Self-adaptive temporary support structure intelligent monitoring system and method based on digital twinning

Abstract

The invention relates to a self-adaptive temporary support structure intelligent monitoring system and method based on digital twinning, the system comprises a multi-source sensing network, a digital twinning model engine, a self-adaptive regulation and control decision center and a hydraulic execution mechanism, and the multi-source sensing network is arranged at a temporary support key node; the digital twin model engine constructs a high-fidelity virtual mirror image through BIM geometric modeling and finite element mechanical simulation, and real-time data is fused to drive dynamic prediction; the adaptive regulation and control decision center generates a regulation and control instruction based on an LSTM deformation prediction algorithm and a reinforcement learning strategy; and the hydraulic executing mechanism dynamically adjusts the supporting force according to the regulation and control instruction, and structural deformation self-compensation is achieved. According to the system and the method, rapid deformation early warning response and high-precision supporting force regulation and control can be realized aiming at the bracket instability risk of a deep water area and a large-span overhanging part of a river channel under the working conditions of concrete pouring and high water level in a flood season, and the construction safety level is remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of water conservancy engineering construction safety technology, specifically relating to an adaptive temporary support structure intelligent monitoring system and method based on digital twins. Background Technology

[0002] In the field of water conservancy engineering construction, the safety of temporary support structures is directly related to the overall progress of the project and the safety of personnel. Especially in complex conditions such as deep-water areas of rivers and large-span cantilever sections, the support structures need to withstand multiple dynamic loads, including concrete pouring loads and high water level impacts during the flood season. Traditional monitoring methods mainly rely on manual inspections and static model analysis, which have significant limitations. First, traditional sensing systems mostly use single-parameter monitoring, lacking the ability to collaboratively perceive multiple physical fields. Moreover, heterogeneous sensor data, due to inconsistent spatiotemporal references and large differences in accuracy, makes it difficult to construct a high-fidelity structural state model through static fusion methods. Second, existing monitoring models are mostly based on idealized mechanical assumptions, failing to effectively capture the nonlinear and spatiotemporal correlation characteristics of load changes and structural responses during construction, resulting in insufficient deformation prediction accuracy and difficulty in achieving early risk warning. In addition, traditional control strategies rely on manual experience and judgment, resulting in strong response lags and a lack of a closed-loop control mechanism from data perception to proactive intervention, making it impossible to achieve adaptive dynamic compensation of support forces when instability risks occur. Although digital twin technology has been introduced into the field of engineering monitoring in recent years, its application in temporary support structures still suffers from problems such as poor synchronization between virtual and real models, insufficient adaptability of dynamic prediction algorithms, and disconnect between decision-making and control links. There is an urgent need to improve the real-time performance, accuracy, and autonomy of the system through deep fusion of multi-source data, integration of intelligent algorithms, and innovation of closed-loop control mechanisms. Summary of the Invention

[0003] To address the shortcomings of existing methods, such as insufficient accuracy in multi-source data fusion, weak ability to capture spatiotemporal dynamic features, lagging deformation prediction, and lack of proactive control mechanisms, this invention aims to provide an adaptive intelligent monitoring system and method for temporary support structures based on digital twins, thereby overcoming the traditional technical bottlenecks in the safety monitoring of temporary support structures during water conservancy engineering construction. This invention integrates multi-source sensor networks, high-fidelity virtual modeling, a hybrid prediction algorithm combining LSTM and ARIMA, and a hydraulic actuator to construct a closed-loop control system encompassing "perception-prediction-decision-execution," achieving a leap from passive early warning to proactive deformation compensation. The ultimate goal is to significantly improve the safety and reliability of temporary support structures under complex conditions such as concrete pouring and high water levels during flood season, promoting the development of intelligent and adaptive monitoring in water conservancy engineering construction.

[0004] To achieve the above objectives, the present invention adopts the following technical solution: an intelligent monitoring system for adaptive temporary support structures based on digital twins, comprising a multi-source sensor network, a digital twin model engine, an adaptive control decision center, and a hydraulic actuator. The multi-source sensor network is deployed at key nodes of the temporary support. The digital twin model engine constructs a high-fidelity virtual image through BIM geometric modeling and finite element mechanical simulation, and integrates real-time data to drive dynamic prediction. The adaptive control decision center generates control commands based on the LSTM deformation prediction algorithm and reinforcement learning strategy. The hydraulic actuator dynamically adjusts the support force according to the control commands to achieve self-compensation of structural deformation.

[0005] Furthermore, the multi-source sensing network includes stress sensors, displacement gauges, inclinometers, and hydraulic gauges, used to collect stress, displacement, inclinometer, and hydraulic pressure data in real time.

[0006] This invention also provides an intelligent monitoring method for adaptive temporary support structures based on digital twins, implemented based on the above system, comprising the following steps:

[0007] S1: A multi-source sensor network is deployed at the stress concentration points of the temporary support structure. The multi-source sensor network collects key parameters in real time at a high frequency and aggregates the raw data to the central processing unit through a wireless transmission protocol.

[0008] S2: The digital twin model engine constructs an accurate geometric model of the temporary support structure based on BIM technology, forming a high-fidelity virtual mirror; the digital twin model engine integrates the real-time sensing data from step S1, and uses Kalman filtering and its extended form algorithm to reduce the difference between the predicted value generated by the digital twin model and the actual measurement value of the sensor.

[0009] S3: The adaptive control decision center receives the prediction results from step S2 and applies the autoregressive integral moving average model and LSTM neural network algorithm to analyze the deformation trend; the adaptive control decision center generates control instructions and optimizes the anti-deformation strategy based on historical data and real-time feedback; the decision output includes early warning signals and control parameters to ensure adaptive decision-making in unstable risk scenarios and improve the level of construction safety.

[0010] S4: The hydraulic actuator executes the control command generated in step S3, and dynamically adjusts the support force through the hydraulic cylinder or servo system to achieve self-compensation of structural deformation; the actuator applies a force feedback mechanism at key nodes to monitor the change of support force in real time, and feeds the adjustment result back to the multi-source sensor network through a closed-loop system.

[0011] Furthermore, a three-dimensional geometric model of the temporary support structure is created using Building Information Modeling (BIM). This three-dimensional geometric model integrates geometric information and non-geometric attribute parameters. It is assumed that the support structure is discretized into n key nodes, and its initial geometric state is determined by a node coordinate matrix. express:

[0012]

[0013] In the formula, Represents the three-dimensional coordinates of the i-th node;

[0014] For linear statics problems, the core governing equations are the system equilibrium equations:

[0015]

[0016] In the formula, K is the global stiffness matrix; U is the nodal displacement vector. , represents the displacement of each node in three spatial directions; F is the node load vector.

[0017] Furthermore, to achieve synchronization between the virtual entity and the physical entity, real-time sensor data streams are introduced. Through data assimilation, the digital model is continuously corrected to make it infinitely close to the real state of the physical entity. The digital twin engine, as the data processing hub, continuously receives real-time data streams transmitted back from the multi-source sensor network in the previous stage. The goal is to find a set of optimal corrections to the model parameters through calibration. This makes the predicted values ​​generated by the corrected digital model... Compared with the actual measured value of the sensor The difference between them is minimized; the calibration process is defined as:

[0018]

[0019] In the formula, Represents the initial parameter set of the model; This represents the obtained optimal parameter correction amount. ;

[0020] Real-time calibration is achieved using Kalman filtering and its extended forms. In the prediction step, Kalman filtering and its extended forms predict the system state at the current moment based on the system's state at the previous time step and the known physical dynamics model.

[0021]

[0022] In the formula, This represents the prior estimate of the system state at time k, before fusing the sensor observations at time k. The state transition matrix describes how the system naturally evolves from the state at time k-1 to the state at time k. This represents the posterior estimate of the system state at time k-1, obtained after fusing the sensor observations from time k-1. To control the input matrix, it is described how the force applied by the hydraulic actuator changes the state of the support; Let be the control vector, representing the known external control input at time k;

[0023] At the same time, the uncertainty of the estimate is predicted:

[0024]

[0025] In the formula, Estimating the state from prior time The corresponding error covariance matrix; The posterior estimated state of the previous time step The error covariance matrix; The process noise covariance matrix;

[0026] Then it proceeds to the update step, when new sensor observations... Upon arrival, calculate the Kalman gain. :

[0027]

[0028] In the formula, The observation matrix reflects the relationship between the internal state x of the system and the sensor observation value z; To observe the noise covariance matrix;

[0029] Then, the posterior estimate of the system state is updated using the Kalman gain and the new observations:

[0030]

[0031] In the formula, This represents the actual observation value of the sensor at time k; This represents the difference between the actual observations from the sensor and the observations predicted by the model.

[0032] Synchronously update the posterior estimate of the error covariance:

[0033]

[0034] In the formula, It is an identity matrix.

[0035] Furthermore, let the displacement of a measuring point be... Time series analysis was performed on continuously collected displacement information to analyze historical data sequences. Input the autoregressive integral moving average (ARIMA) model, which is expressed as follows:

[0036]

[0037] In the formula, For lag operators, These represent the order of autoregression, the order of differencing, and the order of moving average, respectively. and For model parameters, This is the white noise term;

[0038] ARIMA achieves prediction by capturing the autocorrelation and difference stationarity of time series. Furthermore, to handle more complex nonlinear relationships and long-term dependencies, the system employs a Long Short-Term Memory (LSTM) network, the calculation formula of which is as follows:

[0039]

[0040] In the formula, This is the forgetting gate vector, where each element has a value between 0 and 1, representing the cell state. The degree to which the corresponding information is retained; 1 means "completely retained", and 0 means "completely discarded"; Use the Sigmoid activation function; Here is the weight matrix for the forget gate; For the splicing operation, the hidden state from the previous moment is removed. With the input vector at the current time step Connect them into a longer vector; This is the bias vector for the forget gate; The input gate vector; The candidate cell state vector is generated by the tanh function and contains potential new information generated based on the current input and past states. This refers to the cell state updated at the current moment, i.e., long-term memory; This represents the cell state at the previous moment; The Hadamard product is the element-wise multiplication of vectors. Forgetting operation; For memory operations; This is the output gate vector; The weight matrix is ​​a learnable matrix; It is a learnable bias vector; The hidden state at the current moment is the output of the LSTM unit; First, the updated cell state is scaled to between -1 and 1 using the tanh function;

[0041] The predicted key point displacement value at a future time point (t+h) Compared with the maximum safe displacement threshold allowed by structural design specifications Real-time comparisons are performed, and the displacement risk index is defined as follows: Based on the risk index, the system implements dynamic hierarchical early warning, providing corresponding levels of alerts for decision-making.

[0042] Furthermore, the specific rules for the dynamic hierarchical early warning are defined as follows:

[0043] Yellow Alert (Attention Level): When When the value is >0.8, it is triggered; this signal indicates that the structural deformation has entered a period of accelerated development. Although there is no immediate danger, the risk has increased significantly. It prompts engineers to increase the monitoring frequency and notify the decision-making center to start contingency plan analysis, which may require limiting the construction loading rate.

[0044] Red Alert (Action Level): When the value exceeds 0.95, it triggers the highest level of alarm, indicating that the structure is extremely close to its load-bearing limit and the risk of instability is extremely high. The decision-making center must immediately activate the pre-established control strategy, issue instructions to the hydraulic actuators, and actively intervene to compensate for deformation and eliminate risks. At the same time, the system will recommend that personnel be evacuated or related operations be stopped.

[0045] Furthermore, it receives the control command set issued by the adaptive control decision center, and through its internal hydraulic cylinder or high-precision servo system, it converts the digital control parameters into actual mechanical action applied to the temporary support structure, thereby actively and dynamically offsetting the deformation of the structure caused by external loads.

[0046] set up This is the control command vector issued by the decision-making center at time k, whose elements include the target support force. The actuator drives the hydraulic system according to control commands, and its dynamic response process is described by the following simplified discrete state-space equation:

[0047]

[0048] In the formula, This refers to the internal state vector of the actuator system, including physical quantities such as the hydraulic cylinder piston position and oil chamber pressure. The system output is the actual supporting force applied to the structure. Matrices A, B, and C are determined by the dynamic characteristics of the actuator itself; the control objective of the system is to maximize the actual output. Quickly and accurately track instructions That is, to achieve .

[0049] Furthermore, to ensure control accuracy, the hydraulic actuator integrates high-precision force and displacement sensors to form a local closed-loop feedback control; the system monitors the output force in real time. The actual deformation of the structure is monitored and the data is fed back to the multi-source sensor network. This feedback mechanism enables the system to form a continuously iterative intelligent closed loop: the feedback data is sent to the digital twin model engine and used by the Kalman filter algorithm for model state updates and parameter calibration.

[0050] Hydraulic actuators thus constitute the execution terminal connecting the information space and the physical world. By transforming intelligent commands into precise mechanical interventions and using force feedback to close the global control loop, they ultimately ensure the safety and controllability of temporary support structures under complex construction conditions.

[0051] Compared with the prior art, the present invention has the following beneficial effects:

[0052] 1. Achieving high-precision fusion and virtual-real synchronization of multi-source data enhances the reliability of state perception. Traditional monitoring systems rely on data from a single sensor, making it difficult to comprehensively reflect the true state of the supporting structure under complex working conditions. This invention deploys a multi-source sensor network including stress, displacement, tilt angle, and hydraulic pressure gauges, and employs Kalman filtering and its extended algorithms for real-time data assimilation. This algorithm dynamically balances the reliability of theoretical model predictions and sensor measurements, continuously optimizing the parameters of the digital twin model to infinitely approximate the true state of the physical entity. This high-precision data fusion and virtual-real synchronization mechanism fundamentally solves the problem of state misjudgment caused by single data or noise interference in traditional methods, providing highly reliable input for subsequent prediction and decision-making, and ensuring the system's perception reliability.

[0053] 2. Establishing a closed-loop control system encompassing perception, prediction, decision-making, and execution, enhancing the intelligence level of regulation. The core advantage of this system lies in its complete adaptive closed loop. The system uses a digital twin model engine for state estimation and prediction. The adaptive regulation decision center generates regulation commands based on the prediction results, which are ultimately executed precisely by the hydraulic actuator. This actuator is not a simple actuator but integrates a force feedback mechanism, enabling real-time feedback of support force adjustments back to the sensor network, forming a closed loop. This end-to-end intelligent control, characterized by "data-driven decision-making and execution feedback optimization," significantly reduces reliance on human experience, allowing the system to proactively and quickly respond to risks, achieving a qualitative leap from passive monitoring to proactive adaptive regulation.

[0054] 3. Significantly improves construction safety and efficiency, with substantial social and economic value. This invention directly addresses the risk of support instability in deep-water river sections and large-span cantilever sections under extreme conditions such as concrete pouring and high water levels during flood season. Through high-precision early warning and active deformation compensation, it effectively prevents catastrophic accidents such as support collapse, ensuring the safety of construction personnel and demonstrating significant social benefits. Simultaneously, by avoiding project delays and resource waste caused by accidents, and reducing conservative measures taken due to safety concerns, the system significantly improves construction efficiency and reduces overall project costs, exhibiting substantial economic benefits. This system provides an innovative technical solution for construction safety in high-risk fields such as water conservancy projects, with broad prospects for widespread application. Attached Figure Description

[0055] Figure 1 This is a block diagram illustrating the implementation principle of the intelligent monitoring system for adaptive temporary support structures based on digital twins provided in this embodiment of the invention.

[0056] Figure 2 This is a schematic diagram of the structure of the intelligent monitoring system for adaptive temporary support structures based on digital twins provided in an embodiment of the present invention;

[0057] Figure 3 This is a comparison chart of water pressure data at a water depth of 5 meters in an embodiment of the present invention. Detailed Implementation

[0058] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0059] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0060] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0061] like Figure 1-2As shown, this embodiment provides an intelligent monitoring system for adaptive temporary support structures based on digital twins, including a multi-source sensor network 1, a digital twin model engine 6, an adaptive control decision center 7, and a hydraulic actuator 8. The multi-source sensor network 1 is deployed at key nodes of the temporary support and includes stress sensors 2, displacement gauges 3, inclinometers 4, and pressure gauges 5, used to collect stress, displacement, inclinometer, and pressure data in real time. The digital twin model engine 6 constructs a high-fidelity virtual image through BIM geometric modeling and finite element mechanical simulation, integrating real-time data to drive dynamic prediction. The adaptive control decision center 7 generates control commands based on the LSTM deformation prediction algorithm and reinforcement learning strategy. The hydraulic actuator 8 dynamically adjusts the support force according to the control commands, achieving self-compensation for structural deformation.

[0062] This embodiment also provides an intelligent monitoring method for adaptive temporary support structures based on digital twins, implemented based on the above system, including the following steps:

[0063] S1: A multi-source sensor network 1 is deployed at the stress concentration points of the temporary support structure. The multi-source sensor network 1 collects key parameters in real time at a high frequency and aggregates the raw data to the central processing unit through a wireless transmission protocol.

[0064] S2: The digital twin model engine 6 constructs a precise geometric model of the temporary support structure based on BIM technology, forming a high-fidelity virtual mirror. The digital twin model engine 6 integrates the real-time sensor data from step S1 and uses Kalman filtering and its extended forms to reduce the difference between the predicted values ​​generated by the digital twin model and the actual measured values ​​of the sensors.

[0065] S3: The adaptive control decision center 7 receives the prediction results from step S2 and applies an autoregressive integral moving average model and an LSTM neural network algorithm to analyze the deformation trend. The adaptive control decision center 7 generates control commands and optimizes anti-deformation strategies based on historical data and real-time feedback. The decision output includes early warning signals and control parameters, ensuring adaptive decision-making under instability risk scenarios and improving construction safety.

[0066] S4: The hydraulic actuator 8 executes the control command generated in step S3, dynamically adjusting the support force through the hydraulic cylinder or servo system to achieve self-compensation for structural deformation. The actuator applies a force feedback mechanism at key nodes to monitor changes in support force in real time, and feeds the adjustment results back to the multi-source sensor network through a closed-loop system.

[0067] A three-dimensional geometric model of the temporary support structure is created using Building Information Modeling (BIM). This model integrates two main categories of key information: first, precise geometric information such as dimensions, shape, and spatial location; and second, non-geometric attribute parameters such as material type, specifications, and design strength of each component. Assuming the support structure is discretized into n key nodes, its initial geometric state is represented by a node coordinate matrix. express:

[0068]

[0069] In the formula, This represents the three-dimensional coordinates of the i-th node.

[0070] For linear statics problems, the core governing equations are the system equilibrium equations:

[0071]

[0072] In the formula, K is the global stiffness matrix; U is the nodal displacement vector. , represents the displacement of each node in three spatial directions; F is the node load vector.

[0073] To achieve synchronization between the virtual entity and the physical entity, real-time sensor data streams are introduced. Data assimilation is used to continuously refine the digital model, making it infinitely close to the real state of the physical entity. The digital twin engine 6, acting as the data processing hub, continuously receives real-time data streams from the multi-source sensor network of the previous stage (S1). The optimal set of corrections to the model parameters is found through calibration. This makes the predicted values ​​generated by the corrected digital model... Compared with the actual measured value of the sensor The difference between them is minimized. The calibration process is defined as follows:

[0074]

[0075] In the formula, Represents the initial parameter set of the model; This represents the obtained optimal parameter correction amount. .

[0076] To achieve efficient and recursive real-time calibration, this invention utilizes Kalman filtering and its extended forms. The core of this algorithm lies in the "prediction-update" process. In the prediction step, Kalman filtering and its extended forms predict the current system state based on the system's previous state and a known physical dynamic model, using a priori estimates.

[0077]

[0078] In the formula, This represents the prior estimate of the system state at time k, before fusing the sensor observations at time k. The state transition matrix describes how the system naturally evolves from the state at time k-1 to the state at time k. This represents the posterior estimate of the system state at time k-1, obtained after fusing the sensor observations from time k-1. To control the input matrix, it is described how the force applied by the hydraulic actuator changes the state of the support; Let be the control vector, representing the external control input known at time k.

[0079] At the same time, the uncertainty of the estimate is predicted:

[0080]

[0081] In the formula, Estimating the state from prior time The corresponding error covariance matrix; The posterior estimated state of the previous time step The error covariance matrix; Let be the process noise covariance matrix.

[0082] Then it proceeds to the update step, when new sensor observations... Upon arrival, calculate the Kalman gain. :

[0083]

[0084] In the formula, The observation matrix reflects the relationship between the internal state x of the system and the sensor observation value z; To observe the noise covariance matrix.

[0085] This gain balances the confidence level between model predictions and sensor predictions. Then, the posterior estimate of the system state is updated using the Kalman gain and the new observations.

[0086]

[0087] In the formula, This represents the actual observation value of the sensor at time k; The difference between the actual observations of the sensor and the observations predicted by the model is called innovation.

[0088] Synchronously update the posterior estimate of the error covariance:

[0089]

[0090] In the formula, It is an identity matrix.

[0091] Let the displacement of a specific measuring point be... Time series analysis was performed on continuously collected displacement information to analyze historical data sequences. Input the Autoregressive Integral Moving Average (ARIMA) model, which is expressed as follows:

[0092]

[0093] In the formula, For lag operators, These represent the order of autoregression, the order of differencing, and the order of moving average, respectively. and For model parameters, This is the white noise term.

[0094] ARIMA achieves prediction by capturing the autocorrelation and difference stationarity of time series. Furthermore, to handle more complex nonlinear relationships and long-term dependencies, the system employs a Long Short-Term Memory (LSTM) network, the core of which is as follows:

[0095]

[0096] In the formula, This is the forgetting gate vector, where each element has a value between 0 and 1, representing the cell state. The degree to which the corresponding information is retained; 1 means "completely retained", and 0 means "completely discarded"; Use the Sigmoid activation function; Here is the weight matrix for the forget gate; For the splicing operation, the hidden state from the previous moment is removed. With the input vector at the current time step Connect them into a longer vector; This is the bias vector for the forget gate; The input gate vector; The candidate cell state vector is generated by the tanh function and contains potential new information generated based on the current input and past states. This refers to the cell state updated at the current moment, i.e., long-term memory; This represents the cell state at the previous moment; The Hadamard product is the element-wise multiplication of vectors. Forgetting operation; For memory operations; This is the output gate vector; The weight matrix is ​​a learnable matrix; It is a learnable bias vector; The hidden state at the current moment is the output of the LSTM unit; First, the updated cell state is scaled to between -1 and 1 using the tanh function.

[0097] The predicted key point displacement value at a future time point (t+h) Compared with the maximum safe displacement threshold allowed by structural design specifications Real-time comparisons are performed, and the displacement risk index is defined as follows: Based on the risk index, the system implements dynamic tiered early warning systems, providing clear alerts at corresponding levels for decision-making. The specific rules are defined as follows:

[0098] Yellow Alert (Attention Level): When A value >0.8 triggers the signal. This signal indicates that structural deformation has entered a period of accelerated development. Although not immediately dangerous, the risk has increased significantly. It prompts engineers to increase monitoring frequency and notify the decision-making center to begin contingency plan analysis, which may require limiting the construction loading rate.

[0099] Red Alert (Action Level): A value >0.95 triggers the highest level alarm, indicating that the structure is extremely close to its load-bearing limit and the risk of instability is extremely high. The decision-making center must immediately activate the pre-defined control strategy, issuing commands to the hydraulic actuators to proactively intervene, compensate for deformation, and eliminate the risk. Simultaneously, the system will recommend on-site evacuation or cessation of related operations.

[0100] It receives the control command set issued by the adaptive control decision center 7, and through its internal hydraulic cylinder or high-precision servo system, converts the digital control parameters into actual mechanical action applied to the temporary support structure, thereby actively and dynamically offsetting the deformation of the structure caused by external loads.

[0101] set up This is the control command vector issued by the decision-making center at time k, whose elements include the target support force. The actuator drives the hydraulic system according to control commands, and its dynamic response process can be described by the following simplified discrete state-space equation:

[0102]

[0103] In the formula, This refers to the internal state vector of the actuator system, including physical quantities such as the hydraulic cylinder piston position and oil chamber pressure. The system output is the actual supporting force applied to the structure. Matrix A, B, and C are determined by the dynamic characteristics of the actuator itself, including factors such as hydraulic oil compressibility, valve spool response characteristics, and cylinder friction. The control objective of the system is to achieve the desired actual output. Quickly and accurately track instructions That is, to achieve .

[0104] To ensure control accuracy, the hydraulic actuator 8 integrates high-precision force and displacement sensors, forming a local closed-loop feedback control. The system monitors the output force in real time. The actual deformation of the structure is monitored, and the data is fed back to the multi-source sensor network 1. This feedback mechanism enables the system to form a continuously iterative intelligent closed loop: the feedback data is sent to the digital twin model engine 6, where it is used by the Kalman filter algorithm for model state updates and parameter calibration.

[0105] The hydraulic actuator 8 thus constitutes an execution terminal connecting the information space and the physical world. By transforming intelligent commands into precise mechanical intervention and using force feedback to close the global control loop, it ultimately ensures the safety and controllability of the temporary support structure under complex construction conditions.

[0106] like Figure 3 As shown, the Kalman filter algorithm used in this invention exhibits significant advantages in water pressure monitoring: the average absolute error (0.12 kPa) between its processing results and theoretical values ​​is reduced by 52% compared to the traditional moving average method (0.25 kPa), and it can identify trend changes 2-3 time points in advance when sudden pressure fluctuations occur. At the same time, it effectively filters instantaneous interference. These characteristics of high precision, strong real-time performance, and anti-interference provide a reliable data foundation for deformation early warning and adaptive control of temporary support structures under high water level conditions during the flood season.

[0107] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0108] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0109] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0110] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0111] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A digital-twin-based intelligent monitoring system for adaptive temporary support structures, characterized by, The system comprises a multi-source sensing network (1), a digital twin model engine (6), an adaptive regulation and decision center (7), and a hydraulic actuator (8). The multi-source sensing network (1) is arranged at key nodes of the temporary support structure. The digital twin model engine (6) constructs a high-fidelity virtual mirror through BIM geometric modeling and finite element mechanics simulation, and fuses real-time data to drive dynamic prediction. The adaptive regulation and decision center (7) generates regulation instructions based on the LSTM deformation prediction algorithm and the reinforcement learning strategy. The hydraulic actuator (8) dynamically adjusts the support force according to the regulation instructions to realize self-compensation of structural deformation.

2. The digital-twin-based intelligent monitoring system for adaptive temporary support structures according to claim 1, wherein, The multi-source sensing network (1) comprises stress sensors (2), displacement meters (3), inclinometers (4), and water pressure gauges (5) for real-time collection of stress, displacement, inclination, and water pressure data.

3. An intelligent monitoring method for an adaptive temporary support structure based on digital twinning, characterized by, The system is implemented based on the system of claim 1 or 2, comprising the following steps: S1: A multi-source sensing network (1) is arranged at the stress concentration points of the temporary support structure. The multi-source sensing network (1) collects key parameters in real time at a high frequency and transmits the raw data to the central processing unit through a wireless transmission protocol; S2: The digital twin model engine (6) constructs an accurate geometric model of the temporary support structure based on BIM technology to form a high-fidelity virtual mirror. The digital twin model engine (6) fuses the real-time sensing data of step S1 and uses Kalman filtering and its extended form algorithm to reduce the difference between the predicted value of the digital twin model and the actual measured value of the sensor; S3: The adaptive regulation and decision center (7) receives the prediction results of step S2, analyzes the deformation trend using the autoregressive integrated moving average model and the LSTM neural network algorithm, generates regulation instructions, optimizes the anti-deformation strategy based on historical data and real-time feedback, and outputs warning signals and regulation parameters to ensure adaptive decision-making in unstable risk scenarios and improve construction safety; S4: The hydraulic actuator (8) executes the regulation instructions generated in step S3, dynamically adjusts the support force through a hydraulic cylinder or a servo system, and realizes self-compensation of structural deformation. The actuator applies a force feedback mechanism at the key nodes to monitor the change in support force in real time and feeds back the adjustment results to the multi-source sensing network through a closed-loop system.

4. The digital-twin-based intelligent monitoring method of adaptive temporary support structures according to claim 3, characterized in that, A three-dimensional geometric model of a temporary support structure is created using a building information model, BIM, which integrates geometric information and non-geometric attribute parameters; it is assumed that the support structure is discretized into n key nodes, whose initial geometric state is represented by a node coordinate matrix represents: wherein represents the three-dimensional coordinates of the i-th node; For linear static problems, the core control equation is the system balance equation: In the formula, K is a global stiffness matrix; U is a node displacement vector, , which represents the displacement of each node in three directions in space; and F is a node load vector.

5. The digital-twin-based intelligent monitoring method of adaptive temporary support structures as claimed in claim 3, wherein, To achieve synchronization between virtual body and physical entity, real-time sensor data stream is introduced, through data assimilation to continuously correct digital model, so that it infinitely approaches the real state of physical entity; digital twin engine (6) as the data processing center, continuously receives real-time data stream from the previous stage of multi-source sensor network ; through calibration to find a set of optimal correction amount for model parameters , so that the difference between the predicted value generated by the corrected digital model and the actual measured value of the sensor reaches the minimum; the calibration process is defined as: In the formula, denotes the initial parameter set of the model; denotes the obtained optimal parameter correction amount, ; Real-time calibration is achieved using Kalman filtering and its extended form algorithm. In the prediction step, Kalman filtering and its extended form algorithm predict the prior estimate of the system state at the current time based on the state at the previous time and the known physical dynamics model: wherein, denotes the prior estimate of the system state at time k before fusing the sensor observations at time k; denotes the state transition matrix which describes how the system naturally evolves from the state at time k-1 to the state at time k; denotes the posterior estimate of the system state at time k-1 after fusing the sensor observations at time k-1; is the control input matrix which describes how the forces applied by the hydraulic actuators change the state of the support; is the control vector which denotes the external control inputs known at time k; At the same time, the uncertainty of the estimate is predicted: wherein is the a priori estimate of the state is the error covariance matrix corresponding to is the a priori estimate of the state is the error covariance matrix corresponding to is the process noise covariance matrix Subsequently, an update step is entered when new sensor observations arrive, the Kalman gain is computed : In the formula, is an observation matrix, reflecting the relationship between the internal state x of the system and the sensor observation value z; is an observation noise covariance matrix; Then, the posterior estimate of the system state is updated using the Kalman gain and the new observation value: In the formula, represents the actual observation value of the sensor at time k; represents the difference between the actual observation value of the sensor and the model predicted observation value; The posterior estimate of the error covariance is updated synchronously: wherein is the identity matrix.

6. The digital-twin-based intelligent monitoring method of adaptive temporary support structures according to claim 3, characterized in that, Let the displacement of a measuring point be The time series analysis is performed on the continuously collected displacement information, and the historical data sequence An autoregressive integrated moving average model ARIMA is input, and the model is expressed as: wherein is a lag operator, respectively represent the autoregressive order, the differencing order and the moving average order, and are model parameters, is a white noise term; ARIMA achieves prediction by capturing the autocorrelation and difference stationarity of time series. In addition, to handle more complex nonlinear relationships and long-term dependencies, the system uses the long short-term memory network LSTM, whose operation formula is as follows: where, is the forget gate vector, whose elements are between 0 and 1, indicating the degree of retention of the corresponding information in the cell state ; 1 means "completely retain", 0 means "completely discard"; is the Sigmoid activation function; is the weight matrix of the forget gate; is the concatenation operation, which connects the hidden state of the previous moment and the input vector of the current moment into a longer vector; is the bias vector of the forget gate; is the input gate vector; is the candidate cell state vector generated by the tanh function, which contains the potential new information generated according to the current input and the past state; is the updated cell state at the current moment, i.e. long-term memory; is the cell state of the last moment; is the Hadamard product, i.e. element-wise multiplication of vectors; is the forget operation; is the memory operation; is the output gate vector; is the learnable weight matrix; is a learnable bias vector; is the hidden state at the current moment, which is the output of the LSTM unit; First, the updated cell state is scaled to -1 to 1 by the tanh function; The predicted key point displacement value at a future time point (t+h) is compared with the maximum safety displacement threshold value allowed by the structural design specification in real time to define a displacement risk index as follows: The displacement risk index is compared with the maximum safety displacement threshold value allowed by the structural design specification in real time to define a displacement risk index as follows: Based on the risk index, the system implements dynamic grading early warning and provides corresponding level alarms for decision-making.​ 7. The digital-twin-based intelligent monitoring method of adaptive temporary support structures according to claim 6, characterized in that, The specific rules of the dynamic hierarchical warning are defined as follows: Yellow warning (attention level): triggered when 0.8; this signal indicates that the structural deformation has entered an accelerated development stage, although it is not imminent danger, but the risk is significantly increased; it prompts the engineering staff to need to strengthen the monitoring frequency, and informs the decision center to start the pre-plan analysis, which may need to limit the construction loading rate; Red Alert (action level): Trigger at 0.95; This is the highest level of alarm, which means that the structure has been extremely close to its bearing limit, and the risk of instability is extremely high; The decision center must immediately start the pre-prepared control strategy and issue instructions to the hydraulic actuator to actively intervene to compensate for deformation and eliminate risks. At the same time, the system will suggest evacuation or stop related work on site.

8. The digital-twin-based intelligent monitoring method of adaptive temporary support structures as claimed in claim 3, wherein, Receive the control instruction set from the adaptive control decision center (7), and through its internal hydraulic cylinder or high-precision servo system, convert the digital control parameters into actual mechanical actions on the temporary support structure, thereby actively and dynamically offsetting the deformation of the structure due to external loads; Let is the control instruction vector issued by the decision center at time k, whose elements contain the target support force ; the actuator drives the hydraulic system according to the control instruction, and its dynamic response process is described by the following simplified discrete state space equation: wherein is the internal state vector of the actuator system, including physical quantities such as hydraulic cylinder piston position, oil chamber pressure, etc. is the system output, i.e. the support force actually applied to the structure ; matrices A, B, C are determined by the dynamic characteristics of the actuator itself; the control objective of the system is to make the actual output quickly and accurately track the command , i.e. to achieve .

9. The digital-twin-based intelligent monitoring method of adaptive temporary support structures as claimed in claim 3, wherein, To ensure the control accuracy, the hydraulic actuator (8) is integrated with high-precision force and displacement sensors to form a local closed-loop feedback control; the system monitors the output force in real time and the actual deformation of the structure, and feeds the data back to the multi-source sensing network (1); this feedback mechanism forms a continuous iterative intelligent closed loop for the system: the feedback data is fed into the digital twin model engine (6) and is used by the Kalman filtering algorithm for model state updating and parameter calibration; The hydraulic actuator (8) thus constitutes the execution terminal connecting the information space and the physical world, by converting intelligent instructions into precise mechanical interventions, and with the help of force feedback to close the global control loop, ultimately ensuring the safety and controllability of the temporary support structure under complex construction conditions.

Images