matrix; The predicted state value is then corrected using the following formula: ; In the formula: for The state estimate after time correction, for The predicted state value at time 10:
00. for Displacement observations collected by the Beidou receiver (1) at any given time; At the same time, the covariance matrix is updated using the following formula: ; In the formula: for The state covariance matrix after time correction, It is an identity matrix.
9. A method for monitoring the seismic safety of dams based on the fusion of BeiDou and accelerometers as described in claim 1 or 2, characterized in that: In the third step, extracting displacement time history, velocity time history, and acceleration time history estimates from the smoothed state estimation sequence at the same sampling frequency as the accelerometer includes: First, extract the displacement component, velocity component, and acceleration component at the current moment from the smoothed state vector at each moment of the smoothed state estimation sequence. The displacement components extracted at each time point are then arranged in chronological order to obtain displacement time history estimates; the velocity components extracted at each time point are arranged in chronological order to obtain velocity time history estimates; the acceleration components extracted at each time point are arranged in chronological order to obtain acceleration time history estimates; the sampling frequencies of the displacement time history, velocity time history, and acceleration time history estimates are all the same as the sampling frequency of the accelerometer.
10. A method for monitoring the seismic safety of dams based on the fusion of BeiDou and accelerometers as described in claim 1 or 2, characterized in that: In the fourth step, extracting the peak displacement, peak velocity, and peak acceleration from the estimated values of the displacement time history, velocity time history, and acceleration time history includes: For the displacement time history estimate, the absolute value of the displacement value at each moment is taken, and the maximum absolute value is taken as the peak displacement; for the velocity time history estimate, the absolute value of the velocity value at each moment is taken, and the maximum absolute value is taken as the peak velocity; for the acceleration time history estimate, the absolute value of the acceleration value at each moment is taken, and the maximum absolute value is taken as the peak acceleration.
Citation Information
Patent Citations
Reservoir dam operation safety sky-ground work intelligent sensing system and operation method
CN121430740A
