A method and system for local damage detection and evaluation applied to composite structures

Abstract

The application discloses a kind of local damage detection evaluation method and system applied to combination structure, method includes: based on physical size, combination structure surface is evenly divided into several grid units;Extract the gray abnormality feature of each grid unit to calculate surface damage probability and damage aggregation degree;Based on surface damage probability and damage aggregation degree, the scale of each grid unit is adaptively adjusted, and the region to be determined damage is obtained;Obtain the internal detection sequence of each region to be determined damage about multiple internal damage indicators at multiple time points, calculate the influence factor of each region to be determined damage on overall damage, and evaluate the key detection area of combination structure.The method extracts surface abnormality features based on combination structure image to adaptively divide grid, combined with internal acoustic emission, strain and other detection data of each grid area, accurately quantifies the correlation degree of local damage and overall damage of each area, and can accurately identify the key damage area that has the greatest impact on overall structure safety.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of bridge engineering monitoring, in particular to a local damage detection and evaluation method and system applied to a composite structure. BACKGROUND

[0002] Steel plate composite structure is a transverse load-bearing member that is connected into a whole by shear connectors and bears force together. It has been widely used in modern bridge structures because it can fully utilize the mechanical properties of both steel and concrete. However, during its long-term service, factors such as cyclic loading, environmental corrosion, and material aging can lead to the generation and development of various types of local damage.

[0003] Due to the multiphase and non-uniformity of steel plate composite structure, traditional detection methods such as visual inspection have insufficient detection capability and low efficiency for key hidden areas, and cannot quickly respond to changes in structural risk state. Different parts of the steel plate composite structure have significant differences in their importance in force. For example, weld cracks in the tension zone, fracture or loosening of shear connectors, and separation of the bonding surface between the steel beam and the concrete bridge deck, the weakening of the overall load-carrying capacity of the structure by small damage in these key hidden areas is much greater than that of the same size damage in the non-stress area of the web or flange plate. Currently, the detection of these areas often relies on point and contact precision instruments, which require the erection of scaffolding, interruption of traffic, and point-by-point inspection by experienced technicians. This has limited detection range, extremely low efficiency, high cost, and poor automation, making it difficult to achieve efficient and normalized monitoring and screening of the safety state of the entire bridge, especially in hidden areas.

[0004] Therefore, there is an urgent need for an innovative detection and evaluation method and system that can accurately focus on the key stress and hidden areas of the steel plate composite structure, thereby repairing and optimizing the local structure in a timely manner to prevent abnormalities, degradation, and damage during service and operation. SUMMARY

[0005] To solve the above problems of the prior art, the present application provides a local damage detection and evaluation method and system applied to a composite structure, which uses information from limited measurement points to effectively infer the entire bridge, quantitatively evaluate the differential impact of damage location on structural safety, and timely understand the safety state of key and hidden areas to improve the pertinence of composite structure maintenance decisions.

[0006] To achieve the above purpose, the present application provides the following technical solutions:

[0007] In a first aspect, the present application provides a local damage detection and evaluation method applied to a composite structure, comprising:

[0008] The combination structure surface is evenly divided into a plurality of grid units based on physical dimensions of the combination structure;

[0009] extracting a gray abnormality feature of each grid unit, calculating a surface damage probability of each grid unit, and obtaining a damage aggregation degree of each grid unit based on a distribution of the surface damage probability in a grid neighborhood;

[0010] adapting a size of each grid unit based on the surface damage probability and the damage aggregation degree, and obtaining at least one to-be-determined damage region of the combination structure surface;

[0011] obtaining an internal detection sequence of each to-be-determined damage region with respect to a plurality of internal damage indicators at a plurality of time points;

[0012] calculating an influence factor of each to-be-determined damage region on the overall damage based on the internal detection sequence, and evaluating a key detection region of the combination structure.

[0013] Preferably, the extracting a gray abnormality feature of each grid unit, calculating a surface damage probability of each grid unit comprises:

[0014] calculating a global abnormality degree of the grid unit based on a deviation degree of a gray value of a pixel point in the grid unit relative to all pixel points in the gray image;

[0015] calculating a gray difference between each pixel point and a neighborhood pixel point in each grid unit, and extracting a local abnormality degree of each grid unit;

[0016] determining a texture direction of the pixel point in the grid unit, and calculating a continuity degree of the local abnormality degree of the grid unit in the texture direction;

[0017] calculating the surface damage probability of each grid unit based on the global abnormality degree, the local abnormality degree, and the continuity degree.

[0018] Preferably, the determining a texture direction of the pixel point in the grid unit comprises:

[0019] calculating a gradient mean value of all pixel points in the grid unit;

[0020] taking any vertical direction of the direction corresponding to the gradient mean value as the texture direction of the pixel point in the grid unit.

[0021] Preferably, the obtaining a damage aggregation degree of each grid unit based on a distribution of the surface damage probability in a grid neighborhood comprises:

[0022] for each grid unit,

[0023] calculating a first deviation of the grid unit relative to a global surface damage probability based on the surface damage probability;

[0024] calculating a second deviation of each neighborhood grid cell of the grid cell with respect to a global surface damage probability under a spatial distance weighting based on the surface damage probability;

[0025] normalizing the product of the first deviation and the second deviation by a global data to obtain a damage aggregation degree of each grid cell.

[0026] As a preferred, the adaptive adjustment of the grid cell scale based on the surface damage probability and the damage aggregation degree comprises:

[0027] merging adjacent grid cells based on the damage aggregation degree of the grid cells, and taking the maximum value of the surface damage probabilities of the grid cells as the surface damage probability of the new grid cell after merging, and taking the area-weighted average of the damage aggregation degrees of the grid cells as the damage aggregation degree of the new grid cell;

[0028] For each new grid cell, calculating an adjustment coefficient of the grid scale based on the negative correlation mapping of the surface damage probability and the damage aggregation degree of the new grid cell, and calculating a similarity balancing term according to the maximum difference of the damage aggregation degrees of the original grid cells contained in the new grid cell;

[0029] calculating a target grid scale based on the adjustment coefficient and the similarity balancing term for adaptive adjustment.

[0030] As a preferred, the calculation of the influence factor of each pending damage area on the overall damage based on the internal detection sequence comprises:

[0031] constructing a three-dimensional data space of the internal detection sequence, and obtaining the variation trend similarity of the pending damage area and the overall damage in different time dimensions and different damage indicators;

[0032] obtaining the influence factor of the pending damage area on the overall damage based on the variation trend similarity of the same pending damage area in different three-dimensional data planes.

[0033] As a preferred, the construction of the three-dimensional data space of the internal detection sequence comprises:

[0034] the internal detection sequence of one pending damage area and the internal detection sequence of the overall combined structure are respectively expressed as a three-dimensional data matrix, which is denoted as a regional three-dimensional matrix and a global three-dimensional matrix; wherein the three dimensions of the regional three-dimensional matrix are damage indicators, time points and damage values; the three dimensions of the global three-dimensional matrix are damage indicators, time points and average damage values;

[0035] Project the region three-dimensional matrix and the global three-dimensional matrix into three-dimensional space respectively to obtain a plurality of region scatter points and global scatter points.

[0036] Preferably, the acquiring the similarity of the change trend of the three-dimensional data plane of the undetermined damage region and the overall damage in the time dimension and different damage indicators comprises:

[0037] For each undetermined damage region,

[0038] In the three-dimensional space, four region scatter points successively adjacent on the damage indicator and time point coordinate axes are selected to construct two triangular region planes, denoted as a first region plane and a second region plane; four global scatter points corresponding to the four region scatter points on the damage indicator and time point coordinate axes are selected to construct two triangular global planes, denoted as a first global plane and a second global plane;

[0039] Based on the spatial geometric relationship, the included angle of the first region plane and the first global plane and the included angle of the second region plane and the second global plane are calculated, and the change trend similarity of the undetermined damage region and the overall damage is obtained through a normalization transformation.

[0040] Preferably, the method further comprises:

[0041] According to the numerical range of the influence factor, the region is divided into a plurality of risk levels;

[0042] Based on the risk level, the region configuration of the detection sensor is optimized, and different detection frequencies are configured for regions of different risk levels.

[0043] In a second aspect, the present application provides a local damage detection and evaluation system applied to a composite structure, comprising:

[0044] A surface mesh division module is configured to divide the surface of the composite structure into a plurality of mesh units uniformly based on the physical size of the composite structure;

[0045] A surface damage extraction module is configured to extract the gray abnormal features of each mesh unit, calculate the surface damage probability of each mesh unit, and obtain the damage aggregation degree of each mesh unit based on the distribution of the surface damage probability in the mesh neighborhood.

[0046] A mesh scale optimization module is configured to adaptively adjust the scale of each mesh unit based on the surface damage probability and the damage aggregation degree, and obtain at least one undetermined damage region of the surface of the composite structure.

[0047] An internal damage detection module is configured to obtain the internal detection sequence of each undetermined damage region about a plurality of internal damage indicators at a plurality of time points.

[0048] A key area determination module is configured to calculate an influence factor of each to-be-determined damage area on the overall damage based on the internal detection sequence, and evaluate the key detection area of the combined structure.

[0049] The application discloses a local damage detection and evaluation method and system for a combined structure.

[0050] The application discloses a local damage detection and evaluation method and system for a combined structure. The application discloses a local damage detection and evaluation method and system for a combined structure.

[0051] Figure 1 is a flowchart of the application of a local damage detection and evaluation method for a combined structure.

[0052] Figure 2 is a structural block diagram of the application of a local damage detection and evaluation system for a combined structure. DETAILED DESCRIPTION

[0053] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the application.

[0054] To facilitate the understanding of the embodiments, first, a local damage detection and evaluation method for a combined structure disclosed in the embodiments of the application will be described in detail. As shown in Figure 1 the method comprises the following steps.

[0055] In step S1, the surface of the combined structure is uniformly divided into a plurality of grid cells based on the physical size of the combined structure.

[0056] Steel-concrete composite beam structure gives full play to the advantages of two different materials of steel tensile and concrete compression, and is a common structure form of long-span bridges such as arch bridges, cable-stayed bridges and suspension bridges. Traditional damage detection methods often use uniform point setting sensors,

[0057] The detection range and precision coverage of the key parts of the composite structure are insufficient, the local stress concentration area and the dynamic evolution of potential damage are ignored, the screening of the risk area lacks systematicness, the local characteristics and key areas of the damage distribution are not fully considered, resulting in low detection efficiency, waste of resources and easy to miss detection.

[0058] The method first uniformly divides the composite structure into a plurality of grid elements of the same size according to the physical size of the composite structure, maps the network element to a high-resolution grayscale image on the surface of the composite structure, and analyzes the pixel points in each grid element. Each grid area can be used as a small surface damage detection unit. Once an abnormal change occurs in a grid element, such as a crack, corrosion, etc., it can be discovered in time through the abnormality of the pixel points.

[0059] First, image preprocessing is performed, such as denoising, contrast enhancement, geometric correction, etc. In order to better analyze the detail features such as rust and cracks on the surface of the composite structure, the embodiment uses a median filter for image denoising, with a window size of 5x5 pixels; histogram equalization is used to enhance the contrast of the image, eliminate uneven illumination and noise interference, and improve the recognizability of damage features, providing a high-quality image data basis for subsequent feature extraction.

[0060] Step S2, extracting the gray abnormality features of each grid element calculates the surface damage probability of each grid element, and obtains the damage aggregation degree of each grid element based on the distribution of the surface damage probability in the grid neighborhood.

[0061] It can be understood that the surface damage of the composite structure includes rust, cracks, deformation, etc., and it is difficult to accurately identify each type of damage by a single image analysis. For example, the composite structure often has different degrees of cracks as the application time elapses, which will form low gray texture in the two-dimensional image, showing a large gray difference with the normal area which occupies the vast majority of the area, and showing a linear distribution. Therefore, the present application only needs to identify the area with a large deviation from the overall pixel points in the gray image in this step to represent the damage probability of the surface of the composite structure, without accurately identifying whether it is environmental pollution interference or real damage.

[0062] As a preferred, the step S2 of extracting the gray abnormality features of each grid element to calculate the surface damage probability of each grid element comprises the following sub-steps:

[0063] S21, calculate the global anomaly degree of the grid cell based on the degree of deviation of the gray value of the pixel in each grid cell from all pixels in the grayscale image.

[0064] As a specific implementation method, the global anomaly degree of each grid cell can be calculated using the following formula:

[0065]

[0066] in, Indicates the first Global anomaly degree of each grid cell, This represents the average gray level of all pixels in a grayscale image. Indicates the first The average gray level of all pixels in each grid cell is the standard deviation of grayscale values ​​for all pixels in a grayscale image.

[0067] S22, calculate the grayscale difference between each pixel in each grid cell and its neighboring pixels, and extract the local anomaly of each grid cell.

[0068] As a specific implementation method, the local anomaly degree of each grid cell can be calculated using the following formula:

[0069]

[0070] in, Indicates the first Local anomaly of each grid cell, For the first The number of pixels in each grid cell This indicates the first [cell] in the grid. The grayscale value of each pixel Indicates the first The average grayscale value of the neighboring pixels of a pixel is taken as the eight-neighborhood in this embodiment. This reflects the volatility of a pixel relative to its surroundings; the greater the volatility, the greater the likelihood of an anomaly.

[0071] S23, determine the texture direction of the pixels in the grid cell, and calculate the continuity of the local anomaly of the grid cell in the texture direction.

[0072] Preferably, determining the texture direction of pixels in the mesh unit in step S23 includes:

[0073] S231, calculate the average gradient of all pixels in the grid cell.

[0074] In this embodiment, Sobel operator is used to calculate the gray gradient of pixels to represent the texture direction in the grid cell. First, the gray gradient of pixels is calculated 、 , and the gradient direction can be obtained by

[0075] S232, taking any vertical direction of the direction corresponding to the average of the gradient as the texture direction of the pixel in the grid cell.

[0076] This step aims to analyze the texture in the image to find the possible cracks in the image. The crack of the combined structure is one of the textures in the image. The gradient represents the intensity of the change of pixel gray in a certain direction in the image. The position of the gradient change is the texture feature. At this time, the direction of the large change of the gradient is perpendicular to the direction of the crack.

[0077] As preferred, the continuity of the local anomaly degree of the grid cell in the texture direction calculated in step S23 is specifically:

[0078] S233, obtaining the first neighborhood grid of the grid cell within the preset field range, and obtaining the maximum value of the local anomaly degree in the first neighborhood grid;

[0079] S234, screening out the grid cell whose texture direction and the difference is less than a preset threshold value in the texture direction of the grid cell, and recording the grid cell as a second neighborhood grid;

[0080] S235, calculating the average of the local anomaly degree of the second neighborhood grid, and taking the negative correlation mapping of the ratio of the average of the local anomaly degree of the second neighborhood grid to the maximum value of the local anomaly degree in the first neighborhood grid as the continuity in the texture direction.

[0081] It can be understood that if the crack is along the texture direction , then the damage characteristics of the neighborhood grid in this direction should be most similar to the current grid. If the ratio is small, it means that the similarity in the crack direction is much higher than that in other neighborhood directions, reflecting the directional extension of the crack.

[0082] S24, calculating the surface damage probability of each grid cell based on the global anomaly degree, the local anomaly degree and the continuity. Specifically, the surface damage probability of the embodiment is represented as:

[0083]

[0084] wherein, represents a normalization function,​​ Indicates the degree of continuity along the texture direction. Surface damage probability. By combining the degree of anomaly of pixels with the degree of continuity of texture direction, a quantitative surface damage risk assessment index is generated.

[0085] Preferably, obtaining the damage aggregation degree of each grid cell based on the distribution of the surface damage probability in the grid neighborhood includes:

[0086] For each grid cell

[0087] The first deviation of the mesh element relative to the global surface damage probability is calculated based on the surface damage probability.

[0088] The second deviation of each neighboring grid cell of the grid cell relative to the global surface damage probability is calculated based on the surface damage probability, weighted by spatial distance.

[0089] The damage clustering degree of each grid cell is obtained by multiplying the first deviation and the second deviation and then standardizing the global data.

[0090] In one specific implementation, the degree of damage aggregation of mesh elements can be calculated using the following formula:

[0091]

[0092] in, Indicates the first The degree of damage aggregation in each grid cell Indicates the first The first grid cell in the neighborhood of the first grid cell One grid cell, Indicates the first The surface damage probability of each grid cell Indicates the global surface damage probability. Indicates the first The surface damage probability of each grid cell. Represents the spatial weight matrix, representing the position. and The spatial relationship between them, the weight values ​​are based on and The spatial distance between them is determined. This is the first deviation. This is the second deviation. The variance of the surface damage probability of all grid cells is used to standardize the first bias. If the overall variance of the damage probability is large, it means that the damage probability values ​​vary greatly. In this case, the bias of a single grid cell is relatively unimportant, and dividing by the variance will reduce its impact. Reflects the first The abnormal intensity of the probability of damage to a single grid cell is used as a weighting factor for the deviation of neighboring grid cells.

[0093] A larger value indicates stronger spatial clustering of damage in these areas. If a mesh cell is located at a weld, a large positive weighting factor indicates that the damage probability of that mesh is significantly higher than the average. If its neighboring meshes also have high damage probabilities, the degree of damage clustering will be positive and large, confirming that this area is a damage hotspot, requiring mesh refinement for focused inspection.

[0094] This invention identifies “hotspot” regions of damage by quantifying the degree of damage aggregation, i.e., regions with high damage probability and spatial clustering. Combined with the surface damage probability, it guides the optimization of grid division, thereby improving detection efficiency and quality.

[0095] Step S3 involves adaptively adjusting the scale of each mesh element based on the surface damage probability and the degree of damage aggregation to obtain at least one undetermined damage region on the surface of the combined structure. Preferably, this includes the following sub-steps:

[0096] S31, based on the degree of damage aggregation of the grid cells, adjacent grid cells are merged. The surface damage probability of the merged new grid cell is the maximum value among the surface damage probabilities of each grid cell in the new grid cell. The degree of damage aggregation of the new grid cell is the area-weighted average of the degree of damage aggregation of each grid cell in the new grid cell.

[0097] Specifically, if the absolute value of the difference in damage clustering between two adjacent grid cells is less than a threshold, the two grid cells are merged. The defect probability of the merged grid is the maximum of the two grid cells (focusing on the more severe damage), and the damage clustering is the weighted average of the two grid cells (the weights are obtained based on the area of ​​the grid cells).

[0098] For example, the degree of damage aggregation in the merged new mesh elements can be calculated using the following formula:

[0099]

[0100] in, The degree of damage aggregation in the new grid cells, The original first element contained in the new grid cell The area of ​​each grid cell.

[0101] For each new grid cell, an adjustment coefficient of the grid size is calculated based on a negative correlation mapping of the surface damage probability and the damage aggregation degree of the new grid cell, and a similarity balancing term in the new grid cell is calculated according to a maximum difference of the damage aggregation degrees of the original grid cells contained in the new grid cell;

[0102] The target grid size is calculated based on the adjustment coefficient and the similarity balancing term to adaptively adjust.

[0103] In a specific embodiment, the target grid size can be calculated by the following formula:

[0104]

[0105] wherein, is the target grid size, i.e. the adjusted grid cell size, is the initial uniform division grid cell size, , , is a weight adjustment coefficient, denotes a natural logarithm, denotes a standard deviation of the damage aggregation degrees of all grid cells before merging, , respectively denote a maximum value and a minimum value of the damage aggregation degrees of the original grid cells contained in the new grid cell after merging. is the adjustment coefficient, is the similarity balancing term.

[0106] It should be understood that the greater the surface damage probability, the finer the grid cell should be divided, so that the denominator increases and the fraction value decreases, and finally the grid cell size decreases. The damage aggregation degree reflects the spatial aggregation, and the greater the value, the more the region is a damage hotspot, which requires finer grid. If the damage aggregation degrees of each original grid cell contained in the new grid cell are similar, it means that they belong to the same damage mode region, and similar detection accuracy should be used, i.e. the division granularity does not need to be too refined, which is used to balance the similarity restriction inside the new grid cell. And the weight adjustment coefficient is small, ensuring that the similarity balance is a secondary factor and will not cover the main damage feature judgment.

[0107] Step S4, obtaining an internal detection sequence of each to-be-determined damage region at multiple time points with respect to multiple internal damage indicators.

[0108] Given the complexity of composite structures, single image processing has many shortcomings. For example, the surface of the structure is often contaminated with a lot of surface pollution, such as vehicle dust and oil stains, which obscure the true damage features. The connection between steel and concrete (such as shear keys) is a critical stress-bearing part and a region prone to damage, but these parts are often complex in structure and difficult to capture clearly in the image. Creep, shrinkage and foundation settlement of bridges, or vibration and deformation of bridges under the action of vehicles and wind, can cause image blurring or feature displacement, affecting image-based feature extraction.

[0109] Therefore, in the specific implementation process, contact sensors such as acoustic emission sensor arrays, strain gauges, and ultrasonic detectors can be used to collect internal damage index data such as strain data, displacement data, acoustic emission data, and vibration mode data of the combined structure. By combining surface image detection and internal structure detection, a variety of factors affecting the local damage of the combined structure are comprehensively considered, and the actual degree of damage can be grasped more accurately.

[0110] Specifically, sensors, such as acoustic emission sensors, strain gauges, or non-destructive testing equipment such as ultrasonic sensors, are placed inside each area of ​​potential damage to obtain time-series data, which serves as the internal testing sequence.

[0111] It should be understood that most of these sensors are point sensors, which are costly to deploy. Therefore, this invention assesses the impact of different areas on the overall damage through a one-time detection, which can provide a scientific basis for the subsequent deployment of sensors, reduce deployment costs, and avoid ineffective monitoring, making the monitoring of combined structures more targeted.

[0112] Step S5: Calculate the influence factor of each undetermined damage region on the overall damage based on the internal detection sequence, and evaluate the key detection regions of the combined structure. Preferably, this includes the following sub-steps:

[0113] S51, construct the three-dimensional data space of the internal detection sequence.

[0114] S511, the internal detection sequence of a region to be damaged and the internal detection sequence of the overall combined structure are respectively represented as a three-dimensional data matrix, denoted as the regional three-dimensional matrix and the global three-dimensional matrix; wherein the three dimensions of the regional three-dimensional matrix are damage index, time point and damage value; the three dimensions of the global three-dimensional matrix are damage index, time point and average damage value.

[0115] Specifically, the internal detection sequences are each represented as a three-dimensional data matrix. ,in Indicates the first One damage index, Indicates the first Each detection time, represents the first damage index data of the undetermined damage area acquired at the first detection time. The same damage index data of the undetermined damage area at a certain time is averaged to obtain the overall three-dimensional data at the time, and the overall three-dimensional data at all times and all damage indexes form a global three-dimensional matrix , represents the first damage index data of the combined structure as a whole acquired at the first detection time. The first damage index data of the combined structure as a whole acquired at the first detection time is obtained by averaging the first damage index data of all undetermined damage areas acquired at the first detection time.

[0116] S512, respectively project the region three-dimensional matrix and the global three-dimensional matrix into a three-dimensional space to obtain a plurality of region scatter points and global scatter points.

[0117] It can be understood that the present embodiment needs to obtain the damage correlation degree of each undetermined damage area and the combined structure as a whole, and if the internal detection sequence of a certain undetermined damage area is consistent with the internal detection sequence of the overall region in time trend and different indexes, it means that the correlation degree is large, that is, the local damage of the undetermined damage area has a large influence factor on the damage of the combined structure as a whole. Therefore, the present application considers projecting the data points into a three-dimensional space, and reflecting the correlation degree by evaluating the change relationship of two data planes.

[0118] Specifically, a three-dimensional coordinate system with damage index, time point and index data value as three coordinate axes can be established, and the region three-dimensional matrix and represent in this three-dimensional coordinate system, so that a plurality of region scatter points and global scatter points can be obtained, and each scatter point represents a group of three-dimensional data.

[0119] S52, obtain the change trend similarity of the three-dimensional data planes of the undetermined damage area and the overall damage in different time dimensions and different damage indexes;

[0120] S521, for each undetermined damage area,

[0121] In the three-dimensional space, four region scatter points successively adjacent on the damage index and time point coordinate axes are selected to construct two triangular region planes, which are recorded as a first region plane and a second region plane; four global scatter points corresponding to the four region scatter points on the damage index and time point coordinate axes are selected to construct two triangular global planes, which are recorded as a first global plane and a second global plane.

[0122] ​​​​Specifically, four regional scatter points are selected 、 、 、 Three points in space can construct a plane, and in an example embodiment, the first regional plane is constructed by selecting 、 、 、 、 The second regional plane is constructed by selecting 、 、 、 The first global plane is constructed by selecting 、 、 The second global plane is constructed by selecting 、 、

[0123] S522, based on the spatial geometric relationship, the angles between the first regional plane and the first global plane and the angles between the second regional plane and the second global plane are calculated, and a normalized transformation is performed to obtain the change trend similarity of the undetermined damage region and the overall damage.

[0124] Specifically, S5221, first, the normal vector angles of the first regional plane and the first global plane are calculated, and the normal vector of the first regional plane is , and the normal vector of the first global plane is . The comparison of the normal vectors of a single plane may have directional ambiguity, and the combination of two planes provides more stable geometric features, which can accurately reflect the correlation between the damage data of different undetermined damage regions and the overall damage data.

[0125] The normal vector angle of the two planes is

[0126]

[0127] wherein,

[0128]

[0129]

[0130]

[0131]

[0132] ​​As those skilled in the art will understand, similarly, the angle between the normal vectors of the second region plane and the second global plane can be obtained based on spatial geometry. This embodiment will not be described in detail.

[0133] S5222, determine the relative direction of the first region plane and the first global plane. If they are in the same direction, take the positive value of the cosine of the angle between the normal vectors of the first region plane and the first global plane; otherwise, take the negative value of the cosine of the angle between the normal vectors.

[0134] Since the angle between two planes can be represented in space as two complementary angles, representing the same trend change and opposite trend change of the two data planes respectively, this embodiment constructs an auxiliary vector to determine whether the normal vectors of the two planes are on the same side of this auxiliary vector. If they are on the same side, it means that the two data planes have the same trend change, that is, the damage development trend is consistent; otherwise, the two data planes have the opposite trend change, that is, the damage development trend is opposite.

[0135] Specifically, scattered points on the first region plane can be obtained. , The midpoint of the line connects the global scattered points on the first global plane. Construct an auxiliary vector with the midpoint, and calculate the projection of the auxiliary vector onto the normal vectors of the first region plane and the first global plane. If the two projections have the same sign, it means that the two data planes are on the same side of the reference direction. The cosine value of the normal vector is positive, and if they have different signs, it is negative.

[0136] S5223, after normalization transformation, the similarity of the change trend between the damage area to be determined and the overall damage is obtained. Preferably, this embodiment calculates the similarity of the change trend using the following formula:

[0137]

[0138] in, For similarity in changing trends, , Represents the weighting coefficient, and , This represents the cosine value of the normal vectors of the first region plane and the first global plane after taking the sign according to step S5222. The cosine values ​​of the normal vectors of the second region plane and the second global plane are represented by the values ​​after taking the sign according to step S5222.

[0139] S53, based on the similarity of the change trends of the same undetermined damage area under different three-dimensional data planes, obtain the influence factor of the undetermined damage area on the overall damage.

[0140] Specifically, the average of the similarity of the change trends obtained at different indicators and different time points of the same to-be-determined damage area is the influence factor of the to-be-determined damage area on the overall damage.

[0141] The present application can more finely capture the complex different-dimensional correlation mode between the local damage and the overall damage by cross-verification of the correlation between the damage data of different to-be-determined damage areas and the overall damage data through two independent geometric measures, and simultaneously considers the relationship of the data at different time points and different detection indicators, which is helpful to identify the key area in the combined structural damage detection.

[0142] The damage of the steel plate combined structure often occurs in the key but hidden areas such as the weld, the bolt connection, and the shear key joint. The traditional point sensor is difficult to cover comprehensively, and the detection process has low automation degree. The present application quantifies the contribution degree of each local area damage to the overall damage, and identifies the key damage area. The greater the influence factor, the greater the influence of the area damage on the overall structure safety, and the monitoring needs to be strengthened. According to the influence factor, the influence degree of different areas on the overall damage is sorted, so that the subsequent monitoring task can quickly respond to the change of the structure risk state, and ensure the timeliness and accuracy of the evaluation result.

[0143] As a preferred, the method further comprises a step S6:

[0144] According to the numerical range of the influence factor, the area is divided into multiple risk levels;

[0145] Based on the risk level, the area configuration of the detection sensor is optimized, and different detection frequencies are configured for areas of different risk levels.

[0146] In local damage detection, more attention needs to be paid to the areas with high correlation with the overall damage of the combined structure. By accurately identifying the key damage areas that have greater influence on the overall structure safety, the number and position of the detection sensors of different risk level areas are reconfigured, and a differentiated detection frequency and detection means configuration scheme can be formulated. For example, high-frequency multi-sensor fusion detection is used for key areas, important area detection is used for important areas, general area detection is used for general areas, and low-frequency sampling inspection is used for low-risk areas. The present application can provide a scientific priority ranking for maintenance decision-making, so that limited maintenance resources can be preferentially invested in the key parts that most affect the structure safety, and over-detection of low-risk areas can be avoided.

[0147] It can be understood that in engineering applications, it is necessary to understand the behavior and real-time state of the combined structure under load, track detection and damage analysis, timely optimize the design and prevent abnormalities, degradation and local damage that may occur during service and operation, and provide support for long-term performance guarantee. The present application completes the surface anomaly detection of the combined structure through image processing technology, analyzes the current real damage state by using the historical internal damage index data collected by the existing sensor within a period of time, quantifies the damage influence of different regions through adaptive non-uniform grid division, and can identify hidden damage areas and damage aggregation trends that are difficult to find by traditional methods. Based on this identification, the most critical damage area for the safety of the structure is identified to provide a basis for maintenance priority ranking, and the number and position of sensors in each detection area can be further optimized to provide a basis for subsequent targeted monitoring.

[0148] It should be understood that although each step in the flowchart involved in each embodiment as described above is displayed in sequence according to the arrow, these steps are not necessarily executed in the order indicated by the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, at least part of the steps in the flowchart involved in each embodiment as described above can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or steps or stages in other steps.

[0149] Based on the same inventive concept, the embodiments of the present application also provide a system for implementing the above-mentioned local damage detection and evaluation. The implementation scheme for solving the problem provided by the system is similar to the implementation scheme described in the above method, so the specific limitations in the following embodiments of the local damage detection and evaluation system applied to the combined structure can be referred to the limitations of the method in the above text, which will not be repeated here.

[0150] As shown in Figure 2 , the present application also provides a local damage detection and evaluation system applied to a combined structure, comprising:

[0151] a surface grid division module for uniformly dividing the surface of the combined structure into a plurality of grid cells based on the physical size of the combined structure;

[0152] a surface damage extraction module for extracting the gray scale anomaly features of each grid cell, calculating the surface damage probability of each grid cell, and obtaining the damage aggregation degree of each grid cell based on the distribution of the surface damage probability in the grid neighborhood;

[0153] a mesh scale optimization module, configured to adaptively adjust a scale of each mesh unit based on the surface damage probability and the damage aggregation degree, to obtain at least one to-be-determined damage area of the combined structure surface;

[0154] an internal damage detection module, configured to obtain internal detection sequences of each to-be-determined damage area with respect to a plurality of internal damage indexes at a plurality of time points;

[0155] a key area determination module, configured to calculate an influence factor of each to-be-determined damage area on the overall damage based on the internal detection sequences, and evaluate a key detection area of the combined structure.

[0156] It should be noted that the local damage detection and evaluation system for the combined structure provided in the embodiment is used for processing the local damage detection of the steel plate combined structure, and only the division of the above functional modules is used as an example for illustration. In actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the system is divided into different functional modules. Each functional module can be composed of a single execution unit, or two or more execution units can be integrated into a functional module to realize the overall function of the functional module.

[0157] Those skilled in the art can understand that each of the above modules can be realized by software, hardware and a combination thereof in whole or in part. Each of the above modules can be embedded in or independent of a processor in a computer device in hardware form, or can be stored in a memory in a computer device in software form, so as to be called and executed by a processor to perform the operations corresponding to each of the above modules.

[0158] The present application is not limited to the above specific embodiments, and those skilled in the art can make various modifications without creative labor, which are all within the protection scope of the present application.

Claims

1. A method for evaluating local damage detection applied to a composite structure, characterized by, The method comprises the following steps: uniformly dividing a surface of a combined structure into a plurality of grid cells based on physical dimensions of the combined structure; extracting gray-scale anomaly features of each grid cell to calculate a surface damage probability of each grid cell, including: calculating a global anomaly degree of a grid cell based on a deviation degree of a gray-scale value of a pixel point in the grid cell relative to all pixel points in a gray-scale image; calculating a local anomaly degree of each grid cell by calculating gray-scale differences between each pixel point and neighboring pixel points in the grid cell; determining a texture direction of a pixel point in the grid cell to calculate a continuity degree of the local anomaly degree of the grid cell in the texture direction; calculating a surface damage probability of each grid cell based on the global anomaly degree, the local anomaly degree, and the continuity degree; obtaining a damage aggregation degree of each grid cell based on a distribution of the surface damage probability in a grid neighborhood; adaptively adjusting a size of each grid cell based on the surface damage probability and the damage aggregation degree to obtain at least one to-be-determined damage region of the surface of the combined structure; obtaining an internal detection sequence of each to-be-determined damage region with respect to a plurality of internal damage indicators at a plurality of time points; calculating an influence factor of each to-be-determined damage region on overall damage based on the internal detection sequence, including constructing a three-dimensional data space of the internal detection sequence to obtain a change trend similarity of the to-be-determined damage region and the overall damage in different time dimensions and different damage indicators; obtaining the influence factor of the to-be-determined damage region on the overall damage based on the change trend similarity in different three-dimensional data planes of the same to-be-determined damage region; evaluating a key detection region of the combined structure.

2. The method for partial damage detection evaluation applied to a combined structure according to claim 1, wherein The determination of the texture direction of the pixel point in the grid cell comprises: calculating a gradient mean value of all pixel points in the grid cell; taking any vertical direction of the direction corresponding to the gradient mean value as the texture direction of the pixel point in the grid cell.

3. The method for partial damage detection evaluation applied to a combined structure according to claim 1, wherein The obtaining of the damage aggregation degree of each grid cell based on the distribution of the surface damage probability in the grid neighborhood comprises: for each grid cell, calculating a first deviation of the grid cell relative to a global surface damage probability based on the surface damage probability; calculating a second deviation of each neighboring grid cell of the grid cell relative to the global surface damage probability under spatial distance weighting based on the surface damage probability; obtaining the damage aggregation degree of each grid cell by performing global data standardization on a product of the first deviation and the second deviation.

4. The method for partial damage detection evaluation applied to a combined structure according to claim 2, wherein The adaptive adjustment of the size of each grid cell based on the surface damage probability and the damage aggregation degree comprises: merging adjacent grid cells based on the damage aggregation degree of the grid cells, and taking a maximum value of surface damage probabilities of the grid cells constituting a new grid cell as a surface damage probability of the new grid cell after the merging, and taking an area-weighted average of damage aggregation degrees of the grid cells constituting the new grid cell as a damage aggregation degree of the new grid cell. For each new grid cell, an adjustment coefficient of the grid size is calculated based on a surface damage probability of the new grid cell and a negative correlation mapping of a damage aggregation degree, and a similarity balancing term is calculated according to a maximum difference of the damage aggregation degree of the original grid cells contained in the new grid cell; A target grid size is calculated based on the adjustment coefficient and the similarity balancing term to adaptively adjust.

5. The method for partial damage detection evaluation applied to a combined structure according to claim 1, wherein The three-dimensional data space for constructing the internal detection sequence includes: The internal detection sequence of a to-be-determined damage area and the internal detection sequence of the whole combined structure are respectively expressed as a three-dimensional data matrix, denoted as an area three-dimensional matrix and a global three-dimensional matrix; three dimensions of the area three-dimensional matrix are damage indicators, time points and damage values; three dimensions of the global three-dimensional matrix are damage indicators, time points and average damage values; The area three-dimensional matrix and the global three-dimensional matrix are respectively projected into a three-dimensional space to obtain a plurality of area scatter points and global scatter points.

6. The method for partial damage detection evaluation applied to a combined structure according to claim 5, wherein The similarity of the change trend of the three-dimensional data plane of the to-be-determined damage area and the whole damage in the time dimension and different damage indicators includes: For each to-be-determined damage area, In the three-dimensional space, four area scatter points successively adjacent on the damage indicator and time point coordinate axes are selected to construct two triangular area planes, denoted as a first area plane and a second area plane; four global scatter points corresponding to the four area scatter points on the damage indicator and time point coordinate axes are selected to construct two triangular global planes, denoted as a first global plane and a second global plane; Based on the spatial geometric relationship, the included angle of the first area plane and the first global plane and the included angle of the second area plane and the second global plane are calculated, and the change trend similarity of the to-be-determined damage area and the whole damage is obtained through a normalization transformation.

7. The method for partial damage detection evaluation applied to a combined structure according to claim 1, wherein The method further includes: According to the numerical range of the influence factor, the area is divided into a plurality of risk levels; Based on the risk level, the area configuration of the detection sensor is optimized, and different detection frequencies are configured for areas of different risk levels.

8. A system for detecting and assessing localized damage applied to a composite structure, comprising: It includes: A surface grid division module is configured to uniformly divide the surface of the combined structure into a plurality of grid cells based on the physical size of the combined structure; A surface damage extraction module is configured to extract the gray abnormal features of each grid cell to calculate the surface damage probability of each grid cell, including: calculating the global abnormality degree of each grid cell based on the deviation of the gray value of each pixel point in the grid cell from all pixel points in the gray image; calculating the gray difference between each pixel point and the neighborhood pixel points in each grid cell to extract the local abnormality degree of each grid cell; determining the texture direction of the pixel points in the grid cell, and calculating the continuity of the local abnormality degree of the grid cell in the texture direction; calculating the surface damage probability of each grid cell based on the global abnormality degree, the local abnormality degree and the continuity; and obtaining the damage aggregation degree of each grid cell based on the distribution of the surface damage probability in the grid neighborhood. a mesh scale optimization module configured to adaptively adjust a scale of each mesh unit based on the surface damage probability and the damage aggregation degree, to obtain at least one to-be-determined damage region of the combined structure surface; an internal damage detection module configured to obtain internal detection sequences of each to-be-determined damage region with respect to a plurality of internal damage indicators at a plurality of time points; a key region determination module configured to calculate an influence factor of each to-be-determined damage region on the overall damage based on the internal detection sequences, and to evaluate a key detection region of the combined structure; the key region determination module is specifically configured to: construct a three-dimensional data space of the internal detection sequences, and obtain a change trend similarity of a to-be-determined damage region and the overall damage in different time dimensions and different damage indicators; and obtain the influence factor of the to-be-determined damage region on the overall damage based on the change trend similarity of the to-be-determined damage region in different three-dimensional data planes.

Images