A method and terminal for detecting the degree of damage to reinforced concrete components

Abstract

This invention relates to the field of image processing technology and discloses a method and terminal for detecting the degree of damage to reinforced concrete components. The method includes: acquiring an apparent damage image of a damaged reinforced concrete component; extracting the damage texture from the apparent damage image; calculating the damage fractal dimension of the damaged reinforced concrete component based on the damage texture; and obtaining a damage index of the damaged reinforced concrete component based on the damage fractal dimension. This invention identifies the damage texture of an apparent damage image and extracts the damage fractal dimension of the damage texture. By analyzing the intrinsic characteristics of the damage using the damage fractal dimension, it achieves the judgment of damage indexes, eliminating interference caused by subjective human factors and exhibiting high efficiency.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method, terminal, and readable storage medium for detecting the degree of damage to reinforced concrete components. Background Technology

[0002] Earthquakes are among the most severe natural disasters. After an earthquake, rapid detection and post-earthquake assessment of building structures are crucial. There is a certain correlation between seismic action and apparent structural damage. As seismic action intensifies, cracks, concrete spalling, and exposed rebar will appear on the surface of components. These visual characteristics of seismic damage can be used to determine the residual mechanical properties of building structures after the earthquake.

[0003] However, traditional earthquake damage assessment methods require experts to observe and measure damaged components and judge the degree of earthquake damage and remaining seismic resistance based on experience. This assessment method is not only time-consuming but also has certain risks. In addition, different experts may arrive at inconsistent assessment results of the degree of earthquake damage, meaning that this judgment method is subject to errors caused by human subjectivity. Summary of the Invention

[0004] The purpose of this invention is to provide a method and terminal for detecting the degree of damage to reinforced concrete components, which aims to solve the problems of time-consuming assessment and inconsistency caused by subjective errors in the existing technology for detecting damage to reinforced concrete.

[0005] The technical solution adopted by this invention to solve the technical problem is as follows:

[0006] This invention provides a method for detecting the degree of damage to reinforced concrete components, the method comprising:

[0007] Obtain the apparent damage image of the damaged reinforced concrete member, and extract the damage texture from the apparent damage image;

[0008] The damage fractal dimension of the damaged reinforced concrete member is calculated based on the damage texture.

[0009] The damage index of the damaged reinforced concrete member is obtained based on the damage fractal dimension.

[0010] Further, the extraction of damage texture from the apparent damage image specifically includes:

[0011] Obtain an image model and input the apparent damage image into the image model;

[0012] Obtain the damage texture output by the image model.

[0013] Furthermore, the step of calculating the damage fractal dimension of the damaged reinforced concrete member based on the damage texture specifically includes:

[0014] Calculate the single fractal dimension or generalized fractal dimension of the damaged texture;

[0015] The single fractal dimension or the generalized fractal dimension is used as the damage fractal dimension.

[0016] Further, the damage fractal dimension is specifically the single fractal dimension of the damage texture, and calculating the damage fractal dimension of the damage texture specifically includes:

[0017] Box counting is performed on the damaged texture multiple times to obtain the number of boxes with damaged textures under different box sizes;

[0018] Based on the number of boxes with damaged textures at different box sizes, fit the single fractal dimension D0:

[0019]

[0020] Where r is the ratio of box size to image size, and N(r) is the number of boxes with damaged textures when the ratio of box size to image size is r.

[0021] Furthermore, the multiple box counts performed on the damaged texture to obtain the number of boxes with damaged textures at different box sizes, each box count of the damaged texture includes:

[0022] The apparent damage image is uniformly divided into 1 / r boxes;

[0023] Pixel-by-pixel identification is performed on each box to obtain all boxes with damaged textures;

[0024] Count the number of boxes with damaged textures to obtain the number of boxes with damaged textures N(r) when the ratio of box size to image size is r.

[0025] Further, the damage fractal dimension is specifically the generalized fractal dimension of the damage texture, and calculating the damage fractal dimension of the damage texture specifically includes:

[0026] Box probability counting is performed on the damaged texture multiple times to obtain the weighted probability of each box having a damaged texture under different box sizes;

[0027] The generalized fractal dimension is fitted based on the weighted probability that each box of different box sizes has damaged textures:

[0028]

[0029] Where q represents the weight, and Dq Let P represent the generalized fractal dimension with weight q. i q M(r) represents the weight probability of the i-th box when the ratio of box size to image size is r, and M(r) is the total number of boxes when the ratio of box size to image size is r.

[0030] Furthermore, the generalized fractal dimension includes generalized fractal dimensions with weights q of -40, -2, -1, 0, 1, 2, and 40.

[0031] Furthermore, obtaining the damage index of the damaged reinforced concrete member based on the damage fractal dimension specifically includes:

[0032] Obtain the aspect ratio of the concrete component image, and input the damage fractal dimension and the aspect ratio into the classifier;

[0033] The classifier outputs the damage index.

[0034] Furthermore, to achieve the above objectives, the present invention also provides a terminal, the terminal comprising: a memory, a processor, and a reinforced concrete component damage degree detection program stored in the memory and executable on the processor, wherein when the reinforced concrete component damage degree detection program is executed by the processor, the terminal controls the terminal to implement the steps of the reinforced concrete component damage degree detection method as described above.

[0035] In addition, to achieve the above objectives, the present invention also provides a readable storage medium storing a reinforced concrete component damage degree detection program, which, when executed by a processor, implements the steps of a reinforced concrete component damage degree detection method as described above.

[0036] The present invention, by employing the above technical solution, has the following effects:

[0037] This invention identifies damage textures in images of reinforced concrete components and extracts the damage fractal dimension of the damage texture. By analyzing the damage fractal dimension, the intrinsic characteristics of the damage can be determined, thereby enabling the judgment of damage indicators. This method eliminates interference caused by subjective human factors and is highly efficient. Attached Figure Description

[0038] Figure 1 This is a flowchart of the steps of a method for detecting the degree of damage to reinforced concrete components in a preferred embodiment of the present invention;

[0039] Figure 2 This is a flowchart of the steps for constructing each model in a preferred embodiment of the present invention;

[0040] Figure 3 This is a schematic diagram illustrating the effect of the image model extracting damaged textures in a preferred embodiment of the present invention;

[0041] Figure 4 This is a schematic diagram of the parameters of different reinforced concrete frame columns in a preferred embodiment of the present invention;

[0042] Figure 5 This is a schematic diagram showing the dimensions of different reinforced concrete frame columns in a preferred embodiment of the present invention;

[0043] Figure 6 This is a schematic diagram of the static hysteresis test device in a preferred embodiment of the present invention;

[0044] Figure 7 This is a schematic diagram of fractal analysis with different weight values ​​for the multifractal dimension;

[0045] Figure 8 This is a schematic diagram of damage and data comparison in a static hysteresis test according to a preferred embodiment of the present invention;

[0046] Figure 9 This is a schematic diagram of the training steps of the adaptive enhancement algorithm in a preferred embodiment of the present invention;

[0047] Figure 10 This is a schematic diagram illustrating the predictive effect of damage indicators in a preferred embodiment of the present invention;

[0048] Figure 11 This is a schematic diagram of the interface of the integrated platform for damage identification and damage assessment in a preferred embodiment of the present invention.

[0049] Figure 12 A schematic diagram of the operating environment of a preferred embodiment of the terminal of the present invention. Detailed Implementation

[0050] To make the objectives, technical solutions, and advantages of this invention clearer and more explicit, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0051] Example 1

[0052] Specifically, please see Figure 1 and Figure 2 Embodiment 1 of this application is a method for detecting the degree of damage to reinforced concrete components, which includes the following steps:

[0053] S1. Obtain the apparent damage image of the damaged reinforced concrete component and extract the damage texture from the apparent damage image.

[0054] Specifically, in this embodiment, drones or robots can be used to collect images of damaged reinforced concrete components in the earthquake zone to avoid the dangers of manual collection. After the images are collected, they are preprocessed, including rotating, scaling and deforming, to obtain a frontal image with uniform size and format.

[0055] After acquiring images of reinforced concrete components, these images are input into an image model, which outputs damage textures. In this embodiment, the Yolov8 model is specifically used as the image model. A large number of manually labeled apparent damage images are first used as training data to form a training dataset, allowing the image model to learn its ability to judge damage textures. Then, the image model can be used to extract damage textures from the concrete images. The extracted images are shown below. Figure 3 As shown, Figure 3 In the image, the area highlighted by the "crack" box represents the damage texture, and the number following it indicates the confidence level of the damage.

[0056] Specifically, in this embodiment, images of reinforced concrete components are obtained as training data by collecting data from static hysteresis tests of RC (reinforced concrete structure). Please refer to [reference needed] in this embodiment. Figure 4 and Figure 5 , Figure 4 This is a schematic diagram of the parameters of different reinforced concrete frame columns according to the present invention. Figure 5 This is a schematic diagram of the dimensions of different reinforced concrete structural frame columns of the present invention. Static hysteresis tests were conducted on eight reinforced concrete structural frame columns with different dimensions, reinforcement ratios, and shear span ratios to obtain the data required for the research.

[0057] Among them, the static hysteresis test refers to obtaining the load-deformation data of the specimen under the cyclic action of the applied force. The hysteresis curve plotted by the data points can reflect the seismic performance, energy dissipation capacity and other information of the component during repeated stress. Figure 6 The apparatus used for conducting quasi-static loading tests involves anchoring the bottom of a reinforced concrete column, applying axial force at the top of the column, and applying low-cycle repeated horizontal loads through a four-bar linkage until the first cycle horizontal force of the structure drops below 80% of the peak horizontal force. During the test, a mortar surface layer is added to the side surface of the column and coated with latex paint to simulate the earthquake damage phenomenon of actual building components.

[0058] During the experiment, the surface damage phenomena of the components at each stage of loading were recorded using a high-speed, high-definition camera, so that the apparent damage distribution of the reinforced concrete frame columns could be used in subsequent experiments.

[0059] S2. Calculate the damage fractal dimension of the damaged reinforced concrete component based on the damage texture.

[0060] In this embodiment, program code is written on the Matlab (matrix laboratory) platform to determine the damage fractal dimension of the damage texture.

[0061] The fractal dimension of damage is a fractal dimension used to describe the state of damage. Fractal dimension is a concept in fractal geometry. In nature, there are many highly irregular shapes similar to coastlines. Studies have shown that when these shapes are examined at different scales, they exhibit similar complexity and irregularity within a small range. That is, the local structure of these geometric shapes is similar to their overall structure, showing similar morphological characteristics at different magnification scales. This property is called self-similarity. Fractal geometry is a branch of mathematics that studies the self-similarity of complex shapes that repeat at different scales and the complex boundary structures. The core characteristic of fractal geometry is that these shapes exhibit the same or similar structural features at different magnifications. Therefore, a concept is introduced to characterize the complexity of geometric shapes: the fractal dimension. In fractal theory, the fractal dimension, as an important tool for describing and measuring these complex forms, can be either an integer or a fraction.

[0062] It is easy to see from the above description of fractal dimension that fractal dimension is scale independent, and therefore can reveal the intrinsic characteristics of the post-earthquake damage distribution on the surface of reinforced concrete components. Therefore, in this embodiment, fractal parameters are used to describe the damaged surface, that is, to obtain the damage fractal dimension. It can not only truly reflect the spatial distribution and arrangement of the damaged area, but also reflect the degree of damage to the column.

[0063] The damage fractal dimension can be a simple fractal dimension or a multifractal dimension; in this embodiment, it is specifically a multifractal dimension.

[0064] The mainstream method for calculating the single fractal dimension uses the box-counting principle. In this method, the desired image is uniformly divided into 1 / r boxes (cubes), where r is a variable value. Then, pixel-by-pixel identification is performed on each box, and the number of boxes filled with damaged textures is counted. This yields the number of damaged texture-filled boxes N(r) for a given number of boxes r. By adjusting the value of r multiple times, the number of damaged texture-filled boxes N(r) under different values ​​is obtained. Finally, based on the multiple numbers of damaged texture-filled boxes N(r), the single fractal dimension D0 can be fitted and calculated as follows:

[0065]

[0066] Generalized fractal dimension (GFD), also known as multifractal dimension, is used to extract more information from images. Multifractal dimension is represented by D.q express:

[0067]

[0068] Among them, D q Let P represent the generalized fractal dimension with weight q, where M(r) is the total number of boxes (1 / r), and q is a weighting factor, also called a representation value. When q > 1, information from high concentration or high aggregation (dense regions, high probability) is amplified; when q < 1, information from low concentration or low aggregation (sparse regions, low probability) is amplified. i q (r) represents the probability P that the weight factor is q. i (r), probability P i (r) represents the ratio of the number of pixels filled with damaged texture in the i-th box to the total number of pixels filled with damaged texture in the image:

[0069]

[0070] Where, N i (r) represents the number of pixels in the i-th box of the image that are filled with damaged texture, N j (r) represents the number of pixels in the j-th box of the image that are filled with damaged texture, P i (r) is a relative probability quality coefficient, representing the relative damage intensity of the i-th box in the image.

[0071] For multifractal analysis of binarized black and white images, in order to further explain q and P i q (r) value, Figure 7 The fractal analysis results of the damaged area images under different q values ​​are presented. Figure 7 (a) in the diagram is a schematic diagram of the damage distribution. Figure 7 In the diagram, (b) represents the calculated probability P of different boxes. i q (r) Schematic diagram, Figure 7 In the diagram, (c) represents the normalized P of different boxes under different q values. i q (r),

[0072] It can be observed that for boxes with a large proportion of black pixels due to cracking and peeling, P i q The value of (r) is higher. Figure 7 (c) shows the distribution of the normalized probability of the damage region considering the weighting factor q, where the normalized probability λ i The calculation of (q,r) is as follows:

[0073]

[0074] When q = 0, boxes containing cracked and peeling pixels are black, and boxes with no damaged pixels are white. At this point, only the presence or absence of cracks in each box is considered; the intensity of the damage is irrelevant, and the distinction between cracks and breakage is not taken into account. Positive q values ​​increase the intensity of boxes containing pixels with significant peeling or cracking. For example... Figure 7 As shown in (c), when q is positive, the focus is on the most severely damaged boxes, while the remaining sparsely distributed boxes with fewer pixels are almost white. When q is negative, the normalized probability distribution results are the opposite of those for positive q. For example, when q is -1, the boxes with fewer damaged pixels are actually darker.

[0075] In this embodiment, D is selected with q values ​​of -40, -2, -1, 0, 1, 2, and 40. q The value is an important parameter for assessing the degree of damage.

[0076] S3. Determine the damage index of the damaged reinforced concrete component based on the damage fractal dimension.

[0077] In this embodiment, in addition to the seven damage fractal dimensions, the height-to-width ratio (h / b) of the component is also used as a reference when judging the damage index.

[0078] Damage indices are quantitative parameters established based on various engineering requirements parameters (such as strength, stiffness, energy dissipation, deformation, etc.). In this embodiment, the following novel strength-based damage index is used to measure the degree of strength loss of reinforced concrete structural members under seismic loading.

[0079]

[0080] Among them, DI s The damage index is represented by S, which represents the strength of the column at a specific displacement level. c This represents the peak strength of the damaged reinforced concrete column in the skeleton curve. Please refer to [reference needed]. Figure 8 The lateral load-displacement hysteresis curve of a reinforced concrete column reaches its peak strength S c Before and after correspond to the pre-capping and post-capping stages, respectively. The damage distribution images marked on the lateral load-displacement cycle curve are as follows: Figure 8 As shown.

[0081] In this embodiment, experimental data is obtained from static hysteresis tests, and the test photos are matched one-to-one with the corresponding loading points on the hysteresis curve. The damage fractal dimension of each component at each stage is calculated based on the photos. By defining and estimating the degradation state index value of the mechanical properties (i.e. stiffness and strength) of the component at a certain loading point on the hysteresis curve, the degree of post-earthquake damage of the component is output in combination with the regression model.

[0082] Thus, by matching the damage images, multifractal dimension, aspect ratio (h / b), and strength degradation index values ​​of each loading stage, the seismic damage database of the reinforced concrete structural test columns required for the study was successfully established.

[0083] Subsequently, this invention employs the AdaBoost algorithm to determine the impairment index. AdaBoost is an ensemble learning method that improves the accuracy of weak learners by altering the distribution of sample weights. For classification problems, the AdaBoost algorithm adjusts the data weights based on the previous classification results. The weights of samples misclassified in the previous weak learner are increased in the next weak learner, while the weights of correctly classified samples are decreased. A new weak learner is added to the model in each iteration. This process of adjusting weights and training weak learners is repeated until the number of misclassifications falls below a preset value or the number of iterations reaches a specified maximum, ultimately resulting in a strong learner. In short, the core idea of ​​the AdaBoost algorithm is to adjust the weights of misclassified samples and iteratively upgrade them. Its training steps are as follows: Figure 9 As shown.

[0084] Suppose we have a sample set consisting of m samples:

[0085] D = {(x1,y1),(x2,y2),...,(x m ,y m )}

[0086] Each sample data has d features. In this embodiment, the features are the seven damage fractal dimensions and the aspect ratio, y m For sample x m The target value is the corresponding damage index value. Let the initial weight of each sample be D1, and the weight of each sample be initialized to 1 / m. In the k-th iteration, the weight of each sample is D1. k It should be noted that D1 is used to train the first weak classifier, and D... k Used for the k-th weak classifier h k Training.

[0087] The process is repeated n times, and the classifier h is calculated after each iteration. k The maximum error on the training set is calculated using the following formula:

[0088] E k =max|y l -h k (x l )|;

[0089] Among them, h k That is, classifier h k E k For classifier h k The maximum error.

[0090] The relative error for each sample:

[0091]

[0092] Among them, e kl For classifier h k The relative error on the l-th sample.

[0093] Therefore, the current weak classifier h k The error rate is:

[0094]

[0095] That is, the sum of the products of the weights and errors of all samples in the dataset.

[0096] Update the current weak classifier h k The weight is calculated using the following formula:

[0097]

[0098] Among them, w k For weak classifier h k The weight.

[0099] Therefore, for sample x l The formula for calculating the updated weights is:

[0100]

[0101] Among them, Z k D is the normalization factor. k (x l ) is the k-th iteration pair of sample x l The weighted value, D k+1 (x l ) is the (k+1)th iteration on sample x l The weighted value is calculated using the following formula:

[0102]

[0103] After n rounds of iteration, the final strong regressor H(x) is as follows:

[0104]

[0105] Here, f(x) is the median of the weighted output of all weak learners.

[0106] Use the coefficient of determination, i.e., R 2 R0 is a parameter used to measure the goodness of fit of a regression model. The coefficient of determination indicates how well the regression model fits the dataset, that is, it represents the percentage of data explained by the regression model. 2 The closer the value is to 1, the better the performance of the regression model. This can be expressed by the following formula:

[0107]

[0108] in These are the predicted values ​​from the regression model. y is the average of all damage indicators. l The values ​​of the damage index in the dataset.

[0109] The AdaBoost model was used to regress and predict the damage level of each sample in the test set. The results are shown in the figure below. Figure 10 As shown in the figure, the horizontal axis represents the sample number of the test machine, and the vertical axis represents the numerical value of the damage degree. The average prediction accuracy of the sample damage degree can reach 65% or higher.

[0110] Finally, this embodiment completes an integrated platform for damage identification and damage severity assessment based on the method of the present invention, the interface of which is as follows: Figure 11 As shown.

[0111] Example 2

[0112] Please see Figure 12 Based on the above method, the present invention also provides a terminal, which includes a processor 10, a memory 20, and a display 30. However, it should be understood that it is not required to implement all the components shown, and more or fewer components may be implemented instead.

[0113] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory. In other embodiments, the memory 20 may be an external storage device of the terminal, such as a plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the terminal. Further, the memory 20 may include both internal and external storage devices. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code installed on the terminal. The memory 20 can also be used to temporarily store data that has been output or will be output. In one embodiment, the memory 20 stores a reinforced concrete component damage detection program 40, which can be executed by the processor 10 to realize the terminal of this application.

[0114] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data, such as executing the relevant program of the method for detecting the degree of damage to reinforced concrete components.

[0115] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light Emitting Diode) touchscreen. The display 30 is used to display information on the terminal and to display a visual user interface.

[0116] In one embodiment, when the processor 10 executes a reinforced concrete component damage detection program 40 stored in the memory 20, it implements the steps of a reinforced concrete component damage detection method as described above.

[0117] Example 3

[0118] This embodiment provides a storage medium that stores a program for detecting the degree of damage to reinforced concrete components. When the program is executed by a processor, it implements the steps of a method for detecting the degree of damage to reinforced concrete components as described above.

[0119] In summary, this invention identifies the damage texture of apparent damage images and extracts the damage fractal dimension of the damage texture. By analyzing the damage fractal dimension, the intrinsic characteristics of the damage can be determined, thereby eliminating interference caused by subjective human factors and exhibiting high efficiency.

[0120] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal that includes that element.

[0121] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.). The program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The storage medium can be a memory, magnetic disk, optical disk, etc.

[0122] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for detecting the degree of damage to reinforced concrete components, characterized in that, The method for detecting the degree of damage to reinforced concrete components includes: Obtain the apparent damage image of the damaged reinforced concrete member, and extract the damage texture from the apparent damage image; The damage fractal dimension of the damaged reinforced concrete member is calculated based on the damage texture. Based on the damage fractal dimension, the damage index of the damaged reinforced concrete member is obtained; The damage fractal dimension is specifically the generalized fractal dimension of the damage texture, and calculating the damage fractal dimension of the damage texture specifically includes: Box probability counting is performed on the damaged texture multiple times to obtain the weighted probability of each box having a damaged texture under different box sizes; The generalized fractal dimension is fitted based on the weighted probability that each box of different box sizes has damaged textures: in, Indicates weight, Indicates the weight is The generalized fractal dimension in the case of The ratio of box size to image size is The first case The weighted probability of each box, The ratio of box size to image size is The total number of boxes under the given conditions; The generalized fractal dimension includes weights. The generalized fractal dimensions are -40, -2, -1, 0, 1, 2, and 40. The step of obtaining the damage index of the damaged reinforced concrete member based on the damage fractal dimension specifically includes: Obtain the aspect ratio of the concrete component image, and input the generalized fractal dimension and the aspect ratio into the classifier; The classifier outputs the damage index.

2. The method for detecting the degree of damage to reinforced concrete components according to claim 1, characterized in that, The extraction of damage texture from the apparent damage image specifically includes: Obtain an image model and input the apparent damage image into the image model; Obtain the damage texture output by the image model.

3. A terminal, characterized in that, The terminal includes: a memory, a processor, and a reinforced concrete component damage detection program stored in the memory and executable on the processor. When the processor executes the reinforced concrete component damage detection program, it controls the terminal to implement the steps of the reinforced concrete component damage detection method as described in any one of claims 1-2.

4. A readable storage medium, characterized in that, The readable storage medium stores a program for detecting the degree of damage to reinforced concrete components, which, when executed by a processor, implements the steps of a method for detecting the degree of damage to reinforced concrete components as described in any one of claims 1-2.

Images