A deep learning-based concrete box girder defect automatic detection method and system

Abstract

The application discloses a kind of based on deep learning's concrete box girder defect automatic detection method and system, the method includes: the visible light image of box girder, laser point cloud and acoustic emission signal are acquired, and data fusion operation is carried out, generates new box girder data, sets up defect detection model, and according to the new box girder data, crack and damage on box girder are identified;Risk assessment model is set, and according to the new box girder data, crack and damage on box girder, the risk index of box girder is calculated;A plurality of preset risk classification thresholds are set, and the risk index is combined to classify the defects of box girder, so as to complete defect detection.

Description

Technical Field

[0001] This invention belongs to the field of box girder defect detection technology, and more specifically, relates to an automatic detection method and system for concrete box girders based on deep learning. Background Technology

[0002] Traditional methods for detecting defects in concrete box girders rely on visual inspection or a single sensor (such as ultrasound or infrared), which have the following problems:

[0003] Inefficient: Manual inspection requires scaffolding, and the inspection of a single span box girder takes 4-6 hours;

[0004] High rate of missed detection: The manual identification rate of hidden defects such as crack width <0.2mm and internal voids is less than 65%;

[0005] Highly subjective: The test results rely on the engineer's experience, and the evaluation of the same defect by different people can vary by up to ±30%;

[0006] Data isolation: There is a lack of ability to integrate and analyze multi-source data such as images, sound waves, and 3D point clouds.

[0007] Existing automated equipment (such as wall-climbing robots) only achieves data collection, but does not solve the problem of intelligent defect diagnosis and quantitative assessment. Summary of the Invention

[0008] To address the above technical problems, this invention proposes an automatic defect detection method for concrete box girders based on deep learning, comprising:

[0009] The visible light image, laser point cloud, and acoustic emission signal of the box girder are acquired, and data fusion is performed to generate new box girder data. A defect detection model is set up, and cracks and damage on the box girder are identified based on the new box girder data.

[0010] Set up a risk assessment model and calculate the risk index of the box girder based on the new box girder data, cracks and damage on the box girder;

[0011] Multiple preset risk classification thresholds are set, and the box girder is classified into defect risks based on the risk index, thereby completing defect detection.

[0012] Furthermore, the defect detection model is the CrackNet-3D model.

[0013] Furthermore, the risk assessment model includes:

[0014]

[0015] Among them, R t+1 R is the risk index at time t+1. tLet α be the risk index at time t, Δt be the time step, α be the first adjustment factor of the risk assessment model, C(t) be the crack propagation rate at time t, β be the second adjustment factor of the risk assessment model, γ be the third adjustment factor of the risk assessment model, S(t) be the stress concentration factor at time t, δ be the fourth adjustment factor of the risk assessment model, E(t) be the environmental corrosion factor at time t, κ be the fifth adjustment factor of the risk assessment model, λ be the sixth adjustment factor of the risk assessment model, μ be the seventh adjustment factor of the risk assessment model, σ be the eighth adjustment factor of the risk assessment model, and η be the ninth adjustment factor of the risk assessment model.

[0016] Furthermore, the crack propagation rate C(t) at time t includes:

[0017]

[0018] Where C0 is the initial crack propagation rate, ω1 is the first weight of the crack propagation rate, Φ(t) is the crack-induced potential energy at time t, ω2 is the second weight of the crack propagation rate, and v is the adjustment factor of the crack propagation rate.

[0019] Furthermore, the crack-induced potential energy Φ(t) at time t includes:

[0020]

[0021] Where E′ is the elastic modulus of the box girder material, ε(t′) is the local strain at the crack tip at time t′, and V c Let G be the crack volume. c For the fracture toughness of the material, K I (t′) represents the stress intensity at the crack tip at time t′, K IC denoted as the critical value of fracture toughness of the material, and a(t′) is the crack depth at time t′.

[0022] Furthermore, the stress concentration factor S(t) at time t includes:

[0023]

[0024] Where n is the number of regions, σ i Let ψ be the peak stress in the i-th region of the box girder. i Let ζ be the damage induction coefficient for the i-th region of the box girder. i α1 is the adjustment factor for the i-th region of the box girder, σ is the first weight of the damage induction coefficient, and σ is the adjustment factor for the i-th region of the box girder. yield α is the yield strength of the box girder material, α2 is the second weight of the damage induction coefficient, and κ is the yield strength of the box girder material. i Let κ be the stress concentration factor in the i-th region of the box girder. max α is the maximum stress concentration factor, α3 is the third weight of the damage induction factor, and Ai Let A be the damage area of ​​the i-th region of the box girder. ref As the reference area, α4 is the fourth weight of the damage induction coefficient, θ i Let θ be the temperature of the i-th region of the box girder. env The ambient temperature.

[0025] Furthermore, the environmental corrosion factor E(t) at time t includes:

[0026]

[0027] Wherein, ρ1 is the first adjustment factor of the environmental corrosion factor, H(t) is the environmental humidity at time t, ρ2 is the second adjustment factor of the environmental corrosion factor, and Cl - (t) represents the chloride ion concentration at time t, ρ3 represents the third adjustment factor of the environmental corrosion factor, pH(t) represents the environmental acidity or alkalinity at time t, and τ represents the fourth adjustment factor of the environmental corrosion factor.

[0028] Furthermore, the stress concentration factor κ in the i-th region of the box girder is calculated through finite element analysis. i and maximum stress concentration factor κ max .

[0029] Furthermore, the box girder is classified into defect risks to complete defect detection, including: when the risk index is less than or equal to the first risk classification threshold, the box girder is in normal condition; when the risk index is greater than the first risk classification threshold and less than or equal to the second risk classification threshold, the box girder is in observation condition; when the risk index is greater than the second risk classification threshold and less than or equal to the third risk classification threshold, the box girder is in maintenance condition; when the risk index is greater than the third risk classification threshold, the box girder is in emergency handling condition.

[0030] This invention also proposes an automatic defect detection system for concrete box girders based on deep learning, comprising:

[0031] The defect detection module is used to acquire visible light images, laser point clouds and acoustic emission signals of the box girder, perform data fusion operations to generate new box girder data, set up a defect detection model, and identify cracks and damage on the box girder based on the new box girder data.

[0032] The risk assessment module is used to set up a risk assessment model and calculate the risk index of the box girder based on the new box girder data, cracks and damage on the box girder;

[0033] The classification module is used to set multiple preset risk classification thresholds and, in combination with the risk index, classify the box girder for defect risk, thereby completing defect detection.

[0034] In summary, the technical solutions conceived by this invention have the following beneficial effects compared with the prior art:

[0035] The present invention, through the above technical solutions, can improve the accuracy of defect detection in box girders, increase detection efficiency, and avoid safety risks. Attached Figure Description

[0036] Figure 1 This is a flowchart of the method in Embodiment 1 of the present invention;

[0037] Figure 2 This is a system structure diagram of Embodiment 2 of the present invention. Detailed Implementation

[0038] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0039] The method provided by this invention can be implemented in a terminal environment that may include one or more of the following components: a processor, a storage medium, and a display screen. The storage medium stores at least one instruction, which is loaded and executed by the processor to implement the method described in the following embodiments.

[0040] A processor may include one or more processing cores. The processor uses various interfaces and lines to connect various parts of the terminal, and performs various functions and processes data by running or executing instructions, programs, code sets or instruction sets stored in the storage medium, and by calling data stored in the storage medium.

[0041] Storage media can include random access memory (RAM) or read-only memory (ROM). Storage media can be used to store instructions, programs, code, code sets, or instructions.

[0042] The display screen is used to show the user interface of each application.

[0043] In addition, those skilled in the art will understand that the structure of the terminal described above does not constitute a limitation on the terminal. The terminal may include more or fewer components, or combine certain components, or have different component arrangements. For example, the terminal may also include radio frequency circuits, input units, sensors, audio circuits, power supplies, and other components, which will not be described in detail here.

[0044] Example 1

[0045] like Figure 1As shown, this embodiment proposes an automatic defect detection method for concrete box girders based on deep learning. Through a multimodal deep learning framework, it integrates UAV aerial photography, laser 3D scanning, and acoustic emission sensing data to achieve fully automatic identification, location, and evaluation of surface and internal defects in concrete box girders. The detection efficiency is improved by more than 80%, crack identification accuracy reaches 0.05mm level, and void location error is <3cm. The method specifically includes:

[0046] Step 101: Acquire visible light images, laser point clouds, and acoustic emission signals of the box girder, perform data fusion operations to generate new box girder data, set up a defect detection model, and identify cracks and damage on the box girder based on the new box girder data. In this process, the UAV image and the 3D point cloud are registered at the sub-pixel level using the ICP algorithm.

[0047] Specifically, the defect detection model is the CrackNet-3D model, which employs deep convolutional neural networks (CNN) and 3D data processing techniques. It is a deep learning model for detecting cracks in concrete structures, specifically designed for crack detection and localization in 3D data (e.g., laser scan data, point cloud data). The backbone network uses the MobileNetV3+Coordinate Attention module, reducing the number of parameters by 56%; an edge-focal loss function is introduced to improve the robustness of identifying slender cracks; and a TensorRT acceleration engine is deployed, achieving an inference speed of 45 FPS on edge computing devices.

[0048] This embodiment also includes: automatically synthesizing training data for rare defects (such as oblique shear cracks) through a defect sample generative adversarial network (Defect-GAN); and constructing a federated learning framework to achieve continuous optimization of the defect detection model under cross-project data sharing.

[0049] Step 102: Set up a risk assessment model and calculate the risk index of the box girder based on the new box girder data, cracks and damage on the box girder;

[0050] Specifically, the risk assessment model includes:

[0051]

[0052] Among them, R t+1 R is the risk index at time t+1. tLet α be the risk index at time t, Δt be the time step, α be the first adjustment factor of the risk assessment model, C(t) be the crack propagation rate at time t, β be the second adjustment factor of the risk assessment model, γ be the third adjustment factor of the risk assessment model, S(t) be the stress concentration factor at time t, δ be the fourth adjustment factor of the risk assessment model, E(t) be the environmental corrosion factor at time t, κ be the fifth adjustment factor of the risk assessment model, λ be the sixth adjustment factor of the risk assessment model, μ be the seventh adjustment factor of the risk assessment model, σ be the eighth adjustment factor of the risk assessment model, and η be the ninth adjustment factor of the risk assessment model.

[0053] Specifically, the crack propagation rate C(t) at time t includes:

[0054]

[0055] Where C0 is the initial crack propagation rate, ω1 is the first weight of the crack propagation rate, Φ(t) is the crack-induced potential energy at time t, ω2 is the second weight of the crack propagation rate, and ν is the adjustment factor of the crack propagation rate.

[0056] Specifically, the crack-induced potential energy Φ(t) at time t includes:

[0057]

[0058] Where E′ is the elastic modulus of the box girder material, ε(t′) is the local strain at the crack tip at time t′, and V c Let G be the crack volume. c For the fracture toughness of the material, K I (t′) represents the stress intensity at the crack tip at time t′, K IC denoted as the critical value of fracture toughness of the material, and a(t′) is the crack depth at time t′.

[0059] Specifically, the stress concentration factor S(t) at time t includes:

[0060]

[0061] Where n is the number of regions, σ i Let ψ be the peak stress in the i-th region of the box girder. i Let ζ be the damage induction coefficient for the i-th region of the box girder. i α1 is the adjustment factor for the i-th region of the box girder, σ is the first weight of the damage induction coefficient, and σ is the adjustment factor for the i-th region of the box girder. yield α is the yield strength of the box girder material, α2 is the second weight of the damage induction coefficient, and κ is the yield strength of the box girder material. i Let κ be the stress concentration factor in the i-th region of the box girder. max α is the maximum stress concentration factor, α3 is the third weight of the damage induction factor, and Ai Let A be the damage area of ​​the i-th region of the box girder. ref As the reference area, α4 is the fourth weight of the damage induction coefficient, θ i Let θ be the temperature of the i-th region of the box girder. env The ambient temperature.

[0062] Specifically, the stress concentration factor κ of the i-th region of the box girder is calculated through finite element analysis. i and maximum stress concentration factor κ max .

[0063] The stress concentration factor (κ) is obtained through finite element analysis. i Finite element analysis typically involves the following steps. It is a numerical solution method that can calculate physical quantities such as stress, strain, and temperature distribution of a structure under external forces. This method allows for accurate modeling of the structure and the acquisition of stress concentration characteristics in local areas. The specific steps are as follows:

[0064] Step 1: Create the finite element model of the structure

[0065] 1. Geometric Modeling: The first step is to perform geometric modeling of the structure. For concrete box girders, CAD software (such as AutoCAD, SolidWorks, etc.) can be used to draw a three-dimensional geometric model of the structure, and then it can be imported into finite element software (such as ABAQUS, ANSYS, COMSOL, etc.).

[0066] 2. Mesh Generation: The geometric model is decomposed into several small elements to form a finite element mesh. The fineness of the mesh directly affects the accuracy of the analysis. The mesh in local areas should be appropriately refined to accurately capture stress concentration phenomena.

[0067] Step 2: Define material properties and boundary conditions

[0068] 1. Material Properties: Define the physical and mechanical properties of the material for the finite element model (such as elastic modulus, Poisson's ratio, yield strength, etc.). These properties will be used to calculate stress and strain.

[0069] 2. Boundary Conditions and Loads: Set boundary conditions and external loads according to actual working conditions. For example, apply a constant pressure or bending moment at a certain location on the concrete box girder to simulate the working state of the structure.

[0070] Step 3: Perform finite element solution

[0071] 1. Solve for the stress field: Calculate the stress field of the entire structure using finite element software. The analysis results will show the stress values ​​at each node, especially where stress concentrations may occur between nodes or in areas with irregular geometry (such as cracks, holes, sharp corners, etc.).

[0072] 2. Stress Distribution Map: Finite element analysis typically provides a stress distribution map, showing the stress magnitude in different regions. Stress concentration factor (κ) i The value is determined by the stress difference between the local area and the area far from the concentration area.

[0073] Step 4: Calculate the stress concentration factor (κ) i )

[0074] Calculate the stress concentration factor: stress concentration factor κ i It is the ratio of the maximum stress in a localized area to the reference stress far from the concentration area. The specific calculation method is as follows:

[0075]

[0076] Where: σ max σ is the maximum stress in the localized stress concentration region. ref This refers to the stress far from stress concentration areas (usually the average stress in a relatively uniform stress area).

[0077] Step 5: Calculate the maximum stress concentration factor (κ) max )

[0078] Determining the maximum stress concentration factor: For the overall design of a structure, different types of stress concentration areas may exist. The stress concentration factor is calculated at different locations within the structure (such as supports, crack ends, and around openings), and the largest value is selected as the maximum stress concentration factor (κ). max This value represents the most severe stress concentration area that may occur in the structure.

[0079] Specifically, the environmental corrosion factor E(t) at time t includes:

[0080]

[0081] Wherein, ρ1 is the first adjustment factor of the environmental corrosion factor, H(t) is the environmental humidity at time t, ρ2 is the second adjustment factor of the environmental corrosion factor, and Cl - (t) represents the chloride ion concentration at time t, ρ3 represents the third adjustment factor of the environmental corrosion factor, pH(t) represents the environmental acidity or alkalinity at time t, and τ represents the fourth adjustment factor of the environmental corrosion factor.

[0082] This embodiment also sets up another risk assessment model, as shown below:

[0083] Establish a defect hazard assessment model based on GBDT (GBDT is a commonly used ensemble learning method that combines multiple weak learners (usually decision trees) to build a strong learner, improving prediction accuracy by progressively optimizing the model's residuals. GBDT iteratively improves the model's predictive performance, with each iteration using a new decision tree to correct the errors of the previous tree). Input parameters include:

[0084] Crack propagation rate (predicted from time-series data);

[0085] Stress concentration factor (based on FEM finite element simulation);

[0086] Environmental corrosive agents (temperature, humidity, chloride ion concentration);

[0087] Output four levels of early warning (normal / observation / maintenance / emergency response).

[0088] Step 103: Set multiple preset risk classification thresholds, and combine them with the risk index to classify the defect risk of the box girder, thereby completing the defect detection.

[0089] Specifically, the box girder is classified into defect risks to complete defect detection, including: when the risk index is less than or equal to the first risk classification threshold, the box girder is in normal condition; when the risk index is greater than the first risk classification threshold and less than or equal to the second risk classification threshold, the box girder is in observation condition; when the risk index is greater than the second risk classification threshold and less than or equal to the third risk classification threshold, the box girder is in maintenance condition; when the risk index is greater than the third risk classification threshold, the box girder is in emergency handling condition.

[0090] Preferably, an interactive BIM health record is generated, marking the location of defects, risk level, and repair priority; the inspection report is automatically pushed to the maintenance management system, triggering the work order dispatch process.

[0091] Example 2

[0092] like Figure 2 As shown, this embodiment of the invention also provides an automatic defect detection system for concrete box girders based on deep learning, comprising:

[0093] The defect detection module is used to acquire visible light images, laser point clouds and acoustic emission signals of the box girder, perform data fusion operations to generate new box girder data, set up a defect detection model, and identify cracks and damage on the box girder based on the new box girder data.

[0094] Specifically, the defect detection model is the CrackNet-3D model.

[0095] The risk assessment module is used to set up a risk assessment model and calculate the risk index of the box girder based on the new box girder data, cracks and damage on the box girder;

[0096] Specifically, the risk assessment model includes:

[0097]

[0098] Among them, R t+1 R is the risk index at time t+1. t Let α be the risk index at time t, Δt be the time step, α be the first adjustment factor of the risk assessment model, C(t) be the crack propagation rate at time t, β be the second adjustment factor of the risk assessment model, γ be the third adjustment factor of the risk assessment model, S(t) be the stress concentration factor at time t, δ be the fourth adjustment factor of the risk assessment model, E(t) be the environmental corrosion factor at time t, κ be the fifth adjustment factor of the risk assessment model, λ be the sixth adjustment factor of the risk assessment model, μ be the seventh adjustment factor of the risk assessment model, σ be the eighth adjustment factor of the risk assessment model, and η be the ninth adjustment factor of the risk assessment model.

[0099] Specifically, the crack propagation rate C(t) at time t includes:

[0100]

[0101] Where C0 is the initial crack propagation rate, ω1 is the first weight of the crack propagation rate, Φ(t) is the crack-induced potential energy at time t, ω2 is the second weight of the crack propagation rate, and v is the adjustment factor of the crack propagation rate.

[0102] Specifically, the crack-induced potential energy Φ(t) at time t includes:

[0103]

[0104] Where E′ is the elastic modulus of the box girder material, ε(t′) is the local strain at the crack tip at time t′, and V c Let G be the crack volume. c For the fracture toughness of the material, K I (t′) represents the stress intensity at the crack tip at time t′, K IC denoted as the critical value of fracture toughness of the material, and a(t′) is the crack depth at time t′.

[0105] Specifically, the stress concentration factor S(t) at time t includes:

[0106]

[0107] Where n is the number of regions, σ iLet ψ be the peak stress in the i-th region of the box girder. i Let ζ be the damage induction coefficient for the i-th region of the box girder. i α1 is the adjustment factor for the i-th region of the box girder, σ is the first weight of the damage induction coefficient, and σ is the adjustment factor for the i-th region of the box girder. yield α is the yield strength of the box girder material, α2 is the second weight of the damage induction coefficient, and κ is the yield strength of the box girder material. i Let κ be the stress concentration factor in the i-th region of the box girder. max α is the maximum stress concentration factor, α3 is the third weight of the damage induction factor, and A i Let A be the damage area of ​​the i-th region of the box girder. ref As the reference area, α4 is the fourth weight of the damage induction coefficient, θ i Let θ be the temperature of the i-th region of the box girder. env The ambient temperature.

[0108] Specifically, the environmental corrosion factor E(t) at time t includes:

[0109]

[0110] Wherein, ρ1 is the first adjustment factor of the environmental corrosion factor, H(t) is the environmental humidity at time t, ρ2 is the second adjustment factor of the environmental corrosion factor, and Cl - (t) represents the chloride ion concentration at time t, ρ3 represents the third adjustment factor of the environmental corrosion factor, pH(t) represents the environmental acidity or alkalinity at time t, and τ represents the fourth adjustment factor of the environmental corrosion factor.

[0111] Specifically, the stress concentration factor κ of the i-th region of the box girder is calculated through finite element analysis. i and maximum stress concentration factor κ max .

[0112] The classification module is used to set multiple preset risk classification thresholds and, in combination with the risk index, classify the box girder for defect risk, thereby completing defect detection.

[0113] Specifically, the box girder is classified into defect risks to complete defect detection, including: when the risk index is less than or equal to the first risk classification threshold, the box girder is in normal condition; when the risk index is greater than the first risk classification threshold and less than or equal to the second risk classification threshold, the box girder is in observation condition; when the risk index is greater than the second risk classification threshold and less than or equal to the third risk classification threshold, the box girder is in maintenance condition; when the risk index is greater than the third risk classification threshold, the box girder is in emergency handling condition.

[0114] Example 3

[0115] This invention also proposes a storage medium storing multiple instructions for implementing the aforementioned deep learning-based automatic defect detection method for concrete box girders.

[0116] Optionally, in this embodiment, the storage medium may be located in any computer terminal in a group of computer terminals in a computer network, or in any mobile terminal in a group of mobile terminals.

[0117] Optionally, in this embodiment, the storage medium is configured to store program code for performing the following steps: Step 101, acquire visible light image, laser point cloud and acoustic emission signal of box girder, perform data fusion operation to generate new box girder data, set defect detection model, and identify cracks and damage on box girder according to the new box girder data;

[0118] Specifically, the defect detection model is the CrackNet-3D model.

[0119] Step 102: Set up a risk assessment model and calculate the risk index of the box girder based on the new box girder data, cracks and damage on the box girder;

[0120] Specifically, the risk assessment model includes:

[0121]

[0122] Among them, R t+1 R is the risk index at time t+1. t Let α be the risk index at time t, Δt be the time step, α be the first adjustment factor of the risk assessment model, C(t) be the crack propagation rate at time t, β be the second adjustment factor of the risk assessment model, γ be the third adjustment factor of the risk assessment model, S(t) be the stress concentration factor at time t, δ be the fourth adjustment factor of the risk assessment model, E(t) be the environmental corrosion factor at time t, κ be the fifth adjustment factor of the risk assessment model, λ be the sixth adjustment factor of the risk assessment model, μ be the seventh adjustment factor of the risk assessment model, σ be the eighth adjustment factor of the risk assessment model, and η be the ninth adjustment factor of the risk assessment model.

[0123] Specifically, the crack propagation rate C(t) at time t includes:

[0124]

[0125] Where C0 is the initial crack propagation rate, ω1 is the first weight of the crack propagation rate, Φ(t) is the crack-induced potential energy at time t, ω2 is the second weight of the crack propagation rate, and v is the adjustment factor of the crack propagation rate.

[0126] Specifically, the crack-induced potential energy Φ(t) at time t includes:

[0127]

[0128] Where E′ is the elastic modulus of the box girder material, ε(t′) is the local strain at the crack tip at time t′, and V c Let G be the crack volume. c For the fracture toughness of the material, K I (t′) represents the stress intensity at the crack tip at time t′, K IC denoted as the critical value of fracture toughness of the material, and a(t′) is the crack depth at time t′.

[0129] Specifically, the stress concentration factor S(t) at time t includes:

[0130]

[0131] Where n is the number of regions, σ i Let ψ be the peak stress in the i-th region of the box girder. i Let ζ be the damage induction coefficient for the i-th region of the box girder. i α1 is the adjustment factor for the i-th region of the box girder, σ is the first weight of the damage induction coefficient, and σ is the adjustment factor for the i-th region of the box girder. yield α is the yield strength of the box girder material, α2 is the second weight of the damage induction coefficient, and κ is the yield strength of the box girder material. i Let k be the stress concentration factor of the i-th region of the box girder. max α is the maximum stress concentration factor, α3 is the third weight of the damage induction factor, and A i Let A be the damage area of ​​the i-th region of the box girder. ref As the reference area, α4 is the fourth weight of the damage induction coefficient, θ i Let θ be the temperature of the i-th region of the box girder. env The ambient temperature.

[0132] Specifically, the environmental corrosion factor E(t) at time t includes:

[0133]

[0134] Wherein, ρ1 is the first adjustment factor of the environmental corrosion factor, H(t) is the environmental humidity at time t, ρ2 is the second adjustment factor of the environmental corrosion factor, and Cl - (t) represents the chloride ion concentration at time t, ρ3 represents the third adjustment factor of the environmental corrosion factor, pH(t) represents the environmental acidity or alkalinity at time t, and τ represents the fourth adjustment factor of the environmental corrosion factor.

[0135] Specifically, the stress concentration factor κ of the i-th region of the box girder is calculated through finite element analysis. iand maximum stress concentration factor κ max .

[0136] Step 103: Set multiple preset risk classification thresholds, and combine them with the risk index to classify the defect risk of the box girder, thereby completing the defect detection.

[0137] Specifically, the box girder is classified into defect risks to complete defect detection, including: when the risk index is less than or equal to the first risk classification threshold, the box girder is in normal condition; when the risk index is greater than the first risk classification threshold and less than or equal to the second risk classification threshold, the box girder is in observation condition; when the risk index is greater than the second risk classification threshold and less than or equal to the third risk classification threshold, the box girder is in maintenance condition; when the risk index is greater than the third risk classification threshold, the box girder is in emergency handling condition.

[0138] Example 4

[0139] This invention also proposes an electronic device, including a processor and a storage medium connected to the processor. The storage medium stores multiple instructions, which can be loaded and executed by the processor to enable the processor to execute the aforementioned deep learning-based automatic defect detection method for concrete box girders.

[0140] Specifically, the electronic device in this embodiment can be a computer terminal, which may include one or more processors and a storage medium.

[0141] The storage medium can be used to store software programs and modules, such as the deep learning-based automatic defect detection method for concrete box girders in this embodiment of the invention. The processor executes the software programs and modules stored in the storage medium to perform various functional applications and data processing, thus realizing the aforementioned deep learning-based automatic defect detection method for concrete box girders. The storage medium may include high-speed random access storage media, and may also include non-volatile storage media, such as one or more magnetic storage systems, flash memory, or other non-volatile solid-state storage media. In some instances, the storage medium may further include storage media remotely configured relative to the processor, which can be connected to the terminal via a network. Examples of such networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof.

[0142] The processor can call the information and application stored in the storage medium through the transmission system to perform the following steps: Step 101, acquire the visible light image, laser point cloud and acoustic emission signal of the box girder, perform data fusion operation to generate new box girder data, set up a defect detection model, and identify cracks and damage on the box girder according to the new box girder data;

[0143] Specifically, the defect detection model is the CrackNet-3D model.

[0144] Step 102: Set up a risk assessment model and calculate the risk index of the box girder based on the new box girder data, cracks and damage on the box girder;

[0145] Specifically, the risk assessment model includes:

[0146]

[0147] Among them, R t+1 R is the risk index at time t+1. t Let α be the risk index at time t, Δt be the time step, α be the first adjustment factor of the risk assessment model, C(t) be the crack propagation rate at time t, β be the second adjustment factor of the risk assessment model, γ be the third adjustment factor of the risk assessment model, S(t) be the stress concentration factor at time t, δ be the fourth adjustment factor of the risk assessment model, E(t) be the environmental corrosion factor at time t, κ be the fifth adjustment factor of the risk assessment model, λ be the sixth adjustment factor of the risk assessment model, μ be the seventh adjustment factor of the risk assessment model, σ be the eighth adjustment factor of the risk assessment model, and η be the ninth adjustment factor of the risk assessment model.

[0148] Specifically, the crack propagation rate C(t) at time t includes:

[0149]

[0150] Where C0 is the initial crack propagation rate, ω1 is the first weight of the crack propagation rate, Φ(t) is the crack-induced potential energy at time t, ω2 is the second weight of the crack propagation rate, and ν is the adjustment factor of the crack propagation rate.

[0151] Specifically, the crack-induced potential energy Φ(t) at time t includes:

[0152]

[0153] Where E′ is the elastic modulus of the box girder material, ε(t′) is the local strain at the crack tip at time t′, and V c Let G be the crack volume. c For the fracture toughness of the material, K I (t′) represents the stress intensity at the crack tip at time t′, K IC denoted as the critical value of fracture toughness of the material, and a(t′) is the crack depth at time t′.

[0154] Specifically, the stress concentration factor S(t) at time t includes:

[0155]

[0156] Where n is the number of regions, σ i Let ψ be the peak stress in the i-th region of the box girder. i Let ζ be the damage induction coefficient for the i-th region of the box girder. i α1 is the adjustment factor for the i-th region of the box girder, σ is the first weight of the damage induction coefficient, and σ is the adjustment factor for the i-th region of the box girder. yield α is the yield strength of the box girder material, α2 is the second weight of the damage induction coefficient, and κ is the yield strength of the box girder material. i Let κ be the stress concentration factor in the i-th region of the box girder. max α is the maximum stress concentration factor, α3 is the third weight of the damage induction factor, and A i Let A be the damage area of ​​the i-th region of the box girder. ref As the reference area, α4 is the fourth weight of the damage induction coefficient, θ i Let θ be the temperature of the i-th region of the box girder. env The ambient temperature.

[0157] Specifically, the environmental corrosion factor E(t) at time t includes:

[0158]

[0159] Wherein, ρ1 is the first adjustment factor of the environmental corrosion factor, H(t) is the environmental humidity at time t, ρ2 is the second adjustment factor of the environmental corrosion factor, and Cl - (t) represents the chloride ion concentration at time t, ρ3 represents the third adjustment factor of the environmental corrosion factor, pH(t) represents the environmental acidity or alkalinity at time t, and τ represents the fourth adjustment factor of the environmental corrosion factor.

[0160] Specifically, the stress concentration factor κ of the i-th region of the box girder is calculated through finite element analysis. i and maximum stress concentration factor κ max .

[0161] Step 103: Set multiple preset risk classification thresholds, and combine them with the risk index to classify the defect risk of the box girder, thereby completing the defect detection.

[0162] Specifically, the box girder is classified into defect risks to complete defect detection, including: when the risk index is less than or equal to the first risk classification threshold, the box girder is in normal condition; when the risk index is greater than the first risk classification threshold and less than or equal to the second risk classification threshold, the box girder is in observation condition; when the risk index is greater than the second risk classification threshold and less than or equal to the third risk classification threshold, the box girder is in maintenance condition; when the risk index is greater than the third risk classification threshold, the box girder is in emergency handling condition.

[0163] The sequence numbers of the above embodiments of the present invention are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments.

[0164] In the above embodiments of the present invention, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments.

[0165] In the several embodiments provided by this invention, it should be understood that the disclosed technical content can be implemented in other ways. The system embodiments described above are merely illustrative; for example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, indirect coupling or communication connection between units or modules, and may be electrical or other forms.

[0166] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0167] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0168] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: USB flash drives, read-only memory (ROM), random access memory (RAM), portable hard drives, magnetic disks, optical disks, and other media capable of storing program code.

[0169] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A deep learning-based automatic detection method for concrete box girder defects, characterized in that, include: The visible light image, laser point cloud, and acoustic emission signal of the box girder are acquired, and data fusion is performed to generate new box girder data. A defect detection model is set up, and cracks and damage on the box girder are identified based on the new box girder data. Set up a risk assessment model and calculate the risk index of the box girder based on the new box girder data, cracks and damage on the box girder; Risk assessment models include: in, For time Time risk index For time Time risk index For time step, As the first adjustment factor in the risk assessment model, For time Time crack propagation rate, As the second adjustment factor in the risk assessment model, As the third adjustment factor in the risk assessment model, For time Stress concentration factor As the fourth adjustment factor in the risk assessment model, For time Environmental corrosion factors It is the fifth adjustment factor in the risk assessment model. It is the sixth adjustment factor in the risk assessment model. It is the seventh adjustment factor in the risk assessment model. It is the eighth adjustment factor in the risk assessment model. It is the ninth adjustment factor in the risk assessment model; Multiple preset risk classification thresholds are set, and the box girder is classified into defect risks based on the risk index, thereby completing defect detection.

2. The method for automatic defect detection of concrete box girders based on deep learning as described in claim 1, characterized in that, The defect detection model is the CrackNet-3D model.

3. The method for automatic defect detection of concrete box girders based on deep learning as described in claim 2, characterized in that, time Time crack propagation rate include: in, The initial crack propagation rate, As the first weight for crack propagation rate, For time Time-crack induced potential energy, As the second weight for crack propagation rate, This is an adjustment factor for the crack propagation rate.

4. The method for automatic defect detection of concrete box girders based on deep learning as described in claim 3, characterized in that, time Time-crack induced potential energy include: in, The elastic modulus of the box girder material. For time Local strain at the crack tip For the crack volume, For the fracture toughness of the material, For time The stress intensity at the crack tip, This is the critical value for the fracture toughness of the material. For time Crack depth at time.

5. The method for automatic defect detection of concrete box girders based on deep learning as described in claim 4, characterized in that, time Stress concentration factor include: in, For the number of regions, For the box girder Peak stress in each region For the box girder Damage induction coefficient of each region For the box girder Adjustment factors for each region As the first weight of the damage induction coefficient, The yield strength of the box girder material. As the second weight of the damage induction coefficient, For the box girder Stress concentration factor of each region The maximum stress concentration factor. As the third weight of the damage induction coefficient, For the box girder Damage area in each region For reference area, As the fourth weight of the damage induction coefficient, For the box girder Temperature of each region The ambient temperature.

6. The method for automatic defect detection of concrete box girders based on deep learning as described in claim 5, characterized in that, time Environmental corrosion factors include: in, It is the first modulator of environmental corrosion factors. For time ambient humidity, It is the second modifier of environmental corrosion factors. For time Chloride ion concentration at time It is the third adjusting factor among environmental corrosion factors. For time Environmental pH It is the fourth adjusting factor for environmental corrosion factors.

7. The method for automatic defect detection of concrete box girders based on deep learning as described in claim 5, characterized in that, The first section of the box girder was calculated using finite element analysis. Stress concentration factor of each region and maximum stress concentration factor .

8. The method for automatic defect detection of concrete box girders based on deep learning as described in claim 1, characterized in that, The box girder is classified into defect risks to complete defect detection, including: when the risk index is less than or equal to the first risk classification threshold, the box girder is in normal condition; when the risk index is greater than the first risk classification threshold and less than or equal to the second risk classification threshold, the box girder is in observation condition; when the risk index is greater than the second risk classification threshold and less than or equal to the third risk classification threshold, the box girder is in maintenance condition; when the risk index is greater than the third risk classification threshold, the box girder is in emergency handling condition.

9. An automatic defect detection system for concrete box girders based on deep learning, characterized in that, include: The defect detection module is used to acquire visible light images, laser point clouds and acoustic emission signals of the box girder, perform data fusion operations to generate new box girder data, set up a defect detection model, and identify cracks and damage on the box girder based on the new box girder data. The risk assessment module is used to set up a risk assessment model and calculate the risk index of the box girder based on the new box girder data, cracks and damage on the box girder; Risk assessment models include: in, For time Time risk index For time Time risk index For time step, As the first adjustment factor in the risk assessment model, For time Time crack propagation rate, As the second adjustment factor in the risk assessment model, As the third adjustment factor in the risk assessment model, For time Stress concentration factor As the fourth adjustment factor in the risk assessment model, For time Environmental corrosion factors It is the fifth adjustment factor in the risk assessment model. It is the sixth adjustment factor in the risk assessment model. It is the seventh adjustment factor in the risk assessment model. It is the eighth adjustment factor in the risk assessment model. It is the ninth adjustment factor in the risk assessment model; The classification module is used to set multiple preset risk classification thresholds and, in combination with the risk index, classify the box girder for defect risk, thereby completing defect detection.

Images