Breakwater foundation settlement direction real-time monitoring and early warning system and method

Abstract

The invention relates to the technical field of structural health monitoring and geotechnical engineering, and discloses a breakwater foundation settlement direction real-time monitoring and early warning system and method, and the system comprises a data collection module which is used for obtaining the measurement data of at least one sensor disposed on a breakwater and a foundation of the breakwater in real time; the storage module is used for storing a preset physics-based reduced-order model, and the reduced-order model represents the overall deformation rule of the breakwater and the foundation system through a group of core deformation modes; and the processing module is connected with the data acquisition module and the storage module, and the processing module comprises a data assimilation module which is used for fusing the measurement data and the order reduction model in real time. By arranging the deformation field reconstruction module, the problem that only discrete point information can be obtained through traditional monitoring is effectively solved, real-time and high-precision expansion from sparse measurement to a continuous overall deformation field is achieved, and a more visual and reliable basis is provided for comprehensively evaluating the safety state of a structure.

Description

Technical Field

[0001] The present invention relates to the field of structural health monitoring and geotechnical engineering technology, and in particular to a real-time monitoring and early warning system and method for breakwater foundation settlement direction. Background Art

[0002] As a vital port and coastal protection project, the structural safety and stability of breakwaters are crucial for ensuring the normal operation of ports and protecting coastal areas from wave erosion. Under the influence of multiple factors, including long-term cyclic loads such as waves and tides, as well as consolidation and settlement of the foundation itself, breakwater structures and their foundations are prone to uneven deformation, potentially even leading to catastrophic accidents such as soil liquefaction and sliding instability. Therefore, long-term, effective monitoring and early warning of breakwater foundation deformation, particularly its settlement patterns and trends, are critical for ensuring its long-term safe service and implementing preventive maintenance.

[0003] Currently, deformation monitoring of breakwater foundations primarily relies on traditional geotechnical engineering and geodetic methods. These methods typically deploy sensors such as Global Positioning System (GPS) receivers, levels, inclinometers, and settlement gauges on or within key locations on the breakwater. These sensors acquire displacement or tilt data at specific locations through periodic or continuous measurements. However, these monitoring methods, based on discrete physical measurement points, have inherent limitations. They only provide displacement or strain data from a limited number of measurement points and cannot reveal the continuous deformation patterns across the wider area between the measurement points, or even within the entire foundation. This sparse and localized information makes it difficult for engineering managers to grasp the overall deformation patterns of the structure from a macroscopic perspective, and therefore is incapable of identifying early signs of global failure, such as the formation and penetration of potential shear zones.

[0004] Early warning mechanisms based on this type of discrete data are typically simple, often employing fixed thresholds for the measured values ​​at individual measurement points. This approach often produces delayed warning signals, typically only triggering them after deformation has reached a significant level and irreversible structural damage has occurred. This approach lacks insight and foresight into the early stages of a catastrophic process, making it difficult to meet the practical needs of preventative maintenance.

[0005] While high-fidelity numerical simulation methods (such as the finite element method) can theoretically accurately calculate the deformation field of the entire system and reveal its underlying mechanical mechanisms, such models often contain tens of thousands or even millions of degrees of freedom, making them computationally expensive and requiring significant time for a single simulation. This limitation hinders their direct application in online monitoring and early warning systems, which require continuous, real-time feedback. Their application is often limited to design verification or offline post-disaster inversion analysis.

[0006] Existing technologies suffer from a widespread disconnect between "physical measurements" and "physical models": while the former can reflect the actual state, the information is incomplete and early warnings are delayed; while the latter, while providing complete information, cannot meet real-time requirements. Therefore, a new technology is urgently needed that can efficiently integrate sparse real-time measurement data with precise physical model knowledge to achieve real-time, accurate inversion of the continuous deformation field of the breakwater foundation, and based on this, provide intelligent early warning. Summary of the Invention

[0007] In response to the shortcomings of the existing technology, the present invention provides a real-time monitoring and early warning system for the settlement direction of the breakwater foundation, aiming to solve the technical problem that the existing technology is difficult to integrate sparse physical measurement data with complex structural physical models, thereby being unable to obtain the continuous three-dimensional deformation field of the breakwater foundation in real time and accurately and provide effective early warning.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: a real-time monitoring and early warning system for breakwater foundation settlement direction, comprising: a data acquisition module, configured to acquire measurement data from at least one sensor disposed on the breakwater and / or its foundation in real time; a storage module for storing a preset, physically based reduced-order model, wherein the reduced-order model characterizes the overall deformation law of the breakwater and foundation system through a set of core deformation modes; A processing module, connected to the data acquisition module and the storage module, comprising: a data assimilation module for fusing the measurement data with the reduced-order model in real time to optimally estimate modal coefficients representing the contribution of the core deformation mode at the current moment; A deformation field reconstruction module is used to reconstruct the continuous three-dimensional deformation field of the breakwater and foundation system in real time based on the optimal estimated modal coefficients and the core deformation mode; The early warning module is used to analyze the continuous three-dimensional deformation field to generate an early warning signal.

[0009] Preferably, the reduced-order model stored in the storage module is obtained by applying multiple load conditions based on a high-fidelity finite element model to generate a deformation snapshot matrix. , and perform intrinsic orthogonal decomposition on the deformation snapshot matrix to extract a set of optimal orthogonal basis vectors as the core deformation mode and form a basis matrix .

[0010] Preferably, the data assimilation module is based on a preset state space model and uses a Kalman filter for optimal estimation; wherein the state space model includes a state equation for describing the evolution law of the modal coefficients, which is in the form of: ; in, For the moment The modal coefficient vector of ; is the state transfer matrix; is the generalized force vector; is the control input matrix; is the process noise.

[0011] Preferably, the data acquisition module further specifically includes: At least one internal state sensor is used to measure the internal displacement or strain of the breakwater and foundation system, and its measurement data An update process for the data assimilation module; and At least one external load sensor is used to measure the external load acting on the breakwater, and the measurement data thereof is processed and used in the prediction process of the data assimilation module.

[0012] Preferably, the processing module is further configured to: convert the external load vector collected by the external load sensor into , the basis matrix formed by the core deformation mode , projected into the modal space to form the generalized force vector , which is calculated as follows: ; in, is the generalized force vector; is the transpose of the core deformation mode matrix; is the external load vector; And the generalized force vector As a feedforward control input to the state-space model.

[0013] Preferably, the early warning module is specifically configured to perform analysis in at least one of the following ways: Performing spatial differentiation on the continuous three-dimensional deformation field to obtain a global strain field, and generating an early warning based on the formation, expansion, and penetration trends of high shear strain regions in the strain field; or The energy distribution of the modal coefficients is monitored in real time, and an early warning is generated based on an abnormal and sustained growth trend of high-order modal energy representing local complex deformation.

[0014] Preferably, the processing module further comprises an adaptive correction module, and the adaptive correction module is used for online diagnosis of the mismatch between the reduced-order model and the actual physical state of the breakwater and foundation system.

[0015] Preferably, the adaptive correction module is to continuously monitor the innovation sequence in the Kalman filter update step The statistical characteristics of the mismatch are used to quantify the mismatch; wherein the innovation sequence is used to estimate the prior state Correction is performed to obtain the posterior state estimate , and its correction process follows the following equation: ; in, For Kalman; is the observation matrix.

[0016] Preferably, it further comprises an output module, the output module being used to Perform visual display; wherein the continuous three-dimensional deformation field The modal coefficients obtained by the optimal estimate are The basis matrix of the core deformation pattern The reconstruction equation is: .

[0017] The present invention also provides a real-time monitoring and early warning method for the settlement direction of a breakwater foundation, the method comprising the following steps: Pre-building and storing a physics-based reduced-order model, wherein the reduced-order model characterizes the overall deformation law of the breakwater and foundation system through a set of core deformation modes; acquiring in real time measurement data from at least one sensor disposed on the breakwater and / or its foundation; fusing the measured data with the reduced-order model in real time through a data assimilation process to optimally estimate modal coefficients representing the contribution of the core deformation mode at the current moment; reconstructing a continuous three-dimensional deformation field of the breakwater and foundation system in real time based on the optimally estimated modal coefficients and the core deformation mode; and An analysis is performed based on the continuous three-dimensional deformation field to generate a warning signal.

[0018] The present invention provides a real-time monitoring and early warning system and method for the direction of breakwater foundation settlement. It has the following beneficial effects: 1. The present invention provides a deformation field reconstruction module. It utilizes the low-dimensional modal coefficients optimally estimated in real time by the data assimilation module and linearly combines them with the core deformation patterns preset in the storage module that embody the overall deformation laws of the system. This overcomes the limitation of traditional monitoring methods that can only obtain information from discrete measurement points. It achieves real-time reconstruction of the three-dimensional overall deformation field of the breakwater and foundation system from sparse and incomplete sensor point measurements to a continuous, high-resolution, and real-time reconstruction. This makes it possible to comprehensively and intuitively grasp the macroscopic deformation posture and internal state of the structure.

[0019] 2. By setting up an intelligent early warning module, the present invention conducts an in-depth, physics-based secondary analysis of the reconstructed continuous three-dimensional deformation field, rather than relying on a simple single-point displacement threshold. This can generate more reliable and forward-looking early warning signals based on deep physical indicators such as the formation and penetration trend of potential shear failure surfaces in the strain field, or the abnormal growth of high-order modal energy that characterizes the precursors of local instability, significantly improving the intelligence level and timeliness of the early warning.

[0020] 3. By pre-building and storing a physics-based reduced-order model in the offline stage, the present invention transforms the online computational task from solving a high-fidelity finite element model with extremely high degrees of freedom and huge computational cost to a lightweight operation that only requires iterative solution of low-dimensional modal coefficients, thereby greatly reducing the complexity and time cost of online calculations, making real-time and continuous monitoring and analysis of large-scale geotechnical structures technically possible.

[0021] 4. The present invention sets a data assimilation module and preferably adopts a Kalman filter framework to dynamically and optimally integrate the reduced-order model prediction process containing physical laws with the sensor measurement update process containing real-world information, thereby effectively overcoming the error accumulation problem that may exist in a single physical model and the locality and noise interference problems that exist in a single sensor measurement, and obtaining a state estimation result that is more accurate and robust than relying solely on the model or solely on the measurement.

[0022] 5. The present invention further configures an adaptive correction module to continuously monitor the statistical characteristics of the residuals (i.e., innovation sequences) between model predictions and actual measurements during the data assimilation process online. This allows for real-time and quantitative diagnosis of the degree of mismatch between the preset reduced-order model and the actual physical state of the structure that has changed due to factors such as time lapse and disasters. This provides a clear scientific basis for model maintenance and updating, and ensures the long-term reliability and vitality of the monitoring and early warning system. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 Schematic diagram of the system architecture of the present invention; Figure 2 Schematic diagram of the method of the present invention. DETAILED DESCRIPTION

[0024] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the present specification. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0025] Please see the attached Figure 1 , an embodiment of the present invention provides a real-time monitoring and early warning system for breakwater foundation settlement direction, comprising; a data acquisition module, configured to acquire measurement data from at least one sensor disposed on the breakwater and / or its foundation in real time; a storage module for storing a preset, physically based reduced-order model, wherein the reduced-order model characterizes the overall deformation law of the breakwater and foundation system through a set of core deformation modes; A processing module, connected to the data acquisition module and the storage module, comprising: a data assimilation module for fusing the measurement data with the reduced-order model in real time to optimally estimate modal coefficients representing the contribution of the core deformation mode at the current moment; a deformation field reconstruction module for reconstructing a continuous three-dimensional deformation field of the breakwater and foundation system in real time based on the optimally estimated modal coefficients and the core deformation mode; The early warning module is used to analyze the continuous three-dimensional deformation field to generate an early warning signal.

[0026] In this embodiment, the data acquisition module serves as the perception tentacles of the monitoring and early warning system described in the present invention. It is a bridge connecting the breakwater entity in the physical world and the digital twin system. Its core function is to comprehensively, reliably and in real time obtain the measurement data of at least one sensor deployed on the breakwater and / or its foundation.

[0027] The measurement data is multi-source and heterogeneous, encompassing not only internal information reflecting the structural deformation state but also the external environmental factors driving structural deformation. Therefore, to enable the subsequent processing modules to perform effective model-data dual-driven analysis, the data acquisition module in this embodiment is preferably designed to capture and distinguish two key types of information.

[0028] The data acquisition module may include a submodule for acquiring internal state data, which is responsible for monitoring the response behavior of the breakwater and foundation system, that is, the actual internal deformation under various loads.

[0029] In order to accurately capture complex deformations, this submodule can integrate multiple types of sensors, including but not limited to: Global Navigation Satellite System (GNSS) receivers deployed on the top of the dike or other key structural points to obtain high-precision three-dimensional absolute displacement; A static leveling system arranged along a specific profile is used to accurately monitor the relative elevation changes between multiple measuring points caused by uneven settlement; An array of inclinometers buried in the foundation is used to obtain the lateral displacement distribution of the foundation soil at different depths, which is crucial for identifying potential sliding surfaces; and distributed fiber Bragg grating (FBG) sensor arrays embedded along the surface or inside the structure to obtain the strain distribution at key locations of the structure.

[0030] At each discrete computational time step of the system The data acquisition module collects the measurement values ​​of different types and formats from all the above internal state sensors, performs time alignment and formatting, and finally integrates them into a unified observation vector containing part of the real state information of the system at the current moment. The observation vector It is the most direct and critical basis for the subsequent data assimilation module to update the status (i.e., model correction).

[0031] Furthermore, to enhance the foresight and physical authenticity of the system prediction, the data acquisition module preferably includes a submodule for acquiring external load data. This submodule is responsible for monitoring the main external driving forces that cause breakwater deformation and providing real-time input for subsequent physical model predictions.

[0032] In a marine environment, such external loads are primarily dynamic water pressure caused by waves and tides. Therefore, the submodule may include a dynamic water pressure sensor array and a tide gauge deployed on the waterfront of the breakwater.

[0033] It is worth noting that since it is not feasible to achieve full sensor coverage on a wide structural surface in engineering, the external load sensors are usually sparsely distributed. Therefore, the data acquisition module or the processing module closely coupled with it must first reconstruct the approximately continuous external load vector acting on the entire water-facing surface based on the real-time pressure data of these sparse measurement points through a spatial interpolation algorithm (for example, a radial basis function network or Kriging interpolation method can be used). .

[0034] The external load vector It is not directly used for subsequent calculations, but needs to be converted into a driving term that is effective for the reduced-order model. To this end, the system uses the core deformation mode preset in the storage module (i.e., the POD basis matrix ), projecting the external load vector of the physical space into the low-dimensional modal space to calculate the generalized force vector The calculation process follows the following formula: ; in, is the generalized force vector; is the transpose of the core deformation mode matrix; is the external load vector; the obtained generalized force vector Each component of represents the driving force of the external total load on the corresponding core deformation mode. This vector will be input into the state prediction equation of the data assimilation module as a feedforward control term to guide the model to make a more physically accurate prediction of the deformation state at the next moment.

[0035] The data acquisition module in this embodiment is not a simple data collection unit, but an intelligent data pre-processing front end with clear division of labor. It provides two crucial inputs for subsequent processing modules by synchronously and classifiedly collecting internal states and external loads, and performing necessary reconstruction and projection transformation on external load data: observation vectors for model correction and , and the generalized force vector used for model prediction This design lays a solid data foundation for implementing the core idea of ​​dual-driven model and data for the entire system.

[0036] In this embodiment, the storage module constitutes the digital core and knowledge base of the monitoring and early warning system described in the present invention. Its key role is to provide a preset, physically-based reduced-order model for the system's online real-time calculations that can efficiently and accurately characterize the overall deformation law of the breakwater and foundation system.

[0037] It should be noted that directly using traditional high-fidelity finite element models (FEMs) for calculations in real-time monitoring systems is not feasible in engineering due to their extremely high degrees of freedom, enormous computational cost, and inability to meet real-time requirements. Therefore, in the offline phase before system operation, the present invention pre-generates a lightweight proxy model, the Reduced-Order Model (ROM), through a series of mathematical transformations. This model is then stored in the storage module. The core of this reduced-order model is to capture and express the vast majority of the deformation behavior of the original high-fidelity model through a limited set of core deformation modes.

[0038] Specifically, the generation process of the reduced-order model stored in the storage module preferably includes the following steps: First, a high-fidelity finite element model was established that accurately reproduced the actual physical process. This model was constructed based on the breakwater's geometry, material parameters, and geotechnical investigation reports. Its governing equations rigorously describe the consolidation process of the interaction between the solid skeleton and the pore fluid in a saturated porous medium. For example, the Biot consolidation theory can be used, which can be expressed in its basic form as follows: ; in, is the effective stress tensor of the soil skeleton; is the pore water pressure scalar; is the unit tensor; is the body force vector; is the permeability matrix; is the displacement vector of the soil skeleton. This high-fidelity model is the data source and physical basis for generating the reduced-order model.

[0039] Secondly, based on the above high-fidelity model, a deformation snapshot database covering various potential deformation states during the entire life cycle of the system is generated. This process is carried out by applying a series of parameterized load conditions and boundary conditions (for example, simulating different levels of wave loads, different combinations of tidal changes, uncertainty in foundation parameters, etc.) to the high-fidelity model and performing a large number of numerical simulation operations. The simulation results include all the The global displacement response vector of degrees of freedom As a "snapshot" sample, after completing After the simulation, all snapshot samples are combined by column to construct a high-dimensional snapshot matrix

[0040] ; in, is the snapshot matrix; is the snapshot matrix No. column vector. It is a high-dimensional vector with dimensions ; is the total number of snapshots; is the total number of nodal degrees of freedom of the original high-fidelity finite element model used to generate the deformation snapshot.

[0041] Then, the snapshot matrix Perform dimensionality reduction processing to extract the most important core deformation mode. In this embodiment, the Proper Orthogonal Decomposition (POD) method is preferably used. This method aims to find a set of optimal orthogonal bases so that the projection energy of the snapshot sample on this set of bases is most concentrated. Much larger than the number of snapshots In order to improve the calculation efficiency, the snapshot method is used here. The calculation steps of this method are as follows: first calculate the autocorrelation matrix of the snapshot matrix : ; This step converts the analysis of a huge matrix into a square matrix of much smaller dimensions by calculating the correlation between all snapshots. The analysis of , greatly reduces the computational complexity. Then, the autocorrelation matrix Solve its eigenvalue problem: ; in, The eigenvalue The diagonal matrix formed by the equations represents the "energy" contained in the corresponding mode or the contribution to the overall deformation. is its corresponding eigenvector matrix, which provides the coefficients for linearly combining the original snapshots into the final core deformation pattern.

[0042] According to the characteristic value Sort by size in descending order and select the first indivual The mode corresponding to the maximum eigenvalue is selected to ensure that most of the system deformation energy is captured. The modes constitute the core deformation mode that can capture the system deformation information with the highest efficiency. The core deformation modes are combined into a POD basis matrix , which is calculated by the following formula: ; in, and They are respectively The eigenvector matrix and diagonal matrix of the largest eigenvalues.

[0043] At this point, the core content actually stored in the storage module is the low-dimensional POD basis matrix that contains the inherent deformation law of the system. Its physical meaning is that the overall deformation field of the foundation at any time is complex and high-dimensional. , can be approximated with high precision as A linear combination of core deformation modes: ; in, is a low-dimensional vector that varies with time and is called the modal coefficient.

[0044] Therefore, by presetting the reduced-order model (i.e., the POD basis matrix) in the storage module ), the present invention successfully transforms a complex, high-dimensional vector The problem is transformed into a simple one to solve the low-dimensional modal coefficient vector The basis matrix This module is repeatedly called by the processing module during the online monitoring phase. It serves as the basis for the data assimilation module to construct the observation equation, a bridge for the deformation field reconstruction module to restore low-dimensional modal coefficients to a high-dimensional continuous deformation field, and a projection operator for calculating the effect of external loads on the driving effects of various modes. This module is the technical prerequisite for achieving real-time and efficient operation of the entire system.

[0045] In this embodiment, the processing module serves as the computational and decision-making hub of the monitoring and early warning system described herein. Its hardware can be a central processing unit (CPU), digital signal processor (DSP), or application-specific integrated circuit (ASIC), while its software is embodied as a series of collaborative algorithms. This module, bidirectionally connected to the data acquisition module and storage module, is responsible for executing the core technical solution of the present invention: the deep integration of offline physical knowledge with online measured data to achieve real-time, accurate understanding and early warning of the breakwater foundation's condition.

[0046] To achieve this goal, the processing module is preferably designed to include several logically independent and functionally progressive sub-modules, including a data assimilation module, a deformation field reconstruction module, an early warning module, and an adaptive correction module.

[0047] First, the processing module receives the real-time data stream from the data acquisition module, including the observation vector and external load data; at the same time, it calls the core reduced-order model, the POD basis matrix, from the storage module This information will serve as the basic input for subsequent calculations.

[0048] The data assimilation module is the core hub connecting the physical model and the real measurement data. Its technical goal is to dynamically fuse the reduced-order model provided by the storage module, which contains the system's prior physical knowledge, with the sparse measurement data provided by the data acquisition module, which reflects the current real state of the system, through a stable and efficient algorithm framework. Finally, the modal coefficient vector representing the contribution of the core deformation mode at the current moment is solved in real time in an optimal estimation manner. .

[0049] Specifically, to achieve the aforementioned technical effects, the data assimilation module in this embodiment preferably uses the Kalman filter as its core algorithm architecture. This choice is based on the Kalman filter's optimality and computational efficiency when handling linear Gaussian system state estimation problems. The entire data assimilation process is constructed within a state space framework, and real-time tracking of modal coefficients is achieved through continuous, cyclical iterations of the "prediction" and "update" steps.

[0050] Step 1: Prediction This step occurs at each time step The core task is to execute at the beginning of the The optimal state estimation of the current moment is completely dependent on the preset physical model. The state of the system. This process does not involve any actual measurement at the current moment, but is purely based on the deduction of model rules. First, the module predicts the state. It calls the posterior state estimate of the previous moment. , and use the state transfer matrix solidified in the storage module It is forward propagated in time. At the same time, it integrates the generalized force vector obtained by processing the external load sensor As feedforward control input to improve the accuracy of prediction. This process calculates the prior state estimate of the current moment , its mathematical expression is: ; in: is the prior state estimation vector at time k; For the moment The posterior state estimation vector is the final result of the previous iteration; is the state transfer matrix, which describes the modal coefficients from time arrive The inherent evolutionary laws of For the moment The generalized force vector is the projection of the external load in the modal space; To control the input matrix, it applies the influence of the generalized forces to the state vector.

[0051] Secondly, the module performs error covariance prediction. While predicting the state, the uncertainty of its estimate will also increase due to the passage of time and the inherent uncertainty of the model. This step is to quantify the growth of this uncertainty and calculate the prior error covariance matrix The calculation formula is as follows: ; in: For the moment The prior error covariance matrix of ; For the moment The posterior error covariance matrix of ; is the process noise covariance matrix, which represents the errors and uncertainties in the physical model itself that cannot be accurately modeled.

[0052] Step 2: Update After the prediction step is completed, once the data acquisition module provides the current time Real sensor measurement data , the update step starts immediately. The core task of this step is to use this real external information to correct the prior estimate given by the prediction step, so as to obtain a more accurate posterior estimate.

[0053] First, the module calculates a crucial weight matrix - Kalman gain The optimality of this gain matrix is ​​reflected in its ability to perfectly balance the credibility of the model predictions (given by Reflected) and the credibility of the measurement data (reflected by the measurement noise covariance The calculation formula is: ; in, is the Kalman gain matrix; is the observation matrix, which establishes the state vector (modal coefficients ) and the directly observable measurement vector (sensor reading ), that is, ; is the measurement noise covariance matrix, which represents the statistical characteristics of the measurement error of the sensor itself.

[0054] Secondly, the module performs the core state update operation. It uses the Kalman gain just calculated to update the prior state estimate Correction is performed to obtain the posterior state estimate after integrating the measurement information This is the final output of the data assimilation module at the current moment, representing the optimal estimate of the modal coefficients. Its mathematical expression is: ; In this formula This is called the innovation sequence or measurement residual. It reflects the difference between the actual measurement and the model prediction and is the fundamental driving force for state correction. Finally, to complete the iterative closed loop and provide input for the prediction of the next time step, the module also needs to update the error covariance. Due to the introduction of new measurement information, the uncertainty of the system state should be reduced. This step calculates the posterior error covariance matrix using the following formula , to quantify this reduction in uncertainty: ; in, is the identity matrix.

[0055] Through the above-mentioned "prediction-update" cycle, the data assimilation module achieves the dynamic optimal fusion of physical model predictions and real data measurements, thereby being able to continuously and stably output high-precision modal coefficient vectors, providing reliable data input for the subsequent deformation field reconstruction and early warning analysis modules.

[0056] The deformation field reconstruction module is a key functional unit within the processing module, connecting the previous and subsequent modules. Its primary function is to receive and interpret the optimal estimation results output by the data assimilation module. Based on these results and the reduced-order model pre-set in the storage module, it efficiently and in real time transforms the abstract low-dimensional state description into a physically intuitive and complete continuous three-dimensional deformation field of the breakwater and foundation system.

[0057] The deformation field reconstruction module acts as a "decoder" and performs the inverse operation of the offline reduction process. It receives two main inputs: One is from the data assimilation module at time The output a posteriori state estimate, that is, the optimal modal coefficient vector . This vector has a very low dimension ( ), but each of its components accurately quantifies the contribution or "weight" of a corresponding core deformation mode to the total deformation of the system at the current moment.

[0058] The second is the POD basis matrix called from the storage module as the core of the reduced-order model Each column of this matrix is ​​a high-dimensional vector that describes a basic deformation form. The operating mechanism of this module is to utilize the basic principle of reduced-order model, that is, the complex deformation field at any moment can be composed of a set of core deformation modes linearly superimposed. Therefore, the reconstruction process is greatly simplified computationally, requiring only one matrix and vector multiplication operation. Its mathematical expression is as follows: ; in, It is provided by the data assimilation module and contains The input column vector of modal coefficients; is the POD basis matrix called from the storage module, where is the total number of degrees of freedom of the original high-fidelity finite element model, is the number of core deformation modes. List Represents the core deformation modes; This module calculates and outputs a high-dimensional displacement vector. The dimensions of this vector are identical to those of the original high-fidelity finite element model, and each component corresponds to the displacement of a node in the model in one degree of freedom (e.g., X, Y, or Z).

[0059] By performing the above operations, the module can instantly convert the output of the data assimilation module into abstract modal coefficients, “unfolded” or “upsampled” to cover the entire breakwater and its foundation Complete deformation information of each degree of freedom.

[0060] The deformation field reconstruction module plays a crucial role. It represents a significant leap forward from sparse, discrete sensor point measurements to continuous, global understanding of the body's deformation state. Traditional monitoring methods can only obtain displacement values ​​at a limited number of measurement points, but are unable to determine deformation conditions in the areas between these points or deep within the structure. This module overcomes this limitation, enabling the generation of a high-resolution, visual 3D deformation cloud map.

[0061] More importantly, the module outputs a continuous three-dimensional deformation field This isn't just for the final visualization; it also serves as the essential data foundation for subsequent in-depth safety assessments and intelligent early warnings. Only with a continuous displacement field can further calculations (for example, spatial derivatives) be performed to obtain global strain, stress, or rotation fields, which are more indicative of the structural safety status. Therefore, the deformation field reconstruction module is an indispensable bridge connecting "state estimation" and "intelligent early warning." Its efficiency and accuracy are prerequisites for ensuring the entire system can conduct real-time, in-depth analysis.

[0062] The early warning module is the final decision-making and output unit of the processing module. Its technical purpose is to provide real-time, intelligent assessment and classification of the safety status of monitored objects. This module receives the high-resolution, continuous three-dimensional deformation field generated by the deformation field reconstruction module and performs in-depth, physics-based analysis to generate more reliable and forward-looking early warning signals that surpass traditional single-point threshold methods.

[0063] The input of the early warning module is the deformation field reconstruction module at each time step. Calculated global displacement vector covering the entire breakwater and its foundation It should be emphasized that only based on the displacement vector Setting thresholds for individual node displacement values ​​for early warning is not sufficient to reveal the complex failure mechanism within the structure. Therefore, the early warning module of the present invention is preferably designed to use at least one or more of the following analysis strategies to generate early warning signals. A preferred analysis method is field derivative analysis. This method aims to extract physical derivatives that are more indicative of structural stability from the continuous displacement field. The early warning module is embedded with a numerical differentiation algorithm that can be used based on the input displacement field. Calculate the strain tensor field of the entire domain Under the small deformation assumption, the relationship between the strain tensor and the displacement gradient can be expressed as: ; in, is the displacement vector field After obtaining the strain field, the module can further calculate key indicators such as maximum principal strain, volumetric strain or shear strain.

[0064] In soil mechanics, foundation instability often manifests as shear failure. Therefore, a core function of this early warning module is to automatically identify and track the evolution of high shear strain areas. The module has a preset shear strain safety threshold based on material properties or design specifications. During operation, it continuously compares the calculated global shear strain field with this threshold to identify the formation, expansion, and penetration trends of high shear strain areas. When the range, connectivity, or peak value of the high shear strain area meets the preset danger criteria (for example, the formation of a continuous potential sliding zone), the module determines that the system is at risk of instability and generates a warning signal of the corresponding level.

[0065] As a supplement to the above analysis, another preferred analysis method is modal energy analysis. This method does not directly analyze the displacement field, but starts from the intrinsic mode of system deformation in order to capture earlier signs of instability. For this purpose, this module can be directly connected to the data assimilation module to obtain the optimal modal coefficient vector output by it. Theoretically, the total deformation energy of the system is proportional to the sum of the squares of the modal coefficients, and the energy allocated to the first Core deformation modes Energy on It is proportional to the square of its corresponding modal coefficient: ; The early warning module is equipped with an algorithm that calculates the energy of each mode and its contribution to the total energy in real time at each time step. Typically, the vast majority of a system's deformation energy is concentrated in a few low-order modes that represent overall deformation, while the energy of higher-order modes that represent complex local deformations is minimal.

[0066] The core early warning logic of this module is to continuously monitor the distribution pattern of this modal energy. When an abnormal and continuous growth trend is detected in the energy proportion of one or several high-order modes, even if the total deformation at that moment (i.e., the displacement readings of all measuring points) is still within a safe range, the module will identify it as an important early warning signal. This is because the transfer of energy from low-order modes to high-order modes often indicates that localized, more complex deformation mechanisms are being nurtured within the structure, such as local damage or the initiation of nonlinear behavior, which are precursors to macroscopic instability. When the energy proportion of a high-order mode or its growth rate exceeds the dynamic baseline set based on historical data statistics, the module generates a high-priority early warning signal.

[0067] In summary, the early warning module in this embodiment, through the organic combination of field-derived quantity analysis and modal energy analysis, constructs a multi-dimensional, multi-layered comprehensive early warning system, spanning from macroscopic geometric forms to intrinsic energy patterns. This system not only identifies impending dangerous conditions but also captures subtle, structural trends, providing valuable predictive information for the safe operation and maintenance of breakwaters, thereby achieving a shift from "passive response" to "active forecasting."

[0068] To ensure the long-term effectiveness of the system of the present invention, the processing module preferably also integrates an adaptive correction module. The function of this module is to diagnose the mismatch between the reduced-order model in the storage module and the real foundation that undergoes physical changes over time. This module continuously monitors the innovation sequence, a by-product generated by the data assimilation module (Kalman filter) in the update step. Diagnosis is achieved by measuring the statistical properties of the reduced-order model (i.e., the difference between the actual measurement and the model-predicted measurement). In the ideal state where the model and reality completely match, the innovation sequence should be white noise with zero mean. As the mismatch increases, the statistical properties of the sequence will deviate from the ideal state. This module quantifies the mismatch by calculating statistics such as the normalized innovation square (NIS) and performing a chi-square test on them. When the mismatch indicator continuously and significantly exceeds its theoretical confidence interval within the set time window, the module determines that the current reduced-order model has failed and generates a correction signal to prompt the operation and maintenance personnel to update the reduced-order model, thereby achieving closed-loop adaptive maintenance of the system.

[0069] See also Figure 2 The present invention also provides a real-time monitoring and early warning method for the settlement direction of a breakwater foundation, the method comprising the following steps: Pre-building and storing a physics-based reduced-order model, wherein the reduced-order model characterizes the overall deformation law of the breakwater and foundation system through a set of core deformation modes; acquiring in real time measurement data from at least one sensor disposed on the breakwater and / or its foundation; fusing the measured data with the reduced-order model in real time through a data assimilation process to optimally estimate modal coefficients representing the contribution of the core deformation mode at the current moment; reconstructing a continuous three-dimensional deformation field of the breakwater and foundation system in real time based on the optimally estimated modal coefficients and the core deformation mode; and performing analysis based on the continuous three-dimensional deformation field to generate an early warning signal; The method of this embodiment can be used to execute the above system embodiment. Its principles and technical effects are similar and will not be described in detail here.

[0070] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A real-time monitoring and early warning system for breakwater foundation settlement direction, characterized in that: include: a data acquisition module, configured to acquire measurement data from at least one sensor disposed on the breakwater and / or its foundation in real time; a storage module for storing a preset, physically based reduced-order model, wherein the reduced-order model characterizes the overall deformation law of the breakwater and foundation system through a set of core deformation modes; A processing module, connected to the data acquisition module and the storage module, comprising: a data assimilation module for fusing the measurement data with the reduced-order model in real time to optimally estimate modal coefficients representing the contribution of the core deformation mode at the current moment; A deformation field reconstruction module is used to reconstruct the continuous three-dimensional deformation field of the breakwater and foundation system in real time based on the optimal estimated modal coefficients and the core deformation mode; The early warning module is used to analyze the continuous three-dimensional deformation field to generate an early warning signal.

2. A real-time monitoring and early warning system for breakwater foundation settlement direction according to claim 1, characterized in that: The reduced-order model stored in the storage module is obtained by applying multiple load cases based on a high-fidelity finite element model to generate a deformation snapshot matrix. , and perform intrinsic orthogonal decomposition on the deformation snapshot matrix to extract a set of optimal orthogonal basis vectors as the core deformation mode and form a basis matrix .

3. A real-time monitoring and early warning system for breakwater foundation settlement direction according to claim 1, characterized in that: The data assimilation module is based on a preset state space model and uses a Kalman filter for optimal estimation; wherein the state space model includes a state equation for describing the evolution law of the modal coefficients, which is in the form of: ； in, For the moment The modal coefficient vector of ; is the state transfer matrix; is the generalized force vector; is the control input matrix; is the process noise.

4. A real-time monitoring and early warning system for breakwater foundation settlement direction according to claim 3, characterized in that: The data acquisition module further specifically includes: At least one internal state sensor is used to measure the internal displacement or strain of the breakwater and foundation system, and its measurement data An update process for the data assimilation module; and At least one external load sensor is used to measure the external load acting on the breakwater, and the measurement data thereof is processed and used in the prediction process of the data assimilation module.

5. A real-time monitoring and early warning system for breakwater foundation settlement direction according to claim 4, characterized in that: The processing module is further configured to: convert the external load vector collected by the external load sensor into , the basis matrix formed by the core deformation mode , projected into the modal space to form the generalized force vector , which is calculated as follows: ； in, is the generalized force vector; is the transpose of the core deformation mode matrix; is the external load vector; And the generalized force vector As a feedforward control input to the state-space model.

6. A real-time monitoring and early warning system for breakwater foundation settlement direction according to claim 1, characterized in that: The early warning module is specifically configured to perform analysis in at least one of the following ways: performing spatial differentiation of the continuous three-dimensional deformation field to obtain a global strain field, and generating an early warning based on the formation, expansion, and penetration trends of high shear strain regions in the strain field; or The energy distribution of the modal coefficients is monitored in real time, and an early warning is generated based on an abnormal and sustained growth trend of high-order modal energy representing local complex deformation.

7. A real-time monitoring and early warning system for breakwater foundation settlement direction according to claim 3, characterized in that: The processing module further includes an adaptive correction module for online diagnosis of a mismatch between the reduced-order model and an actual physical state of the breakwater and foundation system.

8. A real-time monitoring and early warning system for breakwater foundation settlement direction according to claim 7, characterized in that: The adaptive correction module is to continuously monitor the innovation sequence in the Kalman filter update step The statistical characteristics of the mismatch are used to quantify the mismatch; wherein the innovation sequence is used to estimate the prior state Correction is performed to obtain the posterior state estimate , and its correction process follows the following equation: ； in, For Kalman; is the observation matrix.

9. A real-time monitoring and early warning system for breakwater foundation settlement direction according to claim 1, characterized in that: Also includes an output module, the output module is used to Perform visual display; wherein the continuous three-dimensional deformation field The modal coefficients obtained by the optimal estimate are The basis matrix of the core deformation pattern The reconstruction equation is: 。 10. A real-time monitoring and early warning method for breakwater foundation settlement direction, characterized in that: A system for real-time monitoring and early warning of breakwater foundation settlement direction according to any one of claims 1 to 9, comprising the following steps: Pre-building and storing a physics-based reduced-order model, wherein the reduced-order model characterizes the overall deformation law of the breakwater and foundation system through a set of core deformation modes; acquiring in real time measurement data from at least one sensor disposed on the breakwater and / or its foundation; fusing the measured data with the reduced-order model in real time through a data assimilation process to optimally estimate modal coefficients representing the contribution of the core deformation mode at the current moment; reconstructing a continuous three-dimensional deformation field of the breakwater and foundation system in real time based on the optimally estimated modal coefficients and the core deformation mode; and An analysis is performed based on the continuous three-dimensional deformation field to generate a warning signal.

Images