Historic building structure assessment system based on digital twinning

Abstract

The application relates to the technical field of digital twinning, and particularly discloses a historical building structure evaluation system based on digital twinning, which comprises a multi-source data acquisition module, a dynamic digital twinning modeling module, a structure health evaluation module, an intelligent maintenance decision module and a collaborative optimization analysis module. The scheme captures the geometric, material and environmental information of the historical building, realizes all-around monitoring of the historical building through a digital twinning modeling technology, constructs an intelligent evaluation model in combination with finite element analysis and an SVM algorithm, quickly identifies a high-risk area, and ensures the authenticity and accuracy of the evaluation result; the reinforcement learning algorithm is used to adapt to the complex and changeable environmental requirements of the historical building, automatically generates a maintenance strategy, continuously optimizes the collaboration efficiency among the modules through collaborative optimization analysis, improves the overall operation performance, guarantees the evaluation accuracy, and effectively controls the system resource consumption.

Description

Technical Field

[0001] This invention relates to the field of digital twin technology, specifically to a historical building structure assessment system based on digital twins. Background Technology

[0002] Digital twin technology provides a novel solution for the monitoring and assessment of building structures by constructing virtual mapping models of physical entities and combining them with real-time data acquisition and analysis. However, existing historical building structure assessment technologies still have shortcomings in terms of the comprehensiveness of data acquisition, the dynamism of model construction, and the intelligence of assessment methods; general maintenance decision-making processes are highly subjective, making it difficult to quantify and assess risks and benefits, and the overall system resource consumption is large and the cost is high. Summary of the Invention

[0003] To address the aforementioned issues and overcome the shortcomings of existing technologies, this invention provides a historical building structure assessment system based on digital twins. Addressing the deficiencies in comprehensive data acquisition, dynamic model construction, and intelligent assessment methods of existing historical building structure assessment technologies, this solution captures the geometric, material, and environmental information of historical buildings. Through digital twin modeling technology, it achieves comprehensive monitoring of historical buildings. Combining finite element analysis and SVM (Support Vector Machine) algorithms, it constructs an intelligent assessment model to quickly identify high-risk areas, ensuring the authenticity and accuracy of the assessment results. Furthermore, addressing the issues of subjective nature in general maintenance decision-making processes, difficulty in quantifying risk and benefit assessments, and high overall system resource consumption and costs, this solution uses reinforcement learning algorithms to adapt to the complex and ever-changing environmental requirements of historical buildings, automatically generating maintenance strategies. Through collaborative optimization analysis, it continuously optimizes the collaboration efficiency between modules, improving overall operational performance, ensuring assessment accuracy while effectively controlling system resource consumption.

[0004] The present invention provides a historical building structure assessment system based on digital twins, including a multi-source data acquisition module, a dynamic digital twin modeling module, a structural health assessment module, an intelligent maintenance decision-making module, and a collaborative optimization analysis module;

[0005] The multi-source data acquisition module comprehensively collects the time series of geometric shape, material properties, structural response and environmental parameter information of historical buildings. The data samples collected at each time point are preprocessed to obtain the preprocessed data and sent to the dynamic digital twin modeling module.

[0006] The dynamic digital twin modeling module uses preprocessed data to construct a digital twin model of the historical building, updates the model parameters based on real-time data samples to reflect the current state of the historical building, and sends the digital twin model and its parameters to the structural health assessment module.

[0007] The structural health assessment module uses the finite element analysis method to construct a multidimensional feature vector of the historical building, and combines it with the SVM algorithm to quantitatively assess the structural status of the historical building, identify potential risk areas in the historical building, and output the structural status level.

[0008] The intelligent maintenance decision-making module uses reinforcement learning algorithms to learn maintenance strategies and automatically generate maintenance plans based on the multi-dimensional feature vectors and structural state levels of historical buildings.

[0009] The collaborative optimization analysis module constructs a multi-objective optimization model to optimize the collaboration efficiency between modules and achieve rational allocation of resources.

[0010] Furthermore, the dynamic digital twin modeling module includes a model building unit, a structural characteristic loading unit, a state update unit, and a model output unit;

[0011] The model building unit constructs a basic three-dimensional model of the historical building based on the geometric morphology data in the data sample.

[0012] The structural characteristic loading unit, based on the material properties and structural response data in the data sample, simulates the thermal conductivity and mechanical response of historical building materials through physical simulation and data analysis, and adds material property and structural response information to the basic three-dimensional model.

[0013] The state update unit uses the Kalman filter algorithm to update the model parameters based on real-time data samples, reflecting the current state of the historical building. The formula used is as follows:

[0014] ;

[0015] ;

[0016] In the formula, and These represent the current time point and the previous time point, respectively. Indicates the predicted state. This represents the state estimate at the previous moment. This represents the state transition matrix of a historical building from the previous time point to the current time point. The control input matrix represents the impact of environmental parameters on the condition of historical buildings. This represents a control vector indicating the impact of environmental parameter information on the state of historical buildings. This represents the prediction error covariance. This represents the error covariance at the previous time point. The transpose of the state transition matrix. This represents the process noise covariance in the uncertain evolution of historical buildings;

[0017] The model output unit dynamically adjusts the basic 3D model based on the updated model parameters to generate a visualized digital twin model.

[0018] Furthermore, the structural health assessment module includes: a condition loading unit, a feature extraction unit, and a structural state classification unit;

[0019] The condition loading unit applies boundary conditions and loading conditions to the digital twin model, and uses the finite element analysis method to calculate the stress distribution, displacement field and strain energy of key areas of the historical building, and obtain the changes in model parameters.

[0020] The feature extraction unit identifies potential risk areas in historical buildings based on changes in model parameters, extracts representative features, and constructs a multi-dimensional feature vector for historical buildings.

[0021] The structural status classification unit uses SVM to classify the structural status of historical buildings. It inputs the multi-dimensional feature vectors of historical buildings according to time series, automatically selects support vectors of representative samples, and outputs the structural category label for each time point, mapping it to the structural status level of the historical building. The formula used is as follows:

[0022] ;

[0023] In the formula, Represents a multidimensional feature vector. Represents support vectors, Indicates the index of the support vector. This indicates the number of support vectors. The class label representing the support vector. Represents the Lagrange multipliers. Represents the kernel function. Indicates the bias term. The sign function representing the classification result. This indicates the classification result.

[0024] Furthermore, the intelligent maintenance decision-making module includes a status action recognition unit, a maintenance strategy learning unit, and a maintenance strategy generation unit;

[0025] The state action recognition unit receives multi-dimensional feature vectors and encodes them into state variables, and combines them with the structural state level definition to perform maintenance operations on historical buildings.

[0026] The maintenance strategy learning unit is based on a reinforcement learning algorithm. It learns maintenance strategies through a state-action value function, and maximizes long-term maintenance benefits by combining maintenance costs, construction difficulty, and protection effect objectives. The formula used is as follows:

[0027] ;

[0028] ;

[0029] In the formula, Indicates the current structural state of a historical building. Indicates the currently available maintenance actions. Represents a state-action value function. Indicates the learning rate. This indicates the immediate benefits of the current maintenance action. Indicates the discount factor. This indicates the new state after the maintenance operation is performed. Indicates the next maintenance action. Represents the maximum value function. , , Indicates the target weight coefficient. Indicates the protective effect. Indicates maintenance costs, Indicates the difficulty of construction;

[0030] The maintenance strategy generation unit selects the action with the largest state-action value function as the optimal maintenance action in the current state, generates and outputs the maintenance plan, and uses the following formula:

[0031] ;

[0032] In the formula, Represents the action space. The function representing the maximum value of the point set of the independent variable. This indicates the optimal maintenance action.

[0033] The beneficial effects achieved by the present invention using the above solution are as follows:

[0034] (1) In view of the shortcomings of existing historical building structure assessment technology in terms of the comprehensiveness of data collection, the dynamism of model construction and the intelligence of assessment methods, this solution captures the geometric, material and environmental information of historical buildings, realizes the all-round monitoring of historical buildings through digital twin modeling technology, and constructs an intelligent assessment model by combining finite element analysis and SVM algorithm to quickly identify high-risk areas and ensure the authenticity and accuracy of assessment results.

[0035] (2) In view of the problems that the general maintenance decision-making process is highly subjective, difficult to quantify and assess risks and benefits, and has high overall system resource consumption and cost, this solution uses reinforcement learning algorithm to adapt to the complex and ever-changing environmental requirements of historical buildings, automatically generates maintenance strategies, and continuously optimizes the collaboration efficiency between modules through collaborative optimization analysis, thereby improving overall operating performance, ensuring assessment accuracy, and effectively controlling system resource consumption. Attached Figure Description

[0036] Figure 1 This is a schematic diagram of a historical building structure evaluation system based on digital twins proposed in this invention.

[0037] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0038] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0039] Example 1, see Figure 1 The present invention provides a historical building structure assessment system based on digital twins, including a multi-source data acquisition module, a dynamic digital twin modeling module, a structural health assessment module, an intelligent maintenance decision-making module, and a collaborative optimization analysis module;

[0040] The multi-source data acquisition module integrates a high-precision laser scanner, infrared thermal imaging equipment, vibration sensor, and environmental monitoring sensor to comprehensively collect time series information on the geometric shape, material properties, structural response, and environmental parameters of historical buildings. The data samples collected at each time point are preprocessed to obtain preprocessed data, which is then sent to the dynamic digital twin modeling module.

[0041] The dynamic digital twin modeling module uses preprocessed data to construct a digital twin model of the historical building, updates the model parameters based on real-time data samples to reflect the current state of the historical building, and sends the digital twin model and its parameters to the structural health assessment module.

[0042] The structural health assessment module uses the finite element analysis method to construct a multidimensional feature vector of the historical building, and combines it with the SVM algorithm to quantitatively assess the structural status of the historical building, identify potential risk areas in the historical building, and output the structural status level.

[0043] The intelligent maintenance decision-making module uses reinforcement learning algorithms to learn maintenance strategies and automatically generate maintenance plans based on the multi-dimensional feature vectors and structural state levels of historical buildings.

[0044] The collaborative optimization analysis module constructs a multi-objective optimization model to optimize the collaboration efficiency between modules and achieve rational allocation of resources.

[0045] Example 2, see Figure 1 This embodiment is based on the above embodiment. The multi-source data acquisition module preprocesses the data samples collected at each time point. Specifically, it cleans and normalizes the data points in each data sample across four dimensions: geometric morphology, material properties, structural response, and environmental parameters, eliminating noise interference and unifying the data format. An anomaly detection algorithm based on a convolutional neural network is used to calculate the anomaly probability of the data sample. If the calculated anomaly probability is higher than a threshold of 0.9, the data sample is marked as abnormal data and recorded in the log. The formula used is as follows:

[0046] ;

[0047] In the formula, Indicates the time index of the data sample. The dimension index of the data point. This indicates the number of dimensions of the data points contained in the data sample. Indicates the first In the data sample, the first The value of each data point. Indicates the first Anomalies in a data sample Indicates the first The probability of an anomaly in a data sample. For activation function, Indicates the first The weighting coefficients for each data point For bias terms, This represents a nonlinear transformation.

[0048] Example 3, see Figure 1 This embodiment is based on the above embodiment, and the dynamic digital twin modeling module includes a model building unit, a structural characteristic loading unit, a state update unit, and a model output unit;

[0049] The model building unit constructs a basic three-dimensional model of the historical building based on the geometric morphology data in the data sample.

[0050] The structural characteristic loading unit, based on the material properties and structural response data in the data sample, simulates the thermal conductivity and mechanical response of historical building materials through physical simulation and data analysis, and adds material property and structural response information to the basic three-dimensional model.

[0051] The state update unit uses the Kalman filter algorithm to update the model parameters based on real-time data samples, reflecting the current state of the historical building. The formula used is as follows:

[0052] ;

[0053] ;

[0054] In the formula, and These represent the current time point and the previous time point, respectively. Indicates the predicted state. This represents the state estimate at the previous moment. This represents the state transition matrix of a historical building from the previous time point to the current time point. The control input matrix represents the impact of environmental parameters on the condition of historical buildings. This represents a control vector indicating the impact of environmental parameter information on the state of historical buildings. This represents the prediction error covariance. This represents the error covariance at the previous time point. The transpose of the state transition matrix. This represents the process noise covariance in the uncertain evolution of historical buildings;

[0055] The model output unit dynamically adjusts the basic 3D model based on the updated model parameters to generate a visualized digital twin model.

[0056] Example 4, see Figure 1 This embodiment is based on the above embodiment, and the structural health assessment module includes a condition loading unit, a feature extraction unit, and a structural state classification unit;

[0057] The condition loading unit applies boundary conditions and loading conditions to the digital twin model, and uses the finite element analysis method to calculate the stress distribution, displacement field and strain energy of key areas of the historical building, and obtain the changes in model parameters.

[0058] The feature extraction unit identifies potential risk areas in historical buildings based on changes in model parameters, extracts representative features, and constructs a multi-dimensional feature vector for historical buildings.

[0059] The structural status classification unit uses SVM to classify the structural status of historical buildings. It inputs the multi-dimensional feature vectors of historical buildings according to time series, automatically selects support vectors of representative samples, and outputs the structural category label for each time point, mapping it to the structural status level of the historical building. The formula used is as follows:

[0060] ;

[0061] In the formula, Represents a multidimensional feature vector. Represents support vectors, Indicates the index of the support vector. This indicates the number of support vectors. The class label representing the support vector. Represents the Lagrange multipliers. Represents the kernel function. Indicates the bias term. The sign function representing the classification result. This indicates the classification result.

[0062] By performing the aforementioned operations, this solution addresses the shortcomings of existing historical building structural assessment technologies in terms of the comprehensiveness of data collection, the dynamism of model construction, and the intelligence of assessment methods. It captures the geometric, material, and environmental information of historical buildings, achieves comprehensive monitoring of historical buildings through digital twin modeling technology, and constructs an intelligent assessment model by combining finite element analysis and SVM algorithms to quickly identify high-risk areas and ensure the authenticity and accuracy of the assessment results.

[0063] Example 5, see Figure 1 This embodiment is based on the above embodiment, and the intelligent maintenance decision module includes a status action recognition unit, a maintenance strategy learning unit, and a maintenance strategy generation unit;

[0064] The state action recognition unit receives multi-dimensional feature vectors and encodes them into state variables, and combines them with the structural state level definition to perform maintenance operations on historical buildings.

[0065] The maintenance strategy learning unit is based on a reinforcement learning algorithm. It learns maintenance strategies through a state-action value function, and maximizes long-term maintenance benefits by combining maintenance costs, construction difficulty, and protection effect objectives. The formula used is as follows:

[0066] ;

[0067] ;

[0068] In the formula, Indicates the current structural state of a historical building. Indicates the currently available maintenance actions. Represents a state-action value function. Indicates the learning rate. This indicates the immediate benefits of the current maintenance action. Indicates the discount factor. This indicates the new state after the maintenance operation is performed. Indicates the next maintenance action. Represents the maximum value function. , , Indicates the target weight coefficient. Indicates the protective effect. Indicates maintenance costs, Indicates the difficulty of construction;

[0069] The maintenance strategy generation unit selects the action with the largest state-action value function as the optimal maintenance action in the current state, generates and outputs the maintenance plan, and uses the following formula:

[0070] ;

[0071] In the formula, Represents the action space. The function representing the maximum value of the point set of the independent variable. This indicates the optimal maintenance action.

[0072] Example 6, see Figure 1 This embodiment is based on the above embodiment. The collaborative optimization analysis module constructs a multi-objective optimization model to optimize the collaborative efficiency between modules. Specifically, it sets resource allocation strategies for each module, including: optimizing the data acquisition frequency of the multi-source data acquisition module, the model update cycle of the dynamic digital twin modeling module, the state assessment resolution of the structural health assessment module, and the computational space size of the intelligent maintenance decision module, minimizing the overall cost of the system and maintaining the collaborative efficiency between modules. The formula used is as follows:

[0073] ;

[0074] In the formula, Indicates the module index. Indicates the number of modules. Indicates the first The goals of each module Indicates the weighting coefficient. This represents the penalty term for the constraint condition. Indicates the penalty coefficient. This indicates the resource allocation strategy for each module. This represents a multi-objective optimization model. This represents the minimum value function.

[0075] By performing the aforementioned operations, this solution addresses the issues of highly subjective general maintenance decision-making processes, difficulty in quantifying and assessing risks and benefits, and high overall system resource consumption and costs. It utilizes reinforcement learning algorithms to adapt to the complex and ever-changing environmental requirements of historical buildings, automatically generating maintenance strategies. Furthermore, it continuously optimizes the collaboration efficiency between modules through collaborative optimization analysis, thereby improving overall operational performance, ensuring assessment accuracy, and effectively controlling system resource consumption.

[0076] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0077] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

[0078] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A digital-twin-based historical building structure assessment system, characterized by: The system comprises a multi-source data acquisition module, a dynamic digital twin modeling module, a structure health assessment module, an intelligent maintenance decision module, and a collaborative optimization analysis module. The multi-source data acquisition module comprehensively collects time series of geometric shapes, material properties, structural responses, and environmental parameter information of the historical building, pre-processes data samples collected at each time point, obtains pre-processed data, and sends the pre-processed data to the dynamic digital twin modeling module. The dynamic digital twin modeling module constructs a digital twin model of the historical building using the pre-processed data, updates model parameters according to real-time data samples, reflects the current state of the historical building, and sends the digital twin model and the parameters to the structure health assessment module. The structure health assessment module uses a finite element analysis method to construct a multi-dimensional feature vector of the historical building, combines a SVM algorithm to quantitatively assess the structural state of the historical building, identifies potential risk areas in the historical building, and outputs a structure state grade. The intelligent maintenance decision module learns a maintenance strategy based on the multi-dimensional feature vector and the structure state grade of the historical building, and automatically generates a maintenance scheme using a reinforcement learning algorithm. The collaborative optimization analysis module constructs a multi-objective optimization model to optimize the collaboration efficiency among modules and realize reasonable allocation of resources. The dynamic digital twin modeling module comprises a model construction unit, a structure characteristic loading unit, a state updating unit, and a model output unit. The model construction unit constructs a basic three-dimensional model of the historical building based on geometric shape data in the data samples. The structure characteristic loading unit simulates the thermal conduction characteristics and mechanical responses of the historical building materials through physical simulation and data analysis based on material properties and structural response data in the data samples, and adds material properties and structural response information to the basic three-dimensional model. The state updating unit uses a Kalman filter algorithm to update model parameters according to real-time data samples, reflecting the current state of the historical building. The model output unit dynamically adjusts the basic three-dimensional model based on the updated model parameters, and generates a visual digital twin model. The structure health assessment module comprises a condition loading unit, a feature extraction unit, and a structure state classification unit. The condition loading unit applies boundary conditions and loading conditions to the digital twin model, calculates stress distribution, displacement field, and strain energy of key areas of the historical building using a finite element analysis method, and obtains changes in model parameters. The feature extraction unit identifies potential risk areas in the historical building according to changes in model parameters, extracts representative features, and constructs a multi-dimensional feature vector of the historical building. The structure state classification unit classifies the structural state of the historical building using a SVM, inputs the multi-dimensional feature vector of the historical building in time series, automatically selects support vectors with representative samples, outputs a structure class label at each time point, and maps the structure class label to a structure state grade of the historical building. The intelligent maintenance decision module comprises a state action recognition unit, a maintenance strategy learning unit, and a maintenance strategy generation unit. The state action recognition unit receives a multi-dimensional feature vector and encodes it as a state variable, and defines an action space for maintenance operation of the historic building in combination with a structural state level; The maintenance strategy learning unit learns a maintenance strategy based on a reinforcement learning algorithm through a state-action value function, and maximizes long-term maintenance benefits in combination with multiple targets; The maintenance strategy generation unit selects an optimal maintenance action as the maximum state-action value function in the current state, and generates and outputs a maintenance scheme.

Images