Building structure safety risk assessment system

Abstract

The invention discloses a building structure safety risk assessment system, and relates to the technical field of building engineering.The system comprises a building mechanical parameter acquisition module, a rigidity index analysis module, an image feature extraction module, a comprehensive analysis module and a risk assessment module. The method comprises the following steps of: reflecting the mechanical property and the damage degree of a building structure under the action of a load, comparing and analyzing with a collected concrete surface feature image, extracting the damage condition and the image feature value of the surface of the building structure, fusing a rigidity index and a loss function through a comprehensive analysis module to obtain a loss feature value, and calculating the damage degree of the building structure. The comprehensive assessment method not only improves the comprehensiveness and accuracy of the assessment result, but also provides powerful data support for subsequent risk control and repair measures.

Description

Technical Field

[0001] The invention relates to the technical field of building engineering, and in particular to a building structure safety risk assessment system. Background Art

[0002] Throughout the entire life cycle of a building structure, from design and construction to use and maintenance, its structural safety needs to be continuously monitored and evaluated. Due to the frequent occurrence of natural disasters and the increasing prominence of problems such as building aging, the importance of building structure safety risk assessment has become increasingly prominent. Traditional building structure safety assessment methods mainly rely on manual inspection and experience judgment. This method has many limitations and cannot meet the growing demand for building structure safety assessment.

[0003] At present, in the determination of the rigidity index of existing building structure safety assessment technologies, traditional methods often only consider a single factor and a simple mechanical relationship for calculation, resulting in inaccurate calculation of the rigidity index and failure to fully reflect the actual rigidity and deformation resistance of the building structure under complex working conditions. For the calculation of the loss function, the analysis is based on simple image features, without fully combining the various characteristic information of the concrete surface and the mechanical performance parameters of the building structure, so that the loss function cannot accurately quantify the degree of surface damage of the building structure and its impact on the overall safety, resulting in incomplete and inaccurate safety assessment results of the building structure, making it difficult to discover potential safety risks in advance. Summary of the invention

[0004] The purpose of the present invention is to make up for the shortcomings of the prior art and to provide a building structure safety risk assessment system. The system can comprehensively evaluate the safety status of the building structure under earthquake action by combining the analysis method of the rigidity index and the loss function. At the same time, the analysis of the concrete surface features by the image feature extraction module further supplements the information of the surface damage of the structure, which can more comprehensively reflect the true status of the building structure and reduce the assessment error caused by a single data source.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] A building structure safety risk assessment system, the system includes a building mechanics parameter acquisition module, a rigidity index analysis module, an image feature extraction module, a comprehensive analysis module, and a risk assessment module;

[0007] The building mechanics parameter acquisition module is used to collect building mechanics parameters in each monitoring area of ​​the building structure, including: seismic load value, ultimate strength and steel bar damage degree value, wherein: seismic load value acquisition uses a seismograph to monitor the dynamic response data of the building structure under the action of an earthquake in real time, including acceleration, velocity and displacement parameters, and obtains the seismic load value through integral calculation; ultimate strength acquisition uses stress sensors and strain sensors to monitor the strength changes of building components in real time; steel bar damage degree value acquisition detects the number of steel bar damages in the building structure and calculates the steel bar damage degree value;

[0008] The rigidity index analysis module calculates the rigidity index based on the building mechanics data collected by the building mechanics parameter collection module;

[0009] The image feature extraction module is responsible for collecting the concrete surface feature image and inputting it into the safety monitoring model. After image enhancement processing, the image feature value is obtained, and the loss function is calculated based on the image feature value. The safety monitoring model construction process is as follows:

[0010] Image sample collection and annotation: Collect concrete surface feature image samples, including images of different crack types, different degrees of damage, and images in normal state, to form a sample set Among them I j represents the jth image, N is the total number of samples, and the images in the sample set are labeled, including crack type, crack length, width, and number, to form a labeled set Among them l j Representation and Image I j Corresponding annotation information;

[0011] Feature extraction network design: Construct a convolutional neural network CNN as the architecture of the security monitoring model. The network contains multiple convolutional layers, pooling layers, and fully connected layers. Convolutional layer: For the input image I, the output feature map F of the lth convolutional layer l for: where F l-1 is the output feature map of the previous layer, represents the convolution operation, W l is the convolution kernel weight matrix of layer l, b l is the bias term of the lth layer, f(·) is the activation function; Pooling layer: for the input feature map F l , the output feature map F′ after pooling l is: F′ l =Pooling(F l ), where Pooling(·) represents the pooling operation; Fully connected layer: The feature map processed by multiple convolutional layers and pooling layers is expanded into a one-dimensional vector and connected to the fully connected layer. The output of the fully connected layer is Y = f(W fcX+b fc ), where X is the expanded eigenvector, W fc is the weight matrix of the fully connected layer, b fc is the bias term of the fully connected layer, and f(·) is the activation function;

[0012] Model training: Define the model's loss function J(θ), using a combination of the cross entropy loss function and the mean square error loss function, i.e. J(θ) = λ1J ce (θ)+λ2J mse (θ), where θ represents the parameters of the feature extraction network, λ1 and λ2 are weight coefficients used to balance the contribution of the two loss functions. The cross entropy loss function J ce (θ) is used for classification tasks, i.e., the classification of crack types, and the calculation formula is: C is the number of categories, y ji is the indicator variable, N is the total number of samples, for image I j Belongs to category i when y ki =1, otherwise y ji =0, p ji (θ) is the model prediction image I j The probability of belonging to category i, the mean square error loss function J mse (θ) is used for regression tasks, i.e., the prediction of crack length and width parameters, and the calculation formula is: Where M is the number of regression tasks, z jm For image I j The actual value of the mth regression target, q jm (θ) is the corresponding value predicted by the model. The network parameters θ are trained and updated iteratively multiple times to minimize the loss function and obtain the safety monitoring model.

[0013] The comprehensive analysis module obtains a loss feature value by fusing the rigidity index calculated by the rigidity index analysis module and the loss function of the image feature extraction module;

[0014] The risk assessment module compares the loss characteristic value obtained by the comprehensive analysis module with the damage characteristic value threshold to output a building structure damage instruction.

[0015] Furthermore, the building mechanics parameter acquisition module uses a seismic monitor to collect seismic load values, accurately measures the load value borne by the building structure under the action of an earthquake, and captures the load changes caused by seismic waves of different intensities and frequencies. That is, the seismic wave acceleration time series collected by the seismic monitor is a(t), where t represents time. The displacement d(t) is calculated by integrating the acceleration twice, integrating the acceleration once to obtain the velocity, and integrating the velocity once to obtain the displacement d(t)=∫v(t)dt+d0, where d0 is the initial displacement, and the seismic load value P e The relationship between displacement, velocity and acceleration is: P e =k s d(t)+c s v(t)+m s a(t), where k s is the stiffness coefficient of the structure, c s is the damping coefficient of the structure, m s For the quality of the structure.

[0016] Furthermore, the building mechanics parameter acquisition module collects the ultimate strength through stress sensors and strain sensors to monitor the stress σ(T) and strain ∈(T) states of the building structure under load T to evaluate its ultimate strength. Where k is the strength coefficient, the effect of reaction strain on ultimate strength, and t is the monitoring time period.

[0017] Furthermore, the building mechanics parameter acquisition module collects the steel bar damage degree value by detecting the number of steel bar damages in the building structure and calculating the steel bar damage degree value D r for Where N d is the number of damaged steel bars, N t is the total number of steel bars.

[0018] Furthermore, the rigidity index analysis module comprehensively considers the factors of earthquake load value, ultimate strength and steel bar damage degree value for the calculation of rigidity index, so as to fully reflect the overall rigidity and deformation resistance of the building structure, and calculates the rigidity index RI according to the building mechanics data collected by the building mechanics parameter collection module, that is, Where k is the damage coefficient, C s is the structural coefficient, D r is the steel bar damage degree value, f u is the ultimate strength, P e is the earthquake load value.

[0019] Further, the image feature extraction module inputs the collected concrete surface feature images into the trained safety monitoring model, extracts features through the convolutional layer and pooling layer of the safety monitoring model to obtain multiple feature maps, extracts feature values from the feature maps, and uses the extracted \(i\)-th feature value \(x\) i , and calculates the loss function \(L\) as: where \(n\) represents the number of different types of image features, \(w\) i is the weight coefficient of the \(i\)-th type of image feature, and \(f\) i (\(x\) i ) is the function representation of the \(i\)-th type of image feature.

[0020] Further, the comprehensive analysis module fuses the rigidity index \(RI\) calculated by the rigidity index analysis module and the loss function \(L\) of the image feature extraction module to obtain the loss feature value \(LF\), and its fusion formula is: where \(RI\) min and \(RI\) max are the minimum and maximum values of the rigidity index respectively, \(L\) min and \(L\) max are the minimum and maximum values of the loss function respectively, and \(\omega_1\) and \(\omega_2\) are fusion weight coefficients, and \(\omega_1+\omega_2 = 1\), which are used to adjust the relative importance of the rigidity index and the loss function in the fusion process.

[0021] Further, the risk assessment module compares the loss feature value \(LF\) obtained by the comprehensive analysis module with the damage feature value threshold \(T\), including:

[0022] For the loss feature value exceeding the threshold, that is, \(LF\geq T\), an instruction for damaged building structure is output;

[0023] For the loss feature value not reaching the threshold, that is, \(LF < T\), an instruction for undamaged building structure is output, and at the same time, the threshold \(T\) is adjusted according to factors such as the usage situation, maintenance history, and environmental changes of the building structure for secondary risk assessment, and adjusted to \(T=T_0\times(1 + \alpha\) u \(u+\alpha\) m \(m+\alpha\) e \(e)\), where \(T_0\) is the initial threshold, \(u\), \(m\), and \(e\) are the quantization parameters of the usage situation, maintenance history, and environmental changes respectively, and \(\alpha\) u , \(\alpha\) m , \(\alpha\) e are the adjustment coefficients of the usage situation, maintenance history, and environmental changes respectively, which are used to measure the influence degree of different factors on the threshold adjustment.

[0024] Compared with the prior art, this building structure safety risk assessment system has the following beneficial effects:

[0025] 1. The present invention reflects the mechanical properties and damage degree of the building structure under load by comprehensively considering the mechanical parameters and image characteristics of the building structure, and compares and analyzes the collected concrete surface characteristic images to extract the damage situation and image characteristic values ​​of the building structure surface. The rigidity index and the loss function are integrated through a comprehensive analysis module to obtain the loss characteristic value, thereby realizing a quantitative assessment of the safety risk of the building structure. This comprehensive assessment method not only improves the comprehensiveness and accuracy of the assessment results, but also provides strong data support for subsequent risk control and repair measures.

[0026] 2. This system can monitor various parameters and image information of building structures in real time, and judge the safety risks of building structures in a timely and accurate manner based on the comparison between loss characteristic values ​​and thresholds. This real-time risk warning capability greatly improves the safety of building structures and effectively prevents the occurrence of major safety accidents.

[0027] Other advantages, objectives and features of the present invention will be set forth in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be taught from the practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0029] Figure 1 It is an operation flow chart of a building structure safety risk assessment system;

[0030] Figure 2 Construct a flow chart for the safety monitoring model in a building structure safety risk assessment system. DETAILED DESCRIPTION

[0031] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0032] Embodiment 1

[0033] This embodiment focuses on the application of the building structure safety risk assessment system in large commercial buildings. It elaborates on how the various modules of the system work together under complex working conditions to achieve accurate monitoring and analysis of the mechanical properties and damage conditions of the building structure, and ultimately achieve a comprehensive, real-time and reliable assessment of the safety risks of the building structure, providing a practical scientific basis and technical support for the safety management of large commercial buildings.

[0034] In specific implementation, large commercial buildings have huge scale, complex structural systems and high-density personnel flow. These characteristics determine that they have extremely high requirements for structural safety performance. Real-time monitoring of the performance changes of building structures under various complex working conditions, timely discovery of potential safety hazards, and providing strong support for taking effective prevention and response measures to ensure that large commercial buildings always maintain a safe and reliable operation state throughout their life cycle. First, accurately obtain the dynamic response of the building structure under the action of an earthquake, and deploy high-precision seismic monitors in the building. The seismic monitors are implemented with high frequency and high precision. The acceleration time series a(t) of the earthquake wave is recorded at the same time. When the acceleration is integrated, the velocity v(t) is calculated by one integration. According to the integration principle, v(t) = ∫a(t)dt+v0. Since the building is at rest before the earthquake, the initial velocity v0 is set to 0. The velocity is integrated twice to obtain the displacement d(t). The calculation formula is d(t) = ∫v(t)dt+d0. Similarly, the initial displacement d0 of the building is also 0. Through these integration operations, the motion trajectory and deformation of the building structure under the action of the earthquake can be accurately restored. The earthquake load value P e There is a clear relationship between displacement d(t), velocity v(t) and acceleration a(t), namely, P e =k s d(t)+c s v(t)+m s a(t), where the stiffness coefficient k of the structure s It is an important indicator of the building structure's ability to resist deformation. The damping coefficient c of the structure s It reflects the ability of the building structure to dissipate energy during vibration. The mass of the structure m s It is formed by the cumulative mass of various building materials used in the building structure. By accurately measuring and calculating these parameters, the dynamic changes of seismic load values ​​over time can be accurately calculated, providing key data support for subsequent rigidity index calculations.

[0035] Then, stress sensors and strain sensors are embedded in the main concrete components and steel structure components of the building structure to form a comprehensive and detailed monitoring network. In the concrete components, the sensors can sense the stress and strain state of the concrete under various loads in real time. During the entire service life of the building, these sensors continuously monitor the stress and strain state of the building components under normal use loads and extreme load conditions. According to the calculation formula of the ultimate strength Where k is the strength coefficient, the effect of reaction strain on ultimate strength, and t is the monitoring period. Through long-term data accumulation and analysis, the changing trend of the ultimate strength of the building structure can be accurately grasped, and the possible attenuation or deterioration of the ultimate strength can be discovered in time, which provides an important basis for evaluating the safety of the building structure. For the steel bars in the building structure, the total number of steel bars N can be accurately counted according to the design drawings and construction records of the building structure. t According to the calculation formula of steel bar damage degree value Accurately calculate the damage degree of steel bars to provide reliable data support for assessing the safety of building structures; install high-definition image acquisition equipment on the facade of the building, and set image acquisition points on the surface of the main internal load-bearing structure to monitor the surface condition of the structure under the action of internal loads. These devices are connected to the data processing center via wireless transmission to ensure that images can be transmitted in a timely and stable manner, automatically capture images of the concrete surface, and build a large and rich image sample set While collecting images, each image in the sample set is annotated in detail. The annotation content covers the crack type. By analyzing the crack shape, direction and other characteristics, it is accurately determined whether the crack is horizontal, vertical, oblique crack or other complex cracks; the crack length and width are accurately measured, and the number of cracks is counted to form a complete annotation set L = {l j} j＝1 , providing an accurate reference for subsequent safety monitoring model training and image feature analysis.

[0036] Subsequently, the collected concrete surface feature images are input into the safety monitoring model for image enhancement processing. In the histogram equalization process, for the grayscale range of the image [0, L-1] (where L is the grayscale level), the grayscale histogram h(i) (i = 0, 1, ..., L-1) of the original image is statistically calculated according to the formula (where M is the total number of image pixels) redistributes the grayscale value of the image to make the grayscale distribution of the image more uniform, thereby significantly enhancing the contrast of the image and highlighting the subtle features such as cracks and peeling on the concrete surface. The image after image enhancement processing is sent to the convolution layer and pooling layer of the safety monitoring model for deep feature extraction. In the convolution layer, for the input image I, each convolution layer is combined with the output feature map F of the previous layer through the convolution kernel. l-1Perform convolution operation and add bias term b l , after being processed by the activation function f(·), the output feature map of the current layer is obtained Convolution kernel weight matrix W l In the convolution process, different features of the image are extracted. The pooling layer uses the maximum pooling to downsample the feature map output by the convolution layer. l , the output feature map F′ after pooling l =Pooling(F l ), the pooling operation can effectively reduce the size of the feature map and reduce the amount of calculation while retaining the main feature information of the image. After multiple convolutional layers and pooling layers, the feature map is expanded into a one-dimensional vector and connected to the fully connected layer. The fully connected layer determines the number of neurons according to the task type and data complexity, and calculates Y = f(W fc X+b fc )(where X is the expanded eigenvector, W fc is the fully connected layer weight matrix is the bias term of the fully connected layer, f(·) is the activation function) outputs the final result, extracts the key eigenvalues ​​from the processed feature map, that is, the extracted i-th eigenvalue is x i These eigenvalues ​​include the length, width, depth, distribution density of cracks, area and depth of the peeling area, and its position in the image. Based on these key eigenvalues, the loss function is calculated. Where n represents the number of different types of image features, w i is the weight coefficient of the i-th image feature, reflecting the importance of different image features to the safety assessment of building structures, i (x i ) is the function expression of the i-th image feature. By accurately calculating the loss function, the damage degree of the concrete surface and its impact on the overall safety of the building structure can be quantified.

[0037] According to the rich and accurate data collected by the building mechanics parameter acquisition module, the rigidity index RI is calculated. The calculation formula of the rigidity index is: The ultimate strength f u It is calculated through the long-term monitoring data of the strain sensor, which reflects the maximum bearing capacity of the building structure when it is under load; the steel bar damage degree value D r It is determined by local damage detection, which reflects the damage status of steel bars in the building structure. Steel bar damage will weaken the overall performance of the structure; the earthquake load value P e It is measured and calculated in real time by the seismic monitor when an earthquake occurs. It is the external dynamic load that the building structure bears under the action of an earthquake. The damage coefficient k quantifies the influence of steel bar damage on the rigidity index.s is the structural coefficient. By calculating a rigidity index that accurately reflects the overall stiffness of the building structure and its ability to resist deformation, it provides important structural mechanics performance indicators for subsequent risk assessment. Based on the obtained key characteristic values, according to the loss function calculation formula calculate the loss function. For different types of image features, such as cracks, spalling, holes, etc., according to their respective corresponding characteristic values and carefully determined weight coefficients, accurately calculate their contributions to the loss function. Through the comprehensive calculation of various image features, a loss function value that comprehensively reflects the damage degree of the concrete surface and its impact on the overall safety of the building structure is obtained. The rigidity index RI and the loss function L are fused through the fusion formula to obtain the loss characteristic value LF, where RI min and RI max are the minimum and maximum values of the rigidity index respectively, which are determined through theoretical analysis and simulation calculations of the building structure under different working conditions, reflecting the value range of the building structure's rigidity index. L min and L max are the minimum and maximum values of the loss function respectively, obtained through in-depth analysis and calculations of sample images, reflecting the value range of the loss function under different damage degrees; ω1 and ω2 are fusion weight coefficients, and ω1 + ω2 = 1, which are used to precisely adjust the relative importance of the rigidity index and the loss function in the fusion process. According to factors such as the type of building structure, usage environment, importance, etc., reasonably determine the values of ω1 and ω2 to ensure that the loss characteristic value can accurately reflect the safety status of the building structure.

[0038] Finally, strictly compare the calculated loss characteristic value LF with the preset damage characteristic value threshold T. When LF ≥ T, the system resolutely determines that the building structure is damaged and immediately outputs an instruction indicating that the building structure is damaged. When LF < T, the system determines that the building structure is not damaged and outputs an instruction indicating that the building structure is not damaged. At the same time, adjust the threshold T according to factors such as the usage situation of the building structure (such as quantitative parameters u such as service life, usage frequency, load change situation, etc.), maintenance history (such as quantitative parameters m such as maintenance times, maintenance degree, last maintenance time interval, etc.), environmental changes (such as quantitative parameters e such as temperature change range, humidity change situation, earthquake activity frequency, etc.). The adjustment formula is T = T0 × (1 + α u u + α m m + α e e), where T0 is the initial threshold, and α u , α m , α e are the adjustment coefficients for usage situation, maintenance history, and environmental changes respectively. And according to the secondary evaluation results, adjust the risk assessment threshold in real time to provide real-time data for the safety risk assessment of the building structure.

[0039] In summary, this embodiment fully demonstrates the collaborative working ability and effectiveness of each module of the system through the application of the building structure safety risk assessment system in large commercial buildings. From the acquisition of building mechanics parameters to the extraction of image features, and then to precise calculation and analysis and timely risk assessment decisions, it realizes comprehensive and real-time monitoring and assessment of the structural safety status of large commercial buildings, ensures the accuracy and reliability of the assessment results, and provides a solid technical guarantee for the safe operation of large commercial buildings.

[0040] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

Claims

1. A building structure safety risk assessment system, characterized in that: The system includes a building mechanics parameter acquisition module, a rigidity index analysis module, an image feature extraction module, a comprehensive analysis module, and a risk assessment module; The building mechanics parameter acquisition module is used to collect building mechanics parameters in each monitoring area of ​​the building structure, including seismic load value, ultimate strength and steel bar damage degree value, wherein: Seismic load value collection uses seismographs to monitor the dynamic response of building structures under earthquakes in real time, including acceleration, velocity and displacement parameters, and obtains seismic load values ​​through integral calculations; Ultimate strength acquisition uses stress sensors and strain sensors to monitor the strength changes of building components in real time; Steel bar damage degree value collection is to detect the number of steel bar damages in the building structure and calculate the steel bar damage degree value; The rigidity index analysis module calculates the rigidity index based on the building mechanics data collected by the building mechanics parameter collection module; The image feature extraction module is responsible for collecting the concrete surface feature image and inputting it into the safety monitoring model. After image enhancement processing, the image feature value is obtained, and the loss function is calculated based on the image feature value. The safety monitoring model construction process is as follows:

1. Image sample collection and annotation: Collect concrete surface feature image samples, including images of different crack types, different degrees of damage, and images in normal state, to form a sample set. Annotate the images in the sample set, including crack type, crack length, width, and number, to form an annotation set.

2. Construct a feature extraction network: The convolutional neural network CNN is used as the architecture of the security monitoring model. The network contains multiple convolutional layers, pooling layers and fully connected layers. Convolutional layer: For the input image I, the output feature map F of the lth convolutional layer l ; Pooling layer: for feature map F l , the output feature map F′ after pooling l ; Fully connected layer: expand the feature map processed by multiple convolutional layers and pooling layers into a one-dimensional vector and connect it to the fully connected layer; 3. Model training: Define the model's loss function J(θ), using a combination of the cross entropy loss function and the mean square error loss function, J(θ) = λ1J ce (θ)+λ2J mse (θ), where θ represents the parameters of the feature extraction network, λ1 and λ2 are weight coefficients used to balance the contribution of the two loss functions. ce (θ) is used for classification tasks, i.e., the classification of crack types, and the calculation formula is: C is the number of categories, y ji is the indicator variable, N is the total number of samples, for image I j Belongs to category i when y ji =1, otherwise y ji =0, p ji (θ) is the model prediction image I j The probability of belonging to category i, the mean square error loss function J mse eθ) is used for regression tasks, i.e., the prediction of crack length and width parameters, and the calculation formula is: Where M is the number of regression tasks, z jm For image I j The actual value of the mth regression target, q jm (θ) is the corresponding value predicted by the model. The network parameters θ are trained and updated multiple times to minimize the loss function, and a trained security monitoring model is obtained. The comprehensive analysis module obtains a loss feature value by fusing the rigidity index calculated by the rigidity index analysis module and the loss function of the image feature extraction module; The risk assessment module compares the loss characteristic value obtained by the comprehensive analysis module with the damage characteristic value threshold to output a building structure damage instruction.

2. A building structure safety risk assessment system according to claim 1, characterized in that: The building mechanics parameter acquisition module collects seismic load values ​​by using a seismic monitor to accurately measure the load value borne by the building structure under the action of an earthquake, and captures the load changes caused by seismic waves of different intensities and frequencies. That is, the seismic wave acceleration time series collected by the seismic monitor is a(t), where t represents time. The displacement d(t) is calculated by integrating the acceleration twice, integrating the acceleration once to obtain the velocity, and integrating the velocity once to obtain the displacement d(t)=∫v(t)dt+d0, where d0 is the initial displacement, and the seismic load value P e The relationship between displacement, velocity and acceleration is: P e =k s d(t)+c s v(t)+m s a(t), where k s is the stiffness coefficient of the structure, c s is the damping coefficient of the structure, m s For the quality of the structure.

3. A building structure safety risk assessment system according to claim 1, characterized in that: The building mechanics parameter acquisition module collects the ultimate strength through stress sensors and strain sensors, monitors the stress σ(T) and strain ∈(T) states of the building structure under load T, and evaluates its ultimate strength. Where k is the strength coefficient, the effect of reaction strain on ultimate strength, and t is the monitoring time period.

4. A building structure safety risk assessment system according to claim 1, characterized in that: The building mechanics parameter acquisition module collects the steel bar damage degree value by detecting the number of steel bar damages in the building structure and calculating the steel bar damage degree value D r for Where N d is the number of damaged steel bars, N t is the total number of steel bars.

5. A building structure safety risk assessment system according to claim 1, characterized in that: The rigidity index analysis module comprehensively considers the factors of earthquake load value, ultimate strength and steel bar damage degree value for the calculation of rigidity index, so as to fully reflect the overall rigidity and deformation resistance of the building structure, and calculates the rigidity index RI according to the building mechanics data collected by the building mechanics parameter collection module, that is, Where k is the damage coefficient, C s is the structural coefficient, D r is the steel bar damage degree value, f u is the ultimate strength, P e is the earthquake load value.

6. A building structure safety risk assessment system according to claim 1, characterized in that: The image feature extraction module inputs the collected concrete surface feature image into the trained safety monitoring model, extracts features through the convolution layer and pooling layer of the safety monitoring model, obtains multiple feature maps, extracts feature values ​​from the feature maps, and uses the extracted i-th feature value x i , calculate the loss function L as: Where n represents the number of different types of image features, w i is the weight coefficient of the i-th image feature, f i (x i ) is the function representation of the i-th image feature.

7. A building structure safety risk assessment system according to claim 1, characterized in that: The comprehensive analysis module obtains the loss feature value LF by fusing the rigidity index RI calculated by the rigidity index analysis module and the loss function L of the image feature extraction module. The fusion formula is: Among them, RI min and RI max are the minimum and maximum values ​​of the rigidity index, L min and L max are the minimum and maximum values ​​of the loss function respectively, ω1 and ω2 are the fusion weight coefficients, and ω1+ω2=1 is satisfied, which is used to adjust the relative importance of the rigidity index and the loss function in the fusion process.

8. A building structure safety risk assessment system according to claim 1, characterized in that: The risk assessment module compares the loss characteristic value LF obtained by the comprehensive analysis module with the damage characteristic value threshold T, including: If the loss characteristic value exceeds the threshold, i.e., LF ≥ T, the building structure damage instruction is output; For the loss eigenvalue not reaching the threshold, i.e., LF < T, an instruction indicating that the building structure is not damaged is output. At the same time, the threshold T is adjusted according to factors such as the usage situation, maintenance history, and environmental changes of the building structure for a secondary risk assessment. The adjustment is made to T = T0×(1 + α u u + α m m + α e e), where T0 is the initial threshold, u, m, and e are the quantization parameters of the usage situation, maintenance history, and environmental changes respectively, and α u , α m , α e are the adjustment coefficients of the usage situation, maintenance history, and environmental changes respectively, which are used to measure the influence degree of different factors on the threshold adjustment.

Images