Quantitative Analysis Method for the Influence of Bridge Diseases on Structural Performance

Abstract

The present invention relates to a quantitative analysis method for the influence of bridge diseases on structural performance, comprising the following steps: obtaining bridge information and regular inspection reports of the same type of bridges in a certain area; constructing a bridge structure analysis model according to the bridge information; extracting disease information from the regular inspection reports of the bridges, classifying the diseases into three categories: size attenuation type diseases, stiffness reduction type diseases and non-structural diseases, and the disease information including disease type, location, size and technical condition; at the same time, determining the structural analysis units affected by the diseases according to the disease locations and the node coordinates of the divided structural analysis units; calculating the relevant parameters after the influence of the diseases by using a correction model according to different types of diseases; correcting the element stiffness matrix with the relevant parameters after the influence of the diseases, integrating to obtain the global stiffness matrix S, and further obtaining the strain matrix d of the bridge. It solves the problem of the disconnection between disease information and structural analysis, and fills the gap between traditional disease qualitative assessment and SHM refined analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of civil engineering structure health monitoring and evaluation, and specifically to a quantitative analysis method for the impact of bridge diseases on structural performance, which is used to quantify the impact of bridge diseases on structural performance and realize the reuse of regular inspection data. Background Technique

[0002] With the increase of the service life of bridges, bridge diseases gradually appear and become an important factor threatening the safety of infrastructure. The state monitoring and health assessment of bridges have always been key research fields at home and abroad.

[0003] At present, the regular inspection system is the main method for bridge maintenance management worldwide. This method records and evaluates bridge diseases through manual inspections or equipment such as unmanned aerial vehicles, and rates the diseases according to standardized guidelines (such as AASHTO, the UK "Highway Bridge Inspection Guide", the Chinese "Highway Bridge and Culvert Maintenance Technology Specification", etc.) to evaluate the health status of the bridge. However, the main problems existing in the regular inspection system include: (1) Strong data subjectivity: The assessment process highly depends on the experience of inspection personnel, and the level of disease quantification is low; (2) Lack of quantitative correlation between diseases and performance: Disease information is more used for grade assessment rather than directly analyzing its actual impact on structural stiffness and strength; (3) Low data utilization efficiency: Inspection data is usually not effectively integrated into the bridge information model and is not combined with structural analysis.

[0004] At the same time, the current mainstream analysis methods for regular inspection diseases are limited to qualitative analysis mainly based on experience, including:

[0005] (1) State grade assessment: Usually, the condition of diseases is scored through empirical formulas or grade tables, such as the severity of diseases from level 1 to level 5, for reference in maintenance decisions;

[0006] (2) Disease distribution statistics: Using disease records to generate statistical reports for evaluating the overall state of the bridge;

[0007] (3) Disease trend prediction: Using historical data of diseases to simulate the development trend of diseases and conduct condition prediction;

[0008] (4) Maintenance priority ranking: Determining the bridges and components to be repaired first according to disease scores and the importance of the bridge

[0009] Another type of method for bridge maintenance management during the operation and maintenance stage is the Structural Health Monitoring (SHM) technology. This technology collects structural responses (such as displacement, stress, and vibration) in real time through sensors to evaluate the health status of the bridge structure. Although the SHM system provides high-precision real-time data, the following problems limit its wide application: (1) Limited coverage: Due to the high cost of sensor devices, only a small number of important bridges globally are equipped with SHM systems, and ordinary bridges still mainly rely on manual inspections; (2) Data integrity issues: Sensor failures, environmental interference, etc. lead to missing or distorted data, affecting the evaluation accuracy; (3) Disconnection between disease characteristics and structural performance: SHM data cannot directly reflect disease characteristics, such as information on crack width and spalling range.

[0010] In recent years, scholars at home and abroad have carried out many studies in this field, mainly focusing on the integration of disease information, that is, supporting the standardized integration of disease data through the IFC standard and extended entities. Some scholars have also accurately modeled the impact of diseases through finite element analysis, but this requires the support of high-cost professional analysis software and is difficult to promote to ordinary bridges. Therefore, there is still a lack of an effective conversion method from inspection data to a structural analysis model, especially there is a technical bottleneck in the quantitative analysis of the impact of diseases on the structure. In view of this, there is an urgent need for a method for quantitatively analyzing the impact of bridge diseases on structural performance to overcome the existing deficiencies. Summary of the Invention

[0011] Aiming at the deficiencies of the existing technology, the technical problem to be solved by the present invention is: to provide a method for quantitatively analyzing the impact of bridge diseases on structural performance, solve the problem of the disconnection between disease information and structural analysis, fill the gap between traditional disease qualitative evaluation and SHM refined analysis, and provide an efficient and scalable solution for bridge maintenance management.

[0012] To solve the above technical problems, the present invention adopts the following technical solutions:

[0013] In the first aspect, the present invention provides a method for quantitatively analyzing the impact of bridge diseases on structural performance, and the quantitative analysis method includes the following steps:

[0014] Step 1: Obtain the bridge information and regular inspection reports of the same type of bridges in a certain area to form an optimization analysis database A for disease impact coefficients;

[0015] Step 2: According to the bridge information, construct a bridge structural analysis model based on Matrix-based methods, divide the structural analysis units according to the preset division rules to obtain the node matrix and element matrix, and calculate the element stiffness matrix of the structural analysis units;

[0016] Step 3: Extract the disease information from the regular inspection report of the bridge. Classify the diseases into three categories: size attenuation diseases, stiffness reduction diseases, and non-structural diseases. The disease information includes disease type, location, size, and technical condition;

[0017] Confirm the way it affects the structural performance according to the disease type, that is, size attenuation, stiffness attenuation, or no structural impact;

[0018] At the same time, determine the structural analysis units affected by the disease according to the disease location and the node coordinates of the divided structural analysis units, and calculate the influence of different diseases on the structural characteristics of the structural analysis units;

[0019] Use the correction model according to different types of diseases to calculate the relevant parameters after the disease impact;

[0020] Use the relevant parameters after the disease impact to correct the element stiffness matrix, and integrate the node matrix with the corrected element stiffness matrix to obtain the global stiffness matrix S;

[0021] Step 4: Obtain the load condition P of the bridge. Use P = S·d to obtain the strain matrix d of the bridge. d is the calculation result of the structural response under the disease impact. Further obtain the relevant indicators of the bridge structural response under the disease condition according to d.

[0022] For size attenuation diseases, the reduced member section distance of the relevant structural analysis units affected by them. The relevant parameters after the disease impact are calculated using the following formula:

[0023] I i，z,def =I i，z -∑ M y m，def 2 ·A m,def

[0024] Where, i i，z,def is the section distance of the z-axis of the member of structural analysis unit i after being affected by the size attenuation disease; I i，z is the original section distance of the z-axis of structural analysis unit i, M is the number of size attenuation diseases on structural analysis unit i; y m,def is the distance between the centroid of disease m and structural analysis unit i; A m,def is the area of disease m;

[0025] Non-structural diseases have no significant impact on the overall structural performance and no relevant parameter correction is carried out.

[0026] For stiffness reduction diseases, the reduced member stiffness of the relevant structural analysis units affected by them. The calculation formula for the relevant parameters after the disease impact is:

[0027]

[0028] Among them, k def k is the stiffness of the component after damage; origin is the original stiffness of the component; β n，j is the reduction coefficient of the structural stiffness of the jth technical condition of the nth stiffness reduction defect, j = 1~5; N is the number of all stiffness reduction defects on the component.

[0029] For stiffness reduction type diseases, the stiffness reduction factor β n，j The determination process is:

[0030] Obtain the load conditions of a number of bridges of the same type in a certain area and the strain-related data monitored by SHM sensors.

[0031] At the same time, the locations of all stiffness reduction defects on each bridge and the technical conditions of different defects are obtained, and then the distribution data of the stiffness reduction coefficient of the defect is obtained;

[0032] The above data is used to form a reinforcement learning training database B; a reinforcement learning network is constructed;

[0033] Set the reinforcement learning action rule, i.e., β nj The value selection rule of , where j represents the technical condition level of the disease, taking an integer between 1 and 5, and n takes an integer between 1 and N, where N is the number of all stiffness reduction diseases; β n，j The value range is 0.25~1, set the interval step Δ, the initial state β n，j The value is 0.25, and each action changes β n，j =β n，j +Δ;

[0034] Each action causes the global stiffness matrix S to change, and the strain matrix d of the bridge under the corresponding load condition is obtained according to P = S·d, which is the calculation result of the structural response under the influence of the disease;

[0035] Set the reinforcement learning reward, and calculate the reward by the absolute value of the difference between the calculated result d of the structural response under the influence of the disease and the strain-related data d0 monitored by the SHM sensor, which includes two parts:

[0036] 1) Process reward: Under the current action, Than the previous action Small, reward +1;

[0037] 2) Instant reward: Under the current action, When the error is less than 1%, the reward is +10;

[0038] a t represents the action performed at time t, that is, at time t {βn，j Value of the list; Absolute value of the difference corresponding to the action taken at time t;

[0039] Finally traverse N and j to obtain the optimal {β n，j} combination, and use this {β n，j} combination as the stiffness reduction coefficient in the correction model of the stiffness reduction type disease in step 3, and then calculate the strain matrix d of the bridge.

[0040] Furthermore, the types of stiffness reduction diseases include at least one of structural cracks, steel bar corrosion, interlayer peeling, deflection, etc.

[0041] Furthermore, the calculation formula for the global stiffness matrix is:

[0042]

[0043] Among them, S is the global stiffness matrix, T i is the transformation matrix from the structural analysis unit to the global coordinates, k i is the element stiffness matrix of the structural analysis unit i; T i T represents the transpose matrix of T i ; L is the number of structural analysis units.

[0044] Furthermore, the bridge information is presented in the form of a BIM model; the size of the node matrix is 1*NDOF, where NDOF is the degree of freedom of the entire bridge after dividing the structural analysis units.

[0045] Furthermore, the construction of the bridge structural analysis model based on Matrix-based methods includes the following steps:

[0046] 1) Extract and analyze the global coordinates and relative positioning of the bridge components in the bridge information, decompose the components according to the preset unit size, generate a number of structural analysis units, complete the coordinate transformation, the adjacent structural analysis units are separated by nodes, generate a node matrix with the coordinates of all nodes, and the numbers of all structural analysis units and the corresponding nodes form an element matrix;

[0047] 2) Extract the geometric information of the components, obtain the geometric key characteristics of each structural analysis unit, including geometric key characteristics such as cross-sectional area, moment of inertia, section modulus, length, and geometric center position. The component is composed of several structural analysis units, and establish the mapping between the geometric key characteristics of the structural analysis unit and the element matrix;

[0048] 3) Extract the material information of the components, obtain the mapping relationship between the material elastic modulus E eff of each structural analysis unit and the element matrix;

[0049] 4) Extract the bearing information, determine the node position coordinates corresponding to the bearing and the constraint type of the bearing (unidirectional or bidirectional), and establish the mapping relationship between the bearing and the node matrix;

[0050] 5) Combine the mapping relationship between the bearing and the node matrix to determine the constraint type of the structural analysis element, and then use Matrix-based methods to establish the element stiffness matrix using the elastic modulus of the material information and the geometric information. A structural analysis element has an element stiffness matrix.

[0051] In the second aspect, the present invention provides a quantitative analysis system for the influence of bridge diseases on structural performance, and the system executes the steps of the analysis method.

[0052] Furthermore, the analysis system includes a disease influence coefficient optimization analysis database A, a reinforcement learning training database B, a bridge structure analysis model construction block, a stiffness reduction coefficient optimization module, and a strain matrix calculation module.

[0053] The disease influence coefficient optimization analysis database A is connected to the BIM model, the SHM sensor data module, and the regular inspection report data module;

[0054] The bridge structure analysis model construction block divides the structural analysis elements based on the disease influence coefficient optimization analysis database A and establishes a bridge structure analysis model;

[0055] The reinforcement learning training database B is used to store the strain data in the SHM sensor data module, the load data in the regular inspection report data module, and the disease stiffness reduction coefficient distribution data obtained from the bridge structure analysis model construction block;

[0056] The stiffness reduction coefficient optimization module uses the reinforcement learning training database B to obtain the optimal stiffness reduction coefficient combination through supervised training based on reinforcement learning;

[0057] The strain matrix calculation module uses the optimal stiffness reduction coefficient combination output by the stiffness reduction coefficient optimization module to correct the element stiffness matrix in the structural analysis elements of the bridge structure analysis model, and further calculates the strain matrix d of the bridge.

[0058] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0059] 1) The analysis method of the present invention can scientifically quantify the impact of diseases on the structural performance of bridges, effectively solve the problem of the disconnection between disease information and structural analysis, fill the gap between traditional qualitative disease assessment and refined SHM analysis, and provide an efficient and scalable solution for bridge maintenance management. The traditional regular inspection system relies on expert experience to rate diseases. Although the sensor-based health monitoring technology can obtain bridge response data in real time, its coverage and applicability are limited and it is only applicable to a few high-value bridges.

[0060] 2) The present invention integrates regular inspection data, combines with the bridge structural analysis model, converts disease information into quantifiable stiffness reduction coefficients or cross-section characteristic changes in the structural analysis model, and optimizes and fits the coefficients of disease-reduced stiffness through reinforcement learning technology, so that these data are no longer only used for disease grade assessment, but become important inputs directly driving structural performance assessment, quantifying the impact of diseases on the geometric and material properties of bridges into changes in structural stiffness and internal forces, and greatly improving the value of existing regular inspection data. This method not only avoids the errors of subjective assessment, but also can dynamically reflect the specific impact of the disease location, size and type on the overall structural performance.

[0061] 3) In addition, this method supports automatically extracting key information such as geometry, materials, bearings and loads from the bridge BIM model, reducing manual operations in the process of establishing the analysis model and ensuring the integrity and consistency of data. At the same time, through the adoption of matrix analysis methods, efficient structural performance calculations are realized, providing engineers with key performance indicators including bridge displacement, internal force distribution and stress concentration in the disease area, intuitively reflecting the impact of diseases on the safety and durability of bridges, helping engineers quickly identify potential dangerous areas, and formulating scientific repair and reinforcement plans. The method of the present invention has higher efficiency and stronger applicability, and is particularly suitable for the health status assessment needs of large-scale bridge networks. Brief Description of the Drawings

[0062] Figure 1 Schematic diagram of the structure of the bridge BIM model and the structural analysis unit for the embodiment. Detailed Description of the Specific Embodiment

[0063] The present invention will be described in detail below in conjunction with the drawings and embodiments, but this is not used as a limitation to the protection scope of the present application.

[0064] The method for quantitatively analyzing the impact of bridge diseases on structural performance of the present invention, the steps of the method are as follows:

[0065] Step 1: Obtain bridge information, regular inspection reports and SHM sensor data of the same type of bridges in a certain area to form an optimized analysis database A of disease impact coefficients;

[0066] Among them, the same type means that the materials, structural systems, etc. of the main components of the bridge are the same.

[0067] Step 2: According to the bridge information, construct a bridge structure analysis model based on Matrix-based methods, divide the structural analysis units according to the preset division rules to obtain the node matrix and element matrix, and calculate the element stiffness matrix of the structural analysis units.

[0068] Step 3: Extract the disease information from the regular inspection report of the bridge. Classify the diseases into three categories: size attenuation type diseases, stiffness reduction type diseases, and non-structural diseases. The disease information includes disease type, location, size, and technical condition (rating for each disease), etc.

[0069] Confirm the way of its influence on the structural performance according to the disease type, that is, size attenuation, stiffness attenuation, or no structural influence.

[0070] Meanwhile, determine the structural analysis units affected by the disease according to the disease location and the node coordinates of the divided structural analysis units, and calculate the influence of different diseases on the structural characteristics of the structural analysis units.

[0071] Calculate the relevant parameters after the disease influence by using the correction model according to different types of diseases.

[0072] Correct the element stiffness matrix with the relevant parameters after the disease influence, and obtain the global stiffness matrix S by integrating the node matrix and the corrected element stiffness matrix.

[0073] Step 4: Obtain the load condition P of the bridge, use P = S·d to obtain the strain matrix d of the bridge, d is the calculation result of the structural response under the influence of the disease, and further obtain the relevant indicators of the bridge structural response under the disease condition according to d.

[0074] Among them, the relevant indicators are used for structural performance evaluation and maintenance decision-making, and are integrated into the original bridge BIM model.

[0075] In Step 3, according to the disease type information extracted from the regular inspection report, it is divided into the following three categories according to the existing rules:

[0076] 1) Size attenuation type diseases. Such diseases mainly cause a reduction in the effective size of structural members, such as spalling, exposed reinforcement, etc. The analysis model uses cross-section re-modeling technology, and uses OpenCV tools to calculate the changes in the effective cross-sectional area and moment of inertia to form new cross-sectional characteristics.

[0077] 2) Stiffness reduction type diseases. Such diseases will weaken the material stiffness or the stiffness of the overall component, such as cracks and local damage, etc. The analysis model quantifies the influence of crack width and depth on the component stiffness by introducing a stiffness reduction coefficient.

[0078] 3) Non-structural diseases, which have no significant impact on the overall performance of the structure, such as surface weathering, slight corrosion, etc. Only record the location, size and shape of the diseases as the information for visual presentation to assist engineering decision-making.

[0079] According to the description of the disease location, convert the structural analysis unit affected by it and its position in this structural analysis unit, calculate the impact of the disease on this structural analysis unit, and correct the structural characteristics of this unit (such as sectional distance, stiffness). According to different disease types, select the corresponding correction model for the structural impact. Among them, for diseases with size attenuation, the sectional distance of the relevant unit affected by it after reduction, and the relevant parameters after the disease impact are calculated by the following formula:

[0080] I i，z,def =I i，z -∑ M y m,def 2 ·A m，def

[0081] Among them, I i，z,def is the sectional distance of the z-axis of the component of the structural analysis unit i after considering the impact of the disease with size attenuation; I i，z is the original sectional distance of the z-axis of the structural analysis unit i, M is the number of diseases with size attenuation on the structural analysis unit i; y m,def is the distance between the centroid of the disease m and the structural analysis unit i; A m,def is the area of the disease m;

[0082] The above non-structural diseases are integrated into the bridge information model as additional information and used as a reference for structural performance evaluation, but do not participate in the calculation of stiffness or strength.

[0083] For diseases with stiffness reduction, the stiffness of the relevant unit affected by it after reduction, and the calculation formula for the relevant parameters after the disease impact is:

[0084]

[0085] Among them, k def is the stiffness of the component after the disease; k origin is the original stiffness of the component; β n，j is the reduction coefficient of the j-th technical condition of the n-th stiffness reduction disease on the structural stiffness, j = 1 to 5; N is the number of all stiffness reduction diseases on this component.

[0086] Summarize the unit stiffness matrix after correcting the disease impact, and automatically calculate the global stiffness matrix of the whole bridge. The calculation formula of the global stiffness matrix is:

[0087]

[0088] Among them, S is the global stiffness matrix, and T i is the transformation matrix from the structural analysis element to the global coordinates (obtained by using the node matrix transformation), and k i is the element stiffness matrix of the i-th structural analysis element; T i T represents the transpose matrix of T i ; L is the number of structural analysis elements.

[0089] For the disease of stiffness reduction type, the determination process of its stiffness reduction coefficient β n，j is as follows:

[0090] Obtain the load conditions of several bridges of the same type in a certain area and the strain-related data monitored by SHM sensors,

[0091] and at the same time obtain the positions and technical conditions of all stiffness reduction type diseases on each bridge, and then obtain the distribution data of the disease stiffness reduction coefficient;

[0092] Construct the reinforcement learning training database B with the above data; construct the reinforcement learning network;

[0093] Set the reinforcement learning action rule, that is, the value rule of β n，j , where j represents the technical condition level of the disease, taking an integer between 1 and 5, and n takes an integer between 1 and N, where N is the number of all stiffness reduction type diseases; β n，j The value range is 0.25~1, set the interval step size Δ, and the initial state β n，j takes the value of 0.25 for all. Each time the action changes, β n，j =β n，j +Δ;

[0094] Each time the action causes the global stiffness matrix S to change, the strain matrix d of the bridge under the corresponding load condition is obtained according to P = S·d, which is the calculation result of the structural response under the influence of the disease;

[0095] Set the reinforcement learning reward, and calculate the reward with the absolute value distance of the difference between the calculation result d of the structural response under the influence of the disease and the strain-related data d0 monitored by the SHM sensor, including two parts:

[0096] 1) Process reward: Under the current action, is smaller than that of the previous action, the reward is +1;

[0097] 2) Immediate reward: Under the current action, is less than 1% error, the reward is +10;

[0098] a t represents the action taken at time t, that is, at time t {βn，j the value taken from the list where \(n = 1, 2, \cdots, N\) and \(j = 1, 2, \cdots, 5\); the absolute value of the difference corresponding to the action taken at time \(t\);

[0099] Finally, traverse \(N\) and \(j\) to obtain the optimal \(\{\beta\) n，j}\) combination. Using this \(\{\beta\) n，j} combination as the stiffness reduction coefficient in the correction model of the stiffness reduction type disease in step 3, and then calculate the strain matrix \(d\) of the bridge.

[0100] For the reinforcement learning network, the environment is the bridge structure analysis model of all bridges in the disease impact coefficient optimization analysis database \(A\), and the action is the attenuation coefficient \(\beta\) for different types of stiffness reduction type diseases and different disease technical condition levels on the same bridge n，j value groups, forming a \(\{\beta\) n，j} list. Use Matrix - based methods to calculate the bridge structure response under disease conditions, and the reward is set based on the absolute value of the difference between the calculated result \(d\) of the structure response under disease impact and the structure response \(d_0\) monitored by the SHM sensor. Through supervised learning, finally obtain the optimized \(\{\beta\) n，j} combination.

[0101] The construction of the bridge structure analysis model based on Matrix - based methods includes the following steps:

[0102] 1) Extract and analyze the global coordinates and relative positions of bridge components in the bridge information, decompose the components according to a pre - set unit size to generate a number of structural analysis units, complete coordinate transformation. Adjacent structural analysis units are separated by nodes, generate a node matrix with the coordinates of all nodes, and a unit matrix composed of all structural analysis unit numbers and corresponding nodes;

[0103] 2) Extract the geometric information of the components to obtain the geometric key characteristics of each structural analysis unit, including geometric key characteristics such as cross - sectional area, moment of inertia, section modulus, length, and geometric center position. The component is composed of several structural analysis units, and establish the mapping between the geometric key characteristics of the structural analysis unit and the unit matrix;

[0104] 3) Extract the material information of the components to obtain the mapping relationship between the material elastic modulus \(E\) eff of each structural analysis unit and the unit matrix;

[0105] 4) Extract the bearing information, determine the node position coordinates corresponding to the bearing and the constraint type of the bearing (unidirectional or bidirectional), and establish the mapping relationship between the bearing and the node matrix;

[0106] 5) Determine the constraint type of the structural analysis element in combination with the mapping relationship between the bearing and the node matrix. Then, use Matrix-based methods to establish the element stiffness matrix using the elastic modulus of the material information and the geometric information. Each structural analysis element has an element stiffness matrix.

[0107] In step 2, the bridge information is presented in the form of a BIM model. The conversion process from bridge BIM to the structural analysis model includes extracting geometric information, material properties, and bearing conditions from the bridge BIM model, and converting the information into the structural analysis model through the IFC standard. Use the Ifcopenshell tool to parse the IFC file of the bridge BIM model, specifically including:

[0108] (1) Coordinate transformation: Extract the bridge components (IfcElement) located in IfcBridge through the spatial containment relationship. Use IfcLocalPlacement and IfcAxis2Placement3D to parse the global coordinates and relative positioning of the components. Decompose the components according to the preset element size to generate structural analysis elements, and store them using the IfcStructuralMember entity;

[0109] (2) Extraction and conversion of geometric information: For the components modeled by solid modeling, obtain the specific cross-sectional shape and size of the components by extracting the IfcExtrudedAreaSolid entity. Use the OpenCV tool to convert the cross-sectional geometric information into the cross-sectional characteristics required for engineering analysis (such as geometric key characteristics including cross-sectional area, moment of inertia, section modulus, length, and geometric center position), and store them in the structural analysis model. The calculation method of the moment of inertia is as follows:

[0110] I z =∫ A y 2 dA

[0111] I y =∫ A z 2 dA

[0112] Where I z and I y correspond to the moments of inertia about the z-axis and y-axis respectively.

[0113] (3) Extraction and conversion of material information: Find the material information related to components through the IfcRelAssociatesMaterial entity, and clarify key parameters such as material strength (e.g., concrete C55, steel bar strength grade), elastic modulus, and Poisson's ratio based on rules. For components with multi-material combinations (such as prestressed beams), comprehensively calculate the effective material stiffness and store the information in the structural analysis model.

[0114] (4) Locate the support positions through the IfcElementAssembly entity related to the foundation, use the IfcRelConnectsStructuralMember entity to parse the support constraint types (such as fixed, hinged, or sliding), and convert the support conditions into the input of the nodal constraint matrix.

[0115] For the above-mentioned structural analysis units of the bridge, combine the geometric information and material information, and use the matrix method to establish the element stiffness matrix.

[0116] Example 1

[0117] Please refer to Figure 1 , the division of the structural analysis units of the bridge in the example is as Figure 1 shown. The case bridge is a five-span continuous rigid-frame girder bridge with a span layout of 58 + 182 + 265 + 194 + 70 m. The main span crosses the Red River, the bridge deck width is 22.5 m, and a variable cross-section box girder design is adopted. The materials for the superstructure are C55 concrete and 15.24 mm prestressed steel strands, and the support types include fixed supports and sliding supports. The bridge has been regularly inspected every year since 2013, and traditional manual records of disease information (such as crack width, length, etc.) are used.

[0118] Step 1: Collect the regular inspection reports and SHM sensor data of the same type of bridges in this area, and construct the disease influence coefficient optimization analysis database A. In the example, relevant data of all prestressed concrete elevated continuous girder bridges in XX Province are selected to construct the disease influence coefficient optimization analysis database.

[0119] Step 2: Based on the BIM model of the bridge information, construct the bridge structural analysis model based on Matrix-based methods.

[0120] In this example, the main body of the bridge is a variable cross-section box girder. According to the variable cross-section characteristics, the beam body is divided into structural analysis units with lengths ranging from 2 to 5 m; the piers of the lower structure of the bridge are divided into structural analysis units with a length of 5 m for each section, and the nodal coordinates (x, z) of each structural analysis unit are written into the nodal matrix.

[0121] In an embodiment, for geometric information, the bridge main body adopts a Brep geometric model (Boundary Representation), which is a common model for describing complex bridge cross-sections in the IFC standard. Its characteristic is to define geometric shapes with boundary information of vertices, edges, and faces. Specifically, in the first step, geometric data parsing is performed. First, an IfcFacetedBrep entity associated with the bridge component is obtained from IfcShapeRepresentation, and its boundary definition is parsed. Secondly, the IfcFaces under IfcClosedShell are traversed, and the planes or surfaces forming the cross-section are sequentially extracted. Finally, the three-dimensional coordinates of the vertices are obtained using IfcCartesianPoint, and a polygon of the cross-section boundary is formed through IfcPolyLoop. The second step is to perform cross-section contour projection. First, the local coordinate system where the cross-section is located is determined according to IfcLocalPlacement. Secondly, the cross-section boundary of Brep is projected onto the cross-sectional plane to form a two-dimensional plane contour. Finally, the closed path of the cross-section projection (such as a polygon or a complex curve) is determined. In this embodiment, multiple IfcFaces of the bridge box girder are extracted and the inner and outer contours are distinguished (such as the outer frame and holes of the box girder), and a contour hierarchy is established. In the third step, key characteristics of the cross-section are obtained through geometric calculations, including geometric information such as cross-sectional area, moment of inertia, and geometric center position.

[0122] In an embodiment, for the extraction and conversion of material information, first, the material entity related to the component is found through the IfcRelAssociatesMaterial entity (such as IfcMaterial or IfcMaterialLayerSetUsage). For a single material (such as IfcMaterial), its associated IfcMaterialProperties are directly extracted, including elastic modulus (E), Poisson's ratio (ν), density (ρ), etc. For a composite material (such as prestressed reinforced concrete), the order, thickness, and properties of the material layers are obtained from IfcMaterialLayerSetUsage, and the equivalent material elastic modulus E is calculated in combination with the thickness ratio of the material layers. eff :

[0123]

[0124] where E i and h i are the elastic modulus and thickness of the i-th layer of material respectively, and n is the number of layers of the composite material. The E eff (or the equivalent material properties of the composite material) of the material is associated with the structural analysis unit.

[0125] Combined with the mapping relationship between the bearing and the node matrix, determine the constraint type of the structural analysis unit. Then, use Matrix-based methods to establish the element stiffness matrix using the elastic modulus of the material information and the geometric information. One structural analysis unit has one element stiffness matrix.

[0126] In this embodiment, for a structural analysis unit with 2 non-fixed nodes (i.e., neither end of the structural analysis unit is constrained by a bearing), the calculation formula for its element stiffness matrix k is as follows:

[0127]

[0128] Where E, I, and L are the elastic modulus, section modulus, and length of the structural analysis unit, respectively.

[0129] Step 3: Extract the disease information from the regular inspection report of the bridge. Classify the diseases into three categories: size attenuation type diseases, stiffness reduction type diseases, and non-structural diseases. The disease information includes disease type, location, size, and technical condition.

[0130] Confirm the way it affects the structural performance according to the disease type, that is, size attenuation, stiffness attenuation, or no structural impact.

[0131] At the same time, determine the structural analysis units affected by the disease according to the disease location and the node coordinates of the divided structural analysis units, and calculate the influence of different diseases on the structural characteristics of the structural analysis unit.

[0132] Use the correction model according to different types of diseases to calculate the relevant parameters after the disease impact.

[0133] Use the relevant parameters after the disease impact to correct the element stiffness matrix, and integrate the node matrix and the corrected element stiffness matrix to obtain the global stiffness matrix S.

[0134] In this embodiment, extract the disease type information from the regular inspection report, covering three categories:

[0135] 1) Size attenuation type diseases, including spalling, exposed reinforcement, and reticulated cracks;

[0136] 2) Stiffness reduction type diseases, including structural cracks;

[0137] 3) Non-structural diseases, including surface weathering.

[0138] In this embodiment, according to the location description of the disease, convert the affected structural analysis unit and its position in the structural analysis unit.

[0139] In this embodiment, the effects of various diseases on the structural analysis unit are calculated, and the structural characteristics of the unit (such as section distance and stiffness) are corrected. Among them, for size attenuation diseases (such as spalling, steel bar exposure, and reticulated cracks), the calculation formula for the reduced member section distance of the relevant unit affected is as follows:

[0140] I i，z,def = I i，Z - ∑ M y m,def 2 ·A m，def

[0141] Where, I i，z,def is the section distance of the member's z-axis of the structural analysis unit i after considering the influence of size attenuation diseases; I i，z is the original z-axis section distance of the structural analysis unit i, M is the number of size attenuation diseases on the structural analysis unit i; y m，def is the distance between the disease m and the centroid of the structural analysis unit i; A m,def is the area of the disease m;

[0142] For stiffness reduction diseases (such as structural cracks, steel bar corrosion, delamination, deflection, etc.), the calculation formula for the reduced member stiffness of the relevant unit affected is as follows:

[0143]

[0144] Where, k def is the member stiffness after diseases; k origin is the original stiffness of the member; β n，j is the stiffness reduction coefficient of the jth technical condition of the nth stiffness reduction disease on the structural stiffness, j = 1 - 5; N is the number of all stiffness reduction diseases on the member.

[0145] In this embodiment, there is one structural crack with a technical condition level of 2 (tentatively n = 1) on the structural analysis unit (18). Accordingly,

[0146] k def (18) = k origin (18) ·(1 - β 1，2 )

[0147] Summarize the unit stiffness matrix corrected by the influence of diseases, and automatically calculate the global stiffness matrix of the entire bridge.

[0148] Step 4: Obtain the load condition P of the bridge, and use P = S - d to obtain the strain matrix d of the bridge. d is the calculation result of the structural response under the influence of diseases. Further obtain the relevant indicators of the bridge structural response under disease conditions according to d.

[0149] In this embodiment, (1) a reinforcement learning environment is set up, that is, all bridges in the database A for optimizing the analysis of disease impact coefficients;

[0150] (2) Set the reinforcement learning action rules, that is, the value-taking rules of β n，j , where j represents the technical condition level of the disease, taking an integer between 1 and 5, and n takes an integer between 1 and N, where N is the number of all stiffness reduction diseases; β n，j The value range is 0.25 to 1, set the interval step size Δ, and the initial state β n，j The values are all 0.25, and each action changes β n，j = β n，j + Δ;

[0151] (3) Calculate the initial state of the environment, that is, under the initial {β n，j}, calculate the initial structural response of the bridge. If there is already load and load combination information in the bridge BIM model, use the Ifcopenshell tool to identify the IfcStructuralAction or IfcStructura1LoadGroup entities from the current bridge BIM model, and extract the defined load types and magnitudes (such as dead load, live load, wind load, etc.) and load combination rules; if the load-related information is not defined, manually input through the interactive interface, and finally form the load vector P.

[0152] Specifically, in this embodiment, the load types of the bridge consider two major categories: self-weight load and vehicle live load. Among them, the live load is set according to the Chinese Highway Bridge Design Code (JTG 3362-2018) to simulate the vehicle load, including a uniform load of 10 kN / m 2 applied to the lane range, and a concentrated load of 200 kN acts on the maximum bending moment position of each span respectively. The calculation method of the load condition is as follows:

[0153] Load condition = G + ψ·Q L

[0154] where is the self-weight dead load, ψ is the vehicle live load coefficient, and Q L is the vehicle live load.

[0155] After determining the load condition, the following formula is used for the analysis of the disease structure, and finally the strain matrix d of the bridge under the corresponding load condition is obtained.

[0156] P = S·d

[0157] Combined with the stiffness matrix k of each unit i and the strain matrix d, the performance indexes such as the internal force of the structure at any position can be further calculated.

[0158] (4) Set the reinforcement learning reward. In this embodiment, the reward is calculated based on the absolute value of the difference distance between the calculated result d of the structural response under the disease and the strain-related data d0 monitored by the SHM sensor, which includes two parts:

[0159] 1) Process reward: Under the current action, is smaller than that of the previous action, the reward is +1;

[0160] 2) Immediate reward: Under the current action, when it is less than 1% error, the reward is +10;

[0161] a t represents the action taken at time t, that is, the value taken from the list of {β n，j} at time t; is the absolute value of the difference corresponding to the action taken at time t;

[0162] (5) Train the reinforcement learning network. In this embodiment, a preset action is applied to the environment, and step (3) is repeated to calculate and update the environmental state, and rewards are issued according to the environmental state and the preset reward rules. After the training is completed, the optimal β combination is obtained, which is used for the disease quantification analysis of all similar bridges in this area. A set of {β n，j} lists output by the reinforcement learning process of this application correspond to the stiffness reduction coefficients of all stiffness reduction type diseases on this bridge, and the data on one bridge are trained together. In this embodiment, the final optimized value of β 1，2 is 0.6. In the actual case verification, the difference between the measurement result of the mid-span deflection of the bridge by the SHM sensor system of the bridge and the disease analysis result is 0.6%.

[0163] In summary, the quantification analysis method of the present invention can scientifically quantify the impact of diseases on the structural performance of bridges, effectively solve the problem of the disconnection between disease information and structural analysis, fill the gap between traditional disease qualitative evaluation and SHM refined analysis, and provide an efficient and scalable solution for bridge maintenance management.

[0164] The parts not described in the present invention are applicable to the prior art.

Claims

1. A quantitative analysis method for the influence of bridge diseases on structural performance, characterized in that, The quantitative analysis method includes the following steps: Step 1: Obtain the bridge information and regular inspection reports of the same type of bridges in a certain area to form the disease influence coefficient optimization analysis database A; Step 2: According to the bridge information, construct a bridge structure analysis model, divide the structural analysis units according to the preset division rules, obtain the node matrix and element matrix, and calculate the element stiffness matrix of the structural analysis units; Step 3: Extract the disease information from the regular inspection reports of the bridges, classify the diseases into three categories: size attenuation type diseases, stiffness reduction type diseases, and non-structural diseases. The disease information includes disease type, location, size, and technical condition; Confirm the way of its influence on the structural performance according to the disease type, that is, size attenuation, stiffness attenuation, or no structural influence; At the same time, determine the structural analysis units affected by the disease according to the disease location and the node coordinates of the divided structural analysis units, and calculate the influence of different diseases on the structural characteristics of the structural analysis units; Use the correction model according to different types of diseases to calculate the relevant parameters after the disease influence; Use the relevant parameters after the disease influence to correct the element stiffness matrix, and integrate the node matrix and the corrected element stiffness matrix to obtain the global stiffness matrix S; The calculation formula of the global stiffness matrix is: Among them, S is the global stiffness matrix, T i is the transformation matrix from the structural analysis element to the global coordinates, k i is the element stiffness matrix of the i-th structural analysis element; T i T represents the transpose matrix of T i ; L is the number of structural analysis elements; Step 4: Obtain the load condition P of the bridge, use P = S·d to obtain the strain matrix d of the bridge, d is the calculation result of the structural response under the disease influence, and further obtain the relevant indicators of the bridge structural response under the disease condition according to d; For the size attenuation type disease, the reduced member section distance of the relevant structural analysis units affected by it, the relevant parameters after the disease influence are calculated by the following formula: I i,z,def = I i,z - ∑ M y m,def 2 · A m,def Among them, I i,z,def is the section moment of the z-axis of the component of the structural analysis unit i after being affected by the size attenuation type disease; I i,z is the original section moment of the z-axis of the structural analysis unit i, M is the number of size attenuation type diseases on the structural analysis unit i; y m,def is the distance between the centroid of the disease m and the structural analysis unit i; A m,def is the area of the disease m; Non-structural diseases have no significant influence on the overall structural performance, and no relevant parameter correction is carried out; For the stiffness reduction type disease, the reduced member stiffness of the relevant structural analysis units affected by it, the calculation formula of the relevant parameters after the disease influence is: Among them, k def is the stiffness of the component after the disease; k origin is the original stiffness of the component; β n,j is the reduction coefficient of the j-th technical condition of the n-th stiffness reduction disease on the structural stiffness, j = 1 to 5; N is the number of all stiffness reduction diseases on the component.

2. The method according to claim 1, wherein For the stiffness reduction type disease, its stiffness reduction coefficient β n,j The determination process is as follows: Obtain the load conditions of a certain number of bridges of the same type in a certain area and the strain-related data monitored by SHM sensors; At the same time, obtain the locations of all stiffness reduction type diseases on each bridge and the technical conditions of different diseases, and then obtain the disease stiffness reduction coefficient distribution data; Construct the reinforcement learning training database B with the above data; construct the reinforcement learning network; Set the reinforcement learning action rule, i.e., β n,j The value-taking rule, where j represents the technical condition level of the disease, taking an integer between 1 and 5, n takes an integer between 1 and N, where N is the number of all diseases with stiffness reduction; β n,j The value range is 0.25 to 1, set the interval step size △, and the initial state β n,j The values are all 0.25, and each action change β n,j = β n,j + △; Each time an action causes the global stiffness matrix S to change, the strain matrix d of the bridge under the corresponding load condition is obtained according to P = S·d, which is the calculation result of the structural response under the disease influence; Set the reinforcement learning reward, and calculate the reward according to the absolute value distance of the difference between the calculation result d of the structural response under the disease influence and the strain-related data d0 monitored by the SHM sensor, including two parts: 1) Process Reward: In the current action, compared with that of the previous action is smaller, reward +1; 2) Immediate Reward: Under the current action, When the error is less than 1%, the reward is +10; a t represents the action taken at time t, that is, the value taken from the list of {β n,j} at time t; is the absolute value of the difference corresponding to the action taken at time t; Finally, traverse N and j to obtain the optimal {β n,j} combination, and use this {β n,j} combination as the stiffness reduction coefficient in the correction model of the stiffness reduction type disease in step 3, and then calculate the strain matrix d of the bridge.

3. The method according to claim 1, characterized in that, The types of stiffness reduction diseases include at least one of structural cracks, steel bar corrosion, interlayer peeling, and deflection.

4. The quantitative analysis method according to claim 1, wherein The bridge information is presented in the form of a BIM model; the size of the node matrix is 1*NDOF, where NDOF is the degree of freedom of the whole bridge after dividing the structural analysis units.

5. The quantitative analysis method according to claim 1, characterized in that The construction of the bridge structure analysis model includes the following steps: 1) Extract and analyze the global coordinates and relative positioning of bridge components in bridge information, decompose the components according to a pre-set unit size to generate a number of structural analysis units, complete coordinate transformation, and divide adjacent structural analysis units by nodes. Generate a node matrix with the coordinates of all nodes, and a unit matrix composed of all structural analysis unit numbers and corresponding nodes; 2) Extract the geometric information of the components to obtain the geometric key characteristics of each structural analysis unit, including geometric key characteristics such as cross-sectional area, moment of inertia, section modulus, length, and geometric center position. The component is composed of several structural analysis units, and establish the mapping between the geometric key characteristics of the structural analysis unit and the unit matrix; 3) Extract the material information of the components to obtain the elastic modulus E of the material for each structural analysis unit eff Mapping relationship with the element matrix; 4) Extract the bearing information, determine the node position coordinates corresponding to the bearing and the constraint type of the bearing (unidirectional or bidirectional), and establish the mapping relationship between the bearing and the node matrix; 5) Combine the mapping relationship between the bearing and the node matrix to determine the constraint type of the structural analysis unit, and then use Matrix-based methods to establish the element stiffness matrix using the elastic modulus of the material information and the geometric information. One structural analysis unit has one element stiffness matrix.

6. A quantitative analysis system for the influence of bridge diseases on structural performance, characterized in that, The system executes the steps of the analysis method according to any one of claims 1-5.

7. The analysis system according to claim 6, wherein The analysis system includes a disease influence coefficient optimization analysis database A, a reinforcement learning training database B, a bridge structure analysis model construction block, a stiffness reduction coefficient optimization module, and a strain matrix calculation module. The disease influence coefficient optimization analysis database A is connected to the BIM model, the SHM sensor data module, and the regular inspection report data module; The bridge structure analysis model construction block divides structural analysis units based on the disease influence coefficient optimization analysis database A to establish a bridge structure analysis model; The reinforcement learning training database B is used to store the strain data in the SHM sensor data module, the load data in the regular inspection report data module, and the disease stiffness reduction coefficient distribution data obtained from the bridge structure analysis model construction block; The stiffness reduction coefficient optimization module uses the reinforcement learning training database B to obtain the optimal stiffness reduction coefficient combination through supervised training based on reinforcement learning; The strain matrix calculation module uses the optimal stiffness reduction coefficient combination output by the stiffness reduction coefficient optimization module to correct the element stiffness matrix in the structural analysis unit of the bridge structure analysis model, and further calculates the strain matrix d of the bridge.

Images