Structure surface layer loss state prediction method and system based on dam engineering

Abstract

The invention discloses a structure surface layer loss state prediction method and system based on dam engineering, belongs to the technical field of dam leakage prediction, and aims to solve the problems that a hidden danger area on the surface of a structural surface of a long-shoreline dam in an existing flowing water area is not accurately identified; and the anti-fatigue damage degree of the structure is difficult to predict for the potential safety hazards. The method comprises the following steps: performing core node deployment scanning under multiple points on a risk area in a dam to obtain a comprehensive structure surface image; performing structural surface hidden danger feature recognition and extraction processing of a related pixel point matrix on the initial risk hidden danger image, and determining a structural surface hidden danger image; space-time fusion processing of multi-source external hidden danger factors is carried out on the corresponding dam sections, and external environment hidden danger features are determined; determining a hidden danger growth correlation degree between the structure surface hidden danger feature and the external environment hidden danger feature; and performing damage growth prediction on the hidden danger area of the structural surface layer of the dam, and generating a maintenance inspection scheme of the dam.

Description

Technical Field

[0001] The present application relates to the technical field of dam leakage prediction, and in particular to a method and system for predicting the structural surface loss state based on dam engineering. Background Art

[0002] As water conservancy projects become more and more perfect, the dikes have long coastlines, wide distribution, many dangerous situations, complex and changeable causes of danger, and difficulty in predicting dangerous situations, which seriously threaten people's lives and property safety. There are many factors that affect the safety of dikes, mainly seepage, rain conditions, soil characteristics, etc. The impact of various factors on the safety of dikes is a complex system, so safety warnings for cracks, seepage, flow, settlement and other hidden dangers of dikes are of great significance to ensuring the safety of dikes and improving disaster prevention and mitigation.

[0003] Most of the existing flood and tide prevention dikes and seawalls along the banks of rivers, around lakes and depressions, and along the coast are earth dikes that are built year by year by taking soil from nearby sections. Some dike foundations have not been treated and there are many construction defects. In addition, the dike line is very long and the management is scattered, which is easy to be damaged by nature or man-made damage, which will form hidden dangers. If not handled in time, dangerous situations will occur at high water levels, and even cause dike breach accidents and losses. Dike hidden dangers are not easy to detect on the surface, and if they are not handled thoroughly, they may recur.

[0004] Traditional dam safety monitoring mainly includes methods such as laser radar and manual inspection, or a combination of sensors and manual observation for disaster warning. However, this method cannot achieve non-contact and comprehensive coverage monitoring of the dam protection surface, and there are certain risks in placing and disassembling instruments, which is not conducive to large-scale accurate safety monitoring and risk prevention and control of dams in flowing waters along long coastlines. In addition, in the face of long dam coastlines, the identification error of hidden danger areas is large, and the time cost and labor cost are too high. It is impossible to scientifically target the actual situation of dam safety hazards and form targeted, planned and reasonable hidden danger maintenance and treatment. It is difficult to monitor and predict the loss (fatigue damage strength) of the structural surface of dams along long coastlines in flowing waters. Therefore, there is an urgent need for a convenient and fast dam risk prediction method that can be combined with actual risk characteristics, so as to scientifically and reasonably maintain and repair dams and improve the overall safety of long coastline dams. Summary of the invention

[0005] The embodiments of the present application provide a method and system for predicting the structural surface loss state based on dam projects, which are used to solve the following technical problems: the identification of surface potential hazard areas of structural surfaces of long shoreline dams in existing flowing waters is not very accurate, and it is difficult to predict the degree of fatigue damage of the structure based on these safety hazards, which is not conducive to the planned maintenance and repair of the dams.

[0006] The present application embodiment adopts the following technical solutions: On the one hand, an embodiment of the present application provides a method for predicting the structural surface loss state based on a dam project, comprising: according to a preset long shoreline dam detection route, and through a number of drones, a core node deployment scan is performed at multiple points on the risk area in the dam to obtain a comprehensive structural surface image; wherein the comprehensive structural surface image includes infrared thermal imaging information, contour display information and real structural surface information; the external thermal imaging information, the contour display information and the real structural surface information in the comprehensive structural surface image are pre-screened for risk hazard characteristics to obtain an initial risk hazard image; and the initial risk hazard image is subjected to structural surface hazard feature identification and extraction processing related to the pixel matrix to determine a structural surface hazard image; through the GIS information of the structural surface hazard image, the corresponding dam section is subjected to spatiotemporal fusion processing of multi-source external hazard factors, and based on the degree of decentralized influence of the fused data on the dam section, the dam section is determined. the external environment hidden danger characteristics corresponding to the GIS information; the structural surface hidden danger characteristics in the structural surface hidden danger image and the external environment hidden danger characteristics are calculated based on the grayscale correlation under subjective and objective factors to obtain the hidden danger growth correlation between the structural surface hidden danger characteristics and the external environment hidden danger characteristics; wherein the structural surface hidden danger characteristics are subjective factor characteristics, and the external environment hidden danger characteristics are objective factor characteristics; the hidden danger growth correlation, the structural surface hidden danger characteristics and the external environment hidden danger characteristics are input into the pre-trained structural surface loss prediction model; and through the damage accumulation criterion in the structural surface loss prediction model, the damage growth prediction is performed on the structural surface hidden danger area of ​​the dam to obtain the structural damage loss intensity of each structural surface hidden danger area; based on the risk level of the structural damage loss intensity, the information planning processing of the maintenance route and maintenance time of the structural surface hidden danger area is performed to generate the maintenance inspection plan of the dam.

[0007] The embodiment of the present application can obtain the state of the surface of the dam structure more comprehensively by combining infrared thermal imaging, contour display and real structural surface information, thereby improving the accuracy of prediction. The acquired comprehensive structural surface image can be pre-screened for risk hidden danger features, and potential problem areas can be quickly identified to improve detection efficiency. The hidden danger features of the surface structure can be accurately identified and extracted, providing a reliable basis for loss prediction. The impact of the external environment on the dam structure can also be considered more comprehensively. Through the grayscale correlation calculation of subjective and objective factors, the correlation between the hidden danger features of the surface structure and the hidden danger features of the external environment can be evaluated, which helps to more accurately predict the growth trend of hidden dangers. At the same time, using the pre-trained structural surface loss prediction model, the damage accumulation prediction of the structural surface hidden danger area of ​​the dam can be made, thereby predicting the structural damage loss intensity. And based on the risk level of the structural damage loss intensity, the relevant maintenance routes and maintenance time can be planned, which helps to prevent and maintain the safety of the dam in advance. Finally, by timely identifying and predicting potential structural problems, the safety of the dam can be enhanced and the risk of disasters can be reduced.

[0008] In a feasible implementation manner, before the structural surface hidden danger features of the comprehensive structural surface image of the dam are identified and extracted to determine the structural surface hidden danger image, according to the preset long shoreline dam detection route, and through a number of drones, the risk area in the dam is scanned at multiple points using core nodes to obtain a comprehensive structural surface image, which specifically includes: based on the three-dimensional modeling diagram of the dam section to be detected, the location of the structural hidden danger area is identified; wherein the location of the structural hidden danger area includes at least: the location of the dam corner area, the location of the expansion joint area, and the location of the dam connection; according to the location of the structural hidden danger area, the plan view in the three-dimensional modeling of the dam section is rasterized to generate a number of search grids; and based on the location of the structural hidden danger area, the corresponding search grid is marked and deployed with core nodes to determine the core grid area; the core grid area is determined as the center point, and twice the search grid area is divided into the core grid area. The diagonal length of the search grid is the radius, and the risk area is divided; according to the center point of the search grid occupied by the risk area, and through deep optimization of the search direction and the ox-plowing method with different starting points, the risk area is subjected to full coverage path planning processing of the current cluster of the starting point, and the risk area detection route is obtained; wherein, the risk area detection route is the flight route of the UAV; based on the starting and ending point positions of the risk area detection route, and through the ox-plowing method, the search grid in the plan view is subjected to full coverage path planning processing, and the long shoreline dam detection route is obtained; wherein, the long shoreline dam detection route includes: a general area detection route and the risk area detection route; according to the long shoreline dam detection route, and through the various acquisition devices mounted on the UAV, the surface image of the comprehensive structure is collected; wherein, the various acquisition devices include: ultra-high-definition camera equipment, infrared thermal imager and total station.

[0009] The application embodiment uses a three-dimensional modeling diagram to identify the locations of key risk areas such as dam corners, expansion joints and joints, which helps to concentrate resources for key inspections. Using drones for multi-point scanning can quickly cover a large area and improve inspection efficiency. Through rasterization processing and planar top-view diagrams, search grids and core nodes can be systematically planned to ensure the comprehensiveness and systematicness of inspections. Using the ox-plowing method for path planning can ensure full coverage of risk areas while optimizing the flight path of drones.

[0010] The combined use of ultra-high-definition camera equipment, infrared thermal imagers and total stations can provide high-quality multi-source data, which helps to more accurately assess the status of the dam. Automated inspection route planning and flight paths reduce the need for manual intervention and reduce human errors. Through the efficient inspection of drones, the cost and time required for traditional manual inspections are reduced. Drones can be deployed quickly, increasing the frequency of inspections and ensuring real-time monitoring of dam conditions. Drones can be used to inspect in dangerous or inaccessible areas, reducing safety risks for personnel. The comprehensive structural surface images collected provide rich basic data for subsequent data analysis and risk assessment.

[0011] In a feasible implementation manner, before the structural surface hidden danger feature identification and extraction processing is performed on the comprehensive structural surface image and the structural surface hidden danger image is determined, it is also necessary to pre-screen the risk hidden danger features of the external thermal imaging information, the contour display information and the real structural surface information in the comprehensive structural surface image to obtain an initial risk hidden danger image, specifically including: generating an infrared heat source distribution map for the external thermal imaging information of the surface of the comprehensive dam structure; based on the current weather data information, removing the initial heat source abnormal area in the infrared heat source distribution map, and matching the remaining heat source abnormal area with the temperature difference gradient map of the water level in the current flowing water area by the fluctuation range threshold of the dynamic temperature difference line to determine the final heat source abnormal area based on the comprehensive structural surface image and the corresponding heat source abnormal coordinate position; based on the three-dimensional modeling map of the dam section to be detected, converting the contour display information in the comprehensive structural surface image into a three-dimensional matrix to obtain a three-dimensional contour model map; and matching the contour abnormalities in the three-dimensional contour model map according to the GIS terrain feature information of the current dam section. The abnormal area is processed by abnormal identification, and the contour abnormal area and the corresponding contour abnormal coordinate position under the comprehensive structure surface image are determined; the heat source abnormal coordinate position in the comprehensive structure surface image is matched with the contour abnormal coordinate position within the threshold of the coordinate area to determine the contour-heat source abnormal area; the contour abnormal area corresponding to the contour-heat source abnormal area and the final heat source abnormal area are scored for risk hidden danger characteristics to obtain the contour-heat source abnormal score; if the contour-heat source abnormal score is greater than the first preset threshold, the real structure surface information corresponding to the contour-heat source abnormal area is extracted; through the preset dam surface structure abnormal template, the real structure surface information is matched and queried for information on semantics and interest pixel areas to obtain the structural risk hidden danger characteristics contained in the real structure surface information; wherein the structural risk hidden danger characteristics include: transverse cracks, longitudinal cracks, oblique cracks, holes and settlement extension zones; the contour-heat source abnormal area image with the structural risk hidden danger characteristics is determined as the initial risk hidden danger image.

[0012] The embodiment of the present application generates a heat source distribution map through infrared thermal imaging information, and combines weather data and a water level temperature gradient map to accurately identify abnormal heat source areas and improve the accuracy of hidden danger detection.

[0013] By matching the fluctuation range threshold of the dynamic temperature difference line, the identification criteria of abnormal areas can be adjusted dynamically to adapt to different environmental conditions. The contour display information is converted into a three-dimensional matrix to generate a three-dimensional contour model map, which provides a basis for subsequent three-dimensional analysis. Combined with GIS terrain feature information, the contour abnormal area can be identified more accurately and the reliability of hidden danger detection can be enhanced. The abnormal coordinate position of the heat source and the abnormal coordinate position of the contour line are matched for consistency, and multiple data sources are integrated to improve the comprehensiveness and comprehensiveness of hidden danger detection. Through scoring processing, the characteristics of the contour-heat source abnormal area are quantified to provide a basis for risk assessment.

[0014] Extracting structural risk hidden features from real structural surface information, such as cracks, holes, and settlement, helps to quickly locate potential danger points. Using the dam surface structural anomaly template for information matching query can quickly identify common structural risk hidden features. Automated data processing and risk assessment processes improve detection efficiency and reduce manual analysis time.

[0015] In a feasible implementation manner, the structural surface hidden danger feature identification and extraction processing of the pixel matrix is ​​performed on the initial risk hidden danger image to determine the structural surface hidden danger image, which specifically includes: performing grayscale uniformization processing on the initial risk hidden danger image according to the preset grayscale level and grayscale uniform distribution probability value to obtain a grayscale risk hidden danger image; based on the image pixel value after filtering the grayscale risk hidden danger image, performing detail feature extraction on the pixel features in the structural hidden danger field related to the weight change coefficient to obtain the structural hidden danger detail feature; based on the grayscale feature, color feature and texture feature in the structural hidden danger detail feature, and based on the preset segmentation edge and the corresponding two-dimensional structural hidden danger feature parameters, performing segmentation processing on the complementary overlapping image area of ​​the grayscale risk hidden danger image to obtain the structural hidden danger pixel area; through the attention mechanism, performing positioning processing on the edge pixels of the structural hidden danger pixel area under the edge regularization item, and outputting the edge pixel distribution position; according to the edge pixel distribution position, performing surface size calculation processing on the structural hidden danger pixel area to obtain The surface size parameters of the structural hidden danger pixel area; wherein the surface size parameters include: structural hidden danger length, structural hidden danger width, structural hidden danger area and structural hidden danger extension trend; the color change calculation of the pixel matrix is ​​performed on the central pixel area surrounded by the edge pixel distribution position, and based on the heat source characteristics of the final heat source abnormal area in the initial risk hidden danger image, the spatial depth calculation is performed on the structural hidden danger pixel area to obtain the spatial depth parameter; according to the structural hidden danger extension trend in the surface size parameters, the pixel texture feature of the extended pixel area of ​​the structural hidden danger pixel area is identified, and the extended texture parameter of the extended pixel area is determined; based on the surface size parameters, the spatial depth parameters and the extended texture parameters, the structural hidden danger type is determined; and the surface size parameters, the spatial depth parameters, the extended texture parameters and the structural hidden danger type are determined as the structural surface hidden danger characteristics; according to the structural surface hidden danger characteristics, the initial risk hidden danger image is optimized to obtain the structural surface hidden danger image.

[0016] The embodiment of the present application performs grayscale uniformization processing by using a preset grayscale level and a grayscale uniform distribution probability value, which can reduce noise in the image and improve the accuracy of subsequent processing. The pixel features of the structural hidden danger area are extracted by filtering the image pixel values, which is helpful to identify and locate potential risk points. Based on the weight change coefficient, the pixel features are extracted in detail, which is helpful to capture more detailed structural information. Combining grayscale features, color features and texture features can more comprehensively describe the risk hidden dangers and improve the accuracy of recognition. By using the segmentation processing of complementary overlapping image areas, the grayscale risk hidden danger image can be divided into different pixel areas, which is helpful for further analysis and processing. The edge pixels are located by the attention mechanism, and the surface size parameters are calculated to provide a basis for the subsequent judgment of the type of structural hidden dangers. The spatial depth calculation based on the heat source characteristics is helpful to identify the three-dimensional spatial distribution of the hidden danger area. The pixel texture feature recognition of the extended pixel area is helpful to determine the extension trend of the hidden danger. The type of structural hidden danger can be determined by combining the surface size parameters, spatial depth parameters and extended texture parameters. The initial risk hidden danger image is optimized according to the structural surface hidden danger characteristics to improve the image quality and recognition effect.

[0017] In a feasible implementation, the GIS information of the structural surface hidden danger image is used to perform spatiotemporal fusion processing of multi-source external hidden danger factors on the corresponding dam section, and based on the decentralized influence of the fused data on the dam section, the external environmental hidden danger characteristics corresponding to the GIS information are determined, specifically including: obtaining the GIS information of the structural surface hidden danger image; based on the GIS information, collecting the multi-source external hidden danger factors; wherein the multi-source external hidden danger factors at least include: current flowing water tide factors, water level factors, current dam section maintenance factors, weather forecast factors, current dam section structural strength and flowing water area geostrophic force factors; through three Weibull operation is performed to crop the multi-source external hidden danger factors and the GIS information into the same space to obtain the actual external hidden danger spatial factors; the multi-source external hidden danger factors are processed with a unified time reference under multi-source data to obtain the external hidden danger factors at the current time; the actual external hidden danger spatial factors are spatially and temporally fused with the external hidden danger factors at the current time to obtain the fused data; according to the expert experience score of the external hidden danger factors, each attribute data in the fused data is divided into a decentralized influence degree to determine the weight coefficient of each attribute data; the fused data and the corresponding weight coefficients are integrated to obtain the external environment hidden danger characteristics.

[0018] The embodiment of the present application can more comprehensively evaluate the risk status of the dam by performing spatiotemporal fusion processing on the GIS information of the structural surface hidden danger image of the dam section and the multi-source external hidden danger factors. The external hidden danger factors from different sources, such as tides, water levels, maintenance conditions, weather forecasts, structural strength and Coriolis force, are integrated to improve the comprehensiveness of data and comprehensiveness of analysis. The multi-source data is spatially clipped by three-dimensional Boolean operations to ensure that the analyzed data is consistent with the actual geographical location of the dam section, thereby improving the accuracy of the data. The multi-source external hidden danger factors are processed with a unified time base to ensure the consistency and comparability of the analysis, which helps to more accurately reflect the current risk status. The data obtained by spatiotemporal fusion can more accurately reflect the risk status of the dam section at different times and spaces. The data attributes are scored using expert experience, and the weight coefficient is determined based on the score, which helps to improve the reliability and accuracy of risk assessment. The weight coefficient of each attribute data is determined by dividing the degree of decentralized influence, making the risk assessment more reasonable and scientific. The fused data is integrated with the weight coefficient to obtain the characteristics of external environmental hidden dangers, which provides an important decision-making basis for the maintenance and management of the dam. Overall, the prediction and response capabilities of dam risk management are enhanced, helping to prevent and mitigate potential risks.

[0019] In a feasible implementation manner, before performing grayscale correlation calculation based on subjective and objective factors on the structural surface hidden danger features in the structural surface hidden danger image and the external environment hidden danger features to obtain the hidden danger growth correlation between the structural surface hidden danger features and the external environment hidden danger features, the method further includes: performing data standardization processing on the structural surface hidden danger features and the external environment hidden danger features, and eliminating the data dimensions and scales of different indicators between the structural surface hidden danger features and the external environment hidden danger features; performing normalization processing on the internal indicators in the structural surface hidden danger features and the external indicators in the external environment hidden danger features under the relevant network time slot to obtain Normalized internal indicator data and external indicator data; according to the judgment matrix constructed under the subjective factors, the correlation index between the internal indicator and the external indicator is calculated based on the subjective factors to obtain the subjective correlation index; and based on the subjective correlation index, the total hierarchical weight of the total indicator is determined; wherein the subjective factors are expert evaluation indicator factors; based on the total hierarchical weight, the bottom-level indicator weights are calculated for both the internal indicator data and the external indicator data to obtain the subjective grayscale correlation based on the subjective factors; wherein the subjective grayscale correlation is the degree of mutual influence between the internal indicator and the external indicator under the subjective factors.

[0020] In a feasible implementation manner, after predicting the damage growth of the structural surface hidden danger area of ​​the dam through the damage accumulation criterion in the structural surface loss prediction model and obtaining the structural damage loss intensity, the information planning and processing of the maintenance route and maintenance time of the structural surface hidden danger area is carried out based on the risk level of the structural damage loss intensity, and a maintenance inspection plan for the dam is generated, which specifically includes: arranging the structural damage loss intensity of each structural surface hidden danger area in time series to generate a structural damage loss intensity list; dividing the risk level of each structural damage loss intensity in the structural damage loss intensity list according to a preset loss intensity warning threshold, and determining the intensity risk level of each structural damage loss intensity; wherein the intensity risk level includes: ultra-high risk level, high risk level, Medium risk level, low risk level, low risk level and very low risk level; based on the difficulty of maintenance under the intensity risk level, determine the ideal maintenance time for each intensity risk level; wherein the ideal maintenance time is the ideal time for repair and maintenance of each structural surface hidden danger area; based on the order of the intensity risk levels, generate an initial maintenance route; wherein, the higher the intensity risk level, the higher the planning priority of the corresponding initial maintenance route; according to the ideal maintenance time, prioritize the structural surface hidden danger areas in the same intensity risk level to obtain a priority sequence of the same level; through the priority sequence of the same level, adjust the order of the line nodes of the initial maintenance route to obtain a final maintenance route; and integrate the data of the final maintenance route and the ideal maintenance time to generate the maintenance inspection plan.

[0021] The embodiment of the present application can clearly identify and classify the risk level of the surface hidden dangers of the dam through time sequence arrangement and risk level division, which is helpful for the rational allocation of resources. The preset loss intensity warning threshold can help to timely identify the structural hidden dangers that are about to reach the dangerous level, so as to take preventive measures in advance. Providing risk levels from ultra-high risk to extremely low risk can meet the management needs of hidden dangers of different risk levels. It can also determine the ideal maintenance time according to different risk levels, which is helpful to formulate a reasonable maintenance plan to avoid excessive maintenance or insufficient maintenance. Determine the ideal maintenance time based on the difficulty of maintenance, so that the maintenance plan is closer to the actual operation and improves the maintenance efficiency. At the same time, the initial maintenance line is generated according to the hierarchical order of the risk level to ensure that the maintenance work in the high-risk area is given priority. Priority division is performed within the same risk level to ensure that more urgent or more important maintenance work is given priority at the same risk level. The node order of the maintenance line is adjusted through the priority sequence of the same level, making the maintenance work more efficient and orderly. Integrate the final maintenance line and the ideal maintenance time to generate a maintenance inspection plan to provide detailed guidance for the long-term maintenance of the dam. It can also optimize the allocation and use of maintenance resources and improve the efficiency of maintenance work, which helps to enhance the risk management level of dams and ensure the safety and durability of dams.

[0022] On the other hand, an embodiment of the present application also provides a structural surface loss prediction system based on a dam project, including: an image acquisition module, which is used to perform core node deployment scanning at multiple points on the risk area in the dam according to a preset long shoreline dam detection route and through a number of drones to obtain a comprehensive structural surface image; wherein the comprehensive structural surface image includes infrared thermal imaging information, contour display information and real structural surface information; a structural analysis module, which is used to pre-screen the risk hidden danger characteristics of the external thermal imaging information, the contour display information and the real structural surface information in the comprehensive structural surface image to obtain an initial risk hidden danger image; and the initial risk hidden danger image is subjected to structural surface hidden danger feature identification and extraction processing related to the pixel matrix of the initial risk hidden danger image to determine the structural surface hidden danger image; a large database module, which is used to perform spatiotemporal fusion processing of multi-source external hidden danger factors on the corresponding dam section through the GIS information of the structural surface hidden danger image, and determine the hidden danger related to the dam section based on the degree of decentralized influence of the fused data on the dam section. the external environment hidden danger characteristics corresponding to the GIS information; a correlation calculation module, used to perform grayscale correlation calculation on the structural surface hidden danger characteristics in the structural surface hidden danger image and the external environment hidden danger characteristics based on subjective and objective factors, and obtain the hidden danger growth correlation between the structural surface hidden danger characteristics and the external environment hidden danger characteristics; wherein the structural surface hidden danger characteristics are subjective factor characteristics, and the external environment hidden danger characteristics are objective factor characteristics; a model prediction module, used to input the hidden danger growth correlation, the structural surface hidden danger characteristics and the external environment hidden danger characteristics into the pre-trained structural surface loss prediction model; and through the damage accumulation criterion in the structural surface loss prediction model, the damage growth prediction is performed on the structural surface hidden danger area of ​​the dam to obtain the structural damage loss intensity of each structural surface hidden danger area; an inspection planning module, used to plan and process the information on the maintenance route and maintenance time of the structural surface hidden danger area based on the risk level of the structural damage loss intensity, and generate a maintenance inspection plan for the dam.

[0023] The present application provides a method and system for predicting the structural surface loss state based on dam engineering. Compared with the prior art, the embodiments of the present application have the following beneficial technical effects: 1. Multi-source data fusion: By combining infrared thermal imaging, contour display and real structural surface information, the state of the dam structure surface can be more comprehensively obtained, thereby improving the accuracy of the prediction.

[0024] 2. Pre-screening of risks and hidden dangers: Pre-screening of risk and hidden danger characteristics on the acquired comprehensive structural surface images can quickly identify potential problem areas and improve detection efficiency.

[0025] 3. Identification of hidden dangers on the surface of structures: Through the analysis of the pixel matrix, the characteristics of hidden dangers on the surface of structures can be accurately identified and extracted, providing a reliable basis for loss prediction.

[0026] 4. Space-time fusion processing: The space-time fusion processing of multi-source external hidden danger factors in the dam section can more comprehensively consider the impact of the external environment on the dam structure.

[0027] 5. Grayscale correlation analysis: Through the grayscale correlation calculation of subjective and objective factors, the correlation between the hidden danger characteristics of the surface of the structure and the hidden danger characteristics of the external environment can be evaluated, which helps to more accurately predict the growth trend of hidden dangers.

[0028] 6. Loss prediction model: Using the pre-trained structural surface loss prediction model, the damage accumulation of the potential danger area on the structural surface of the dam can be predicted, thereby predicting the intensity of structural damage loss.

[0029] 7. Maintenance route and time planning: Based on the risk level of structural damage and loss intensity, relevant maintenance routes and maintenance time can be planned, which helps to prevent and maintain the safety of the dam in advance.

[0030] 8. Improve detection efficiency: Drone scanning and multi-point deployment can cover a wider area and improve the efficiency and quality of detection.

[0031] 9. Reduce maintenance costs: Through accurate risk identification and loss prediction, maintenance work can be reasonably planned and unnecessary maintenance costs can be reduced.

[0032] 10. Enhance dam safety: By timely identifying and predicting potential structural problems, the safety of dams can be enhanced and the risk of disasters can be reduced. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings: Figure 1 A flow chart of a method for predicting the structural surface loss state based on a dam project provided in an embodiment of the present application; Figure 2 A schematic diagram of a regional detection route for a drone provided in an embodiment of the present application; Figure 3 A schematic diagram of a potential hazard area on a surface of a structure provided in an embodiment of the present application; Figure 4A schematic diagram of a structural surface hidden danger image provided in an embodiment of the present application; Figure 5 A schematic diagram of the structure of a structural surface loss prediction system based on a dam project provided in an embodiment of the present application. DETAILED DESCRIPTION

[0034] In order to enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.

[0035] It should be noted that the structural surface hidden danger areas mainly include: transverse crack areas, longitudinal crack areas, oblique crack areas, hole areas, settlement extension zone areas, etc. The structural surface hidden danger image is the image data of the structural surface hidden danger area.

[0036] The present application embodiment provides a method for predicting the structural surface loss state based on a dam project, such as Figure 1 As shown, the method for predicting the structural surface loss state based on the dam project specifically includes steps S101-S106: S101. According to the preset long shoreline dam inspection route, a number of drones are used to perform core node deployment scanning at multiple points on the risk area of ​​the dam to obtain a comprehensive structural surface image, which includes infrared thermal imaging information, contour display information, and real structural surface information.

[0037] Specifically, it is necessary to first identify the location of the structural hidden danger area based on the three-dimensional modeling diagram of the dam section to be detected, wherein the location of the structural hidden danger area at least includes: the location of the dam corner area, the location of the expansion joint area, and the location of the dam connection.

[0038] Furthermore, according to the location of the structural hazard area, the plane top view in the three-dimensional modeling of the dam section is rasterized to generate several search grids. Based on the location of the structural hazard area, the corresponding search grid is marked and deployed with core nodes to determine the core grid area. It is also necessary to determine the core grid area as the center point and use twice the diagonal length of the search grid as the radius to divide the risk area.

[0039] In one embodiment, Figure 2 A schematic diagram of a regional detection route of a drone provided in an embodiment of the present application, such as Figure 2As shown in the figure, by utilizing core key technologies such as fully autonomous UAV cabin technology, UAV mission planning technology for dam monitoring, real-time high-speed transmission technology of UAV data and levee data, UAV data processing and levee disease information extraction, the search grid and core grid area can be obtained based on the analysis. At the same time, the general structural hazard areas mainly include levee corners, expansion joints, connection points, turning points of flowing water areas, and areas with steep slopes, and then these areas are divided into risk areas.

[0040] Furthermore, according to the search grid center point occupied by the risk area, and through deep optimization of the search direction and the cattle-ploughing method with different starting points, the full coverage path planning processing of the current cluster of the starting point is carried out on the risk area to obtain the risk area detection route. Among them, the risk area detection route is the flight route of the drone.

[0041] Furthermore, by using the starting and ending points of the risk area detection route and the ox-ploughing method, the search grid in the plane top view is fully covered by path planning and processing to obtain the long shoreline dam detection route. The long shoreline dam detection route includes: general area detection route and risk area detection route.

[0042] Furthermore, according to the long shoreline dam inspection route, the comprehensive structure surface image is finally collected through various collection devices mounted on the drone. Among them, the various collection devices include: ultra-high-definition camera equipment, infrared thermal imager and total station. That is, the mounted ultra-high-definition camera equipment is used to collect real structural surface information, the infrared thermal imager is used to collect external thermal imaging information, and the total station is used to collect contour display information.

[0043] In one embodiment, Figure 2 As shown in the figure, based on the ox-ploughing method and deep optimization of the search direction, the full coverage path planning processing of the current cluster of the starting point is first carried out for the risk area, that is, the path planning of the risk area is carried out first, and the full coverage path planning of the detection area that needs to be focused on is realized, that is, the risk area detection route. Then the remaining search grids are generalized for path planning to form a general area detection route, which can greatly reduce the data collection time, and at the same time, the main cost and energy are all used for the accurate detection of the risk area, and then the various mounted sensor devices are used to finally collect an image set, that is, a comprehensive structure surface image set.

[0044] S102, pre-screening the risk hidden danger features of the infrared thermal imaging information, contour display information and real structural surface information in the comprehensive structural surface image to obtain an initial risk hidden danger image; and identifying and extracting the structural surface hidden danger features of the pixel matrix of the initial risk hidden danger image to determine the structural surface hidden danger image.

[0045] Specifically, the infrared thermal imaging information of the surface of the comprehensive dam structure is first processed to generate an infrared heat source distribution map. Then, based on the current weather data information, the initial heat source anomaly area in the infrared heat source distribution map is eliminated, and the remaining heat source anomaly area is matched with the temperature difference gradient map of the water level in the current flowing water area by the fluctuation range threshold of the dynamic temperature difference line, and the final heat source anomaly area based on the comprehensive structure surface image and the corresponding heat source anomaly coordinate position are determined.

[0046] Furthermore, based on the 3D modeling diagram of the dam section to be detected, the contour line display information in the comprehensive structure surface image is converted into a 3D matrix to obtain a 3D contour line model diagram. Then, according to the GIS terrain feature information of the current dam section, the contour line abnormal area in the 3D contour line model diagram is identified, and the contour line abnormal area based on the comprehensive structure surface image and the corresponding contour line abnormal coordinate position are determined.

[0047] As a feasible implementation method, the infrared thermal imaging information is first used to generate the corresponding infrared heat source distribution map, and then based on the weather temperature and climate conditions during the acquisition period, the abnormal signals are first removed, that is, error filtering is performed, and then the remaining heat source abnormal area is matched with the temperature difference gradient map of the water level in the current flowing water area for the fluctuation range threshold of the dynamic temperature difference line, that is, the temperature difference gradient map of the water level in the current flowing water area is used as the temperature fluctuation range threshold and the temperature difference reference, and then the heat source abnormal area at abnormal temperature is screened out, that is, the final heat source abnormal area and the corresponding heat source abnormal coordinate position. At the same time, the GIS terrain feature information of the current dam section is analyzed for the contour abnormal area using the three-dimensional contour model map, so as to determine the contour abnormal area and the corresponding contour abnormal coordinate position.

[0048] Furthermore, the heat source anomaly coordinate position in the comprehensive structural surface image and the contour line anomaly coordinate position are matched within the relevant coordinate region threshold to determine the contour-heat source anomaly region. The contour line anomaly region corresponding to the contour-heat source anomaly region and the final heat source anomaly region are scored for risk hazard characteristics to obtain the contour-heat source anomaly score.

[0049] Furthermore, if the contour-heat source anomaly score is greater than the first preset threshold, the real structural surface information corresponding to the contour-heat source anomaly area is extracted. Then, the real structural surface information is processed by information matching query of the semantics and interest pixel area through the preset dam surface structural anomaly template to obtain the structural risk hidden danger features contained in the real structural surface information. Among them, the structural risk hidden danger features include: transverse cracks, longitudinal cracks, oblique cracks, holes and settlement extension zones. Finally, the contour-heat source anomaly area image with structural risk hidden danger features is determined as the initial risk hidden danger image.

[0050] In one embodiment, Figure 3 A schematic diagram of a structural surface hidden danger area provided in an embodiment of the present application, such as Figure 3 As shown in the figure, when the same area satisfies both the heat source anomaly and the contour anomaly, the area can be scored to obtain the contour-heat source anomaly score. When the contour-heat source anomaly score is greater than the first preset threshold, the real structural surface information corresponding to the contour-heat source anomaly area is extracted. This can greatly reduce the amount of recognition calculation for the structural hidden danger area corresponding to the real structural surface information, thereby finding the structural surface hidden danger area more quickly. Then, combined with the dam surface structural anomaly template, the real structural surface information is processed for information matching and query processing on the semantics and interest pixel area, and finally the structural risk hidden danger characteristics are parsed.

[0051] Furthermore, the grayscale uniformity processing is performed on the initial risk hidden danger image using the preset grayscale level and grayscale uniform distribution probability value to obtain the grayscale risk hidden danger image. And based on the image pixel value after filtering the grayscale risk hidden danger image, the detailed feature extraction of the weight change coefficient of the pixel features in the structural hidden danger field is performed to obtain the structural hidden danger detail feature.

[0052] Furthermore, according to the grayscale features, color features and texture features in the structural hidden danger detail features, and based on the preset segmentation edges and the corresponding two-dimensional structural hidden danger feature parameters, the grayscale risk hidden danger image is segmented into complementary overlapping image areas to obtain the structural hidden danger pixel area.

[0053] Furthermore, it is also necessary to use the attention mechanism to perform positioning processing on the edge pixels of the structural hidden danger pixel area under the relevant edge regularization items, and output the edge pixel distribution position. Then, according to the edge pixel distribution position, the surface size of the structural hidden danger pixel area is calculated and processed to obtain the surface size parameters of the structural hidden danger pixel area. Among them, the surface size parameters include: structural hidden danger length, structural hidden danger width, structural hidden danger area and structural hidden danger extension trend.

[0054] Furthermore, the color change calculation of the pixel matrix is ​​performed on the central pixel area surrounded by the edge pixel distribution position, and based on the heat source characteristics of the final heat source abnormal area in the initial risk hidden danger image, the spatial depth calculation is performed on the structural hidden danger pixel area to obtain the spatial depth parameter. Then, according to the structural hidden danger extension trend in the surface size parameter, the pixel texture feature of the extended pixel area of ​​the structural hidden danger pixel area is identified, and the extended texture parameter of the extended pixel area is determined.

[0055] Furthermore, based on the surface size parameters, spatial depth parameters and extended texture parameters, the type of structural hidden danger is determined. The surface size parameters, spatial depth parameters, extended texture parameters and the type of structural hidden danger are determined as the structural surface hidden danger characteristics. Finally, according to the structural surface hidden danger characteristics, the initial risk hidden danger image is optimized to obtain the structural surface hidden danger image.

[0056] As a feasible implementation method, Figure 4 A schematic diagram of a structural surface hidden danger image provided in an embodiment of the present application, such as Figure 4 As shown in the figure, on the basis of identifying the crack features on the surface of the dam, the image is segmented into multiple complementary overlapping areas according to the grayscale, color, texture and other features of the grayscale processing and filtering processing image, so that the cracks in the image are clearer. After segmenting the grayscale risk hidden danger image, the edge regularization term of the structural hidden danger pixel is located to address the problem that the edge of the structural hidden danger is not significant. From the perspective of machine vision, the attention mechanism is captured to the pixel area of ​​visual interest, and the surrounding irrelevant pixel area is suppressed. The structural hidden danger pixel area is obtained by outputting the index and function features, that is, the relevant features are associated in each dimension of the image through the attention channel to enhance the interdependence between the images of the structural hidden danger pixel area.

[0057] In one embodiment, by calculating the surface size of the structural hidden danger pixel area, calculating the spatial depth of the structural hidden danger pixel area, and identifying the pixel texture features of the extended pixel area of ​​the structural hidden danger pixel area, the type of structural hidden danger can be determined, such as a transverse crack area, a longitudinal crack area, an oblique crack area, a hole area, and a settlement extension zone area, etc., and then the surface size parameters, spatial depth parameters, extended texture parameters and structural hidden danger types are determined as structural surface hidden danger characteristics, and then these characteristics are finally optimized by data labeling and identification results, so that a structural surface hidden danger image containing the above-mentioned characteristics can be obtained, that is, the main core feature image of the structural surface hidden danger that can be extracted from the structural surface hidden danger image.

[0058] S103. Through the GIS information of the structural surface hidden danger image, the corresponding dam section is subjected to spatiotemporal fusion processing of multi-source external hidden danger factors, and based on the decentralized influence degree of the fused data on the dam section, the external environmental hidden danger characteristics corresponding to the GIS information are determined.

[0059] Specifically, the GIS information of the structural surface hidden danger image is first obtained. Then, based on the GIS information, multi-source external hidden danger factors are collected. Among them, the multi-source external hidden danger factors at least include: current flowing water tide factors, water level factors, current dam section maintenance factors, weather forecast factors, current dam section structural strength and flowing water Coriolis force factors.

[0060] Furthermore, through three-dimensional Boolean operations, the multi-source external hidden danger factors and GIS information are clipped in the same space to obtain the actual external hidden danger spatial factors. The multi-source external hidden danger factors are processed with a unified time base under multi-source data to obtain the current time external hidden danger factors. The actual external hidden danger spatial factors and the current time external hidden danger factors are spatiotemporally fused to obtain fused data.

[0061] Furthermore, it is necessary to divide the degree of influence of each attribute data in the fused data according to the expert experience score of the external hidden danger factors, and determine the weight coefficient of each attribute data. Finally, the fused data and the corresponding weight coefficient are integrated to obtain the characteristics of the external environmental hidden dangers.

[0062] It should be noted that a unified spatial reference system is the basis for data integration and fusion. The reference systems of data of different types or sources are often inconsistent, which will result in different types of models being unable to be displayed and applied in the same scene. This causes data accuracy loss during the fusion process. Based on the full-space three-dimensional model data standard, this paper caches and slices the model data in a unified manner without changing the actual coordinates of the original data. In the process of slicing, the data fusion correction is completed. At the same time, three-dimensional Boolean operations are used to cut different elements overlapping in the same spatial position, so that the characteristics of each element are more in line with the actual situation, and then the fused data is obtained.

[0063] In one embodiment, during the actual growth of the hidden dangers of the dam structure, it will be mainly affected by a large number of external factors, thereby accelerating the erosion of the structure surface, etc. Therefore, it is necessary to combine the GIS information of the structure surface hidden danger image, that is, the actual environmental conditions of the flowing water area where each structure surface hidden danger image is located, and combine the specific impact results of the spatiotemporal variables to determine the actual external hidden danger spatial factors and the current time external hidden danger factors, and then perform spatiotemporal fusion after a unified time reference to obtain fused data, and then under the expert experience scoring mechanism of the external hidden danger factors, divide the attribute data in the fused data according to the degree of decentralized influence, and determine the weight coefficient of each attribute data. Finally, the fused data and the corresponding weight coefficient are integrated to obtain the importance and influence of each item of data, that is, the influence degree on the subsequent loss intensity of the structure surface hidden danger area and the degree of damage and aging.

[0064] S104, performing grayscale correlation calculation on the structural surface hidden danger features and the external environment hidden danger features in the structural surface hidden danger image based on subjective and objective factors, and obtaining the hidden danger growth correlation degree between the structural surface hidden danger features and the external environment hidden danger features. Among them, the structural surface hidden danger features are subjective factor features, and the external environment hidden danger features are objective factor features.

[0065] It should be noted that the grey correlation analysis method is an effective tool for dealing with uncertain information and small sample data problems, and is particularly applicable in complex data model predictions, especially in scenarios involving multiple factors affecting time slot allocation. The grey correlation analysis method provides an effective method for evaluating and ranking the needs of nodes with different degrees of influence, and aggregates the needs of each node into a comprehensive score.

[0066] Specifically, firstly, the data of the structural surface hidden danger characteristics and the external environment hidden danger characteristics are standardized, and the data dimensions and scales of different indicators between the structural surface hidden danger characteristics and the external environment hidden danger characteristics are eliminated. Then, the internal indicators in the structural surface hidden danger characteristics and the external indicators in the external environment hidden danger characteristics are normalized under the relevant network time slot to obtain the normalized internal indicator data and external indicator data.

[0067] Furthermore, according to the judgment matrix constructed under subjective factors, the correlation index between internal and external indicators is calculated based on subjective factors to obtain subjective correlation index. Based on the subjective correlation index, the total hierarchical weight of the total indicator is determined. Among them, the subjective factor is the expert evaluation indicator factor.

[0068] As a feasible implementation method, since the judgment matrix is ​​constructed through subjective factor evaluation, there may be a certain degree of subjective bias. Therefore, it is necessary to calculate the internal and external indicators to determine whether the consistency is within an acceptable range. If the consistency is less than 0.1, the judgment matrix is ​​considered to have good consistency and the obtained weight vector can be used in the decision-making process. Otherwise, the judgment matrix needs to be re-evaluated and readjusted.

[0069] Furthermore, based on the total hierarchical weight, the underlying indicator weights are calculated for both the internal indicator data and the external indicator data to obtain the subjective grayscale correlation based on subjective factors. The subjective grayscale correlation is the degree of mutual influence between the internal and external indicators under subjective factors.

[0070] Furthermore, the normalized internal indicator data and external indicator data are extracted, and then the relative importance of the internal indicator data and external indicator data under each node is calculated through the preset entropy weight method to obtain the objective indicator correlation.

[0071] In one embodiment, based on the total hierarchical weight, both the internal indicator data and the external indicator data are subjected to the underlying indicator weight calculation, that is, it is necessary to first use the internal indicator data as the data criterion basis, and both the internal indicator data and the external indicator data are subjected to the underlying indicator weight calculation, so as to obtain the subjective grayscale correlation based on the underlying indicator weight, that is, the subjective grayscale correlation is the degree of influence of the existence of the external indicator on the internal indicator. Then, based on the entropy weight method, the relative importance of the associated nodes between each internal indicator and the external indicator is calculated, that is, the relative importance of the existence of the internal indicator to the external indicator, so as to calculate the objective indicator correlation in this case.

[0072] Furthermore, according to the objective indicator correlation, the information entropy is calculated for both the internal indicator data and the external indicator data to obtain the information utility value of each indicator data. The information utility value is normalized to obtain the entropy weight method weight based on objective factors. Among them, the objective factors are the correlation influencing factors between various indicators based on the large database. Then, based on the entropy weight method weight, the criterion indicator weight is calculated for both the internal indicator data and the external indicator data to determine the objective grayscale correlation based on objective factors.

[0073] Furthermore, the subjective grayscale correlation and the objective grayscale correlation are combined to obtain the comprehensive grayscale correlation. Finally, through the comprehensive grayscale correlation, each association node between the structural surface hidden danger characteristics and the external environment hidden danger characteristics is adaptively adjusted within the grayscale correlation range to determine the hidden danger growth correlation. Among them, the hidden danger growth correlation is the degree of influence of the external environment hidden danger characteristics on the structural surface hidden danger characteristics.

[0074] As a feasible implementation method, the objective grayscale correlation degree based on objective factors is determined by first calculating the criterion index weights for both the internal index data and the external index data. Then, the subjective grayscale correlation degree and the objective grayscale correlation degree are combined, that is, the subjective grayscale correlation degree and the objective grayscale correlation degree are used as the reference standards of the correlation degree, and each associated node between the hidden danger characteristics of the surface layer of the structure and the hidden danger characteristics of the external environment is adaptively adjusted within the grayscale correlation range, so that each associated node meets the subjective grayscale correlation degree and the objective grayscale correlation degree, that is, the adaptive adjustment within the grayscale correlation range, and finally the hidden danger growth correlation degree based on subjective and objective factors can be determined.

[0075] The embodiment of the present application can provide a comprehensive assessment of the hidden dangers of the surface layer of the structure and the hidden dangers of the external environment by combining subjective and objective factors to perform gray correlation calculation, so as to more comprehensively understand the complexity and interaction of the hidden dangers. Extracting and normalizing the internal and external indicator data ensures the comparability of data of different dimensions in the calculation process and improves the accuracy of the analysis. The entropy weight method is used to calculate the correlation of objective indicators. This method can objectively reflect the relative importance between indicators and reduce subjective bias. The data utility value of each indicator is calculated by information entropy, which further refines the importance of the indicator and helps to make more accurate choices in the decision-making process. Based on the information utility value, normalization is performed to obtain the entropy weight method weights, which can reflect the contribution of the indicator to the overall evaluation. The weight of the criterion indicator is calculated using the entropy weight method weight, and the objective gray correlation is determined, which enhances the objectivity of the evaluation. By combining the subjective gray correlation and the objective gray correlation, a comprehensive gray correlation is obtained. This method can balance subjective and objective factors and improve the comprehensiveness of the evaluation. By adaptively adjusting the associated nodes based on the comprehensive grayscale correlation, the growth correlation of hidden dangers can be determined more accurately, that is, the degree of influence of external environmental hidden dangers on structural surface hidden dangers. It can help decision makers identify and assess risks more accurately, thereby improving the quality of risk management decisions. It can also enhance the understanding of the growth correlation of hidden dangers and improve the ability to predict the development trend of potential risks.

[0076] S105, input the hidden danger growth correlation, structural surface hidden danger characteristics and external environment hidden danger characteristics into the pre-trained structural surface loss prediction model. And through the damage accumulation criterion in the structural surface loss prediction model, the damage growth of the structural surface hidden danger area of ​​the dam is predicted to obtain the structural damage loss intensity of each structural surface hidden danger area.

[0077] Specifically, the collected historical structural surface hidden danger feature data set and the corresponding external environment hidden danger feature data set are first used. Then, based on the preset SN curve, the historical hidden danger growth correlation set, the historical structural surface hidden danger feature data set and the external environment hidden danger feature data set, the prediction model is iteratively optimized and trained to generate a structural surface loss prediction model. Among them, the historical hidden danger growth correlation set is the output end, and the SN curve, the historical structural surface hidden danger feature data set and the external environment hidden danger feature data set are the input end. The prediction model is a neural network model that can realize data fitting and data residual processing; the SN curve is a damage accumulation estimation curve based on the number of water flow impacts N and the dam stress S.

[0078] Furthermore, it is necessary to input the hidden danger growth correlation within the current time node, the structural surface hidden danger characteristics and the external environment hidden danger characteristics into the pre-trained structural surface loss prediction model.

[0079] As a feasible implementation method, the prediction model is trained by fitting the hyperfunction relationship between input and output, wherein the input layer Q includes: SN curve, historical hidden danger growth correlation set, historical structural surface hidden danger feature data set and external environment hidden danger feature data set, hidden layer (multi-layer neurons can be connected), and the output layer CNN includes: historical hidden danger growth correlation set. At the same time, the input layer, convolution calculation layer, relu excitation layer, pooling layer, and fully connected layer network in the convolutional neural network can also be used to update the core algorithm. During the training process, the weights are initialized first, and the weights of the network are initialized to very small random numbers. Each sample is forward propagated input and backward propagated error.

[0080] Further, according to , predict and obtain the structural damage loss intensity based on the structural surface loss prediction model .in, is the SN curve; It is the hidden danger feature of the structure surface; It is the hidden danger characteristics of the external environment; To increase the relevance of hidden dangers; i is a mathematical constant; is the theoretical structural loss strength; Mathematical symbols; is the time variable; It is the input variable of the structural surface loss prediction model; structural damage loss intensity It is used to represent the predicted service life time during which the potential danger area on the surface of the structure can continuously exceed the minimum structural safety strength threshold.

[0081] S106. Based on the risk level of the structural damage and loss intensity, information planning and processing on the maintenance routes and maintenance time of the potential danger areas on the surface of the structure is carried out to generate a maintenance inspection plan for the dam.

[0082] Specifically, the structural damage loss intensity of each structural surface hidden danger area is first arranged in time series to generate a structural damage loss intensity list. And according to the preset loss intensity warning threshold, each structural damage loss intensity in the structural damage loss intensity list is divided into risk levels to determine the intensity risk level of each structural damage loss intensity. Among them, the intensity risk level includes: ultra-high risk level, high risk level, medium risk level, low risk level, low risk level and extremely low risk level.

[0083] Furthermore, based on the maintenance difficulty under the intensity risk level, the ideal maintenance time for each intensity risk level is determined. The ideal maintenance time is the ideal time for repairing and maintaining each surface potential hazard area of ​​the structure. Then, based on the order of the intensity risk levels, the initial maintenance route is generated. The higher the intensity risk level, the higher the planning priority of the corresponding initial maintenance route. The ideal maintenance time is then used to prioritize the surface potential hazard areas of the structure at the same intensity risk level to obtain a priority sequence of the same level.

[0084] Furthermore, the initial maintenance line is processed by adjusting the order of line nodes through the same-level priority sequence to obtain the final maintenance line. The final maintenance line is integrated with the ideal maintenance time to generate a maintenance inspection plan.

[0085] In one embodiment, after calculating the structural damage loss intensity of each structural surface hidden danger area, firstly, based on the structural damage loss intensity list, each structural surface hidden danger area is sorted by the degree of danger, that is, the area with low current strength loss and major structural surface hidden danger is first defined as an ultra-high risk level, and then needs to be processed first, and then recursively. Then, the structural surface hidden danger areas at the same level are divided into maintenance time, that is, the repair and maintenance time corresponding to the structural surface hidden danger areas of different types or positions is different, and it is necessary to first perform maintenance and repair on the area with longer time, so as to prevent delaying too long in other areas and further aggravating the dangerous situation of the current hidden danger area, so as to use the priority sequence of the same level to adjust the order of the line nodes of the initial maintenance line to obtain the final maintenance line, and integrate the data of the final maintenance line with the ideal maintenance time to generate a maintenance inspection plan, and the inspection personnel can process the structural surface hidden danger areas one by one according to the maintenance inspection plan, and return the processing results to the management system to form a closed-loop processing of the entire maintenance and repair.

[0086] In addition, the present application embodiment also provides a structural surface loss prediction system based on dam engineering, such as Figure 5 As shown, the structural surface loss state prediction system 500 based on the dam project specifically includes: The image acquisition module 510 is used to perform core node deployment scanning at multiple points on the risk area in the dam according to the preset long shoreline dam inspection route and through a number of drones to obtain a comprehensive structural surface image; wherein the comprehensive structural surface image includes infrared thermal imaging information, contour display information and real structural surface information.

[0087] The structural analysis module 520 is used to pre-screen the risk hidden danger characteristics of the infrared thermal imaging information, contour display information and real structural surface information in the comprehensive structural surface image to obtain an initial risk hidden danger image; and to identify and extract the structural surface hidden danger characteristics of the relevant pixel matrix of the initial risk hidden danger image to determine the structural surface hidden danger image.

[0088] The big database module 530 is used to perform spatiotemporal fusion processing of multi-source external hazard factors on the corresponding dam section through the GIS information of the structural surface hazard image, and determine the external environmental hazard characteristics corresponding to the GIS information based on the decentralized influence of the fused data on the dam section.

[0089] The correlation calculation module 540 is used to perform grayscale correlation calculation on the structural surface hidden danger features in the structural surface hidden danger image and the external environment hidden danger features based on subjective and objective factors, so as to obtain the hidden danger growth correlation between the structural surface hidden danger features and the external environment hidden danger features; wherein the structural surface hidden danger features are subjective factor features, and the external environment hidden danger features are objective factor features.

[0090] The model prediction module 550 is used to input the hidden danger growth correlation, structural surface hidden danger characteristics and external environment hidden danger characteristics into the pre-trained structural surface loss prediction model; and through the damage accumulation criterion in the structural surface loss prediction model, the damage growth of the structural surface hidden danger area of ​​the dam is predicted to obtain the structural damage loss intensity of each structural surface hidden danger area.

[0091] The inspection planning module 560 is used to plan and process the information on the maintenance routes and maintenance time for the potential danger areas on the surface of the structure based on the risk level of the structural damage and loss intensity, and generate a maintenance inspection plan for the dam.

[0092] The embodiment of the present application can obtain the state of the surface of the dam structure more comprehensively by combining infrared thermal imaging, contour display and real structural surface information, thereby improving the accuracy of prediction. The acquired comprehensive structural surface image can be pre-screened for risk hidden danger features, and potential problem areas can be quickly identified to improve detection efficiency. The hidden danger features of the surface structure can be accurately identified and extracted, providing a reliable basis for loss prediction. The impact of the external environment on the dam structure can also be considered more comprehensively. Through the grayscale correlation calculation of subjective and objective factors, the correlation between the hidden danger features of the surface structure and the hidden danger features of the external environment can be evaluated, which helps to more accurately predict the growth trend of hidden dangers. At the same time, using the pre-trained structural surface loss prediction model, the damage accumulation prediction of the structural surface hidden danger area of ​​the dam can be made, thereby predicting the structural damage loss intensity. And based on the risk level of the structural damage loss intensity, the relevant maintenance routes and maintenance time can be planned, which helps to prevent and maintain the safety of the dam in advance. Finally, by timely identifying and predicting potential structural problems, the safety of the dam can be enhanced and the risk of disasters can be reduced.

[0093] Each embodiment in this application is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0094] The above specific embodiments of the present application are described. In some cases, it is possible to perform in an order different from that in the embodiments and still achieve the desired result. In addition, the process depicted in the accompanying drawings does not necessarily require the specific order or continuous order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0095] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the embodiments of the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the embodiments of the present application should be included in the scope of the specification of the present application.

Claims

1. A method for predicting the structural surface loss state based on dam engineering, characterized in that: The method comprises: The structure surface hidden danger features are identified and extracted from the comprehensive structure surface image of the dam to determine the structure surface hidden danger image; Through the GIS information of the structural surface hidden danger image, the corresponding dam section is subjected to the spatiotemporal fusion processing of multi-source external hidden danger factors to determine the hidden danger characteristics of the external environment; Perform grayscale correlation calculation on the structural surface hidden danger characteristics and the external environment hidden danger characteristics to obtain the hidden danger growth correlation degree; By using the damage accumulation criterion in the structural surface loss prediction model, the damage growth of the potential danger area on the structural surface of the dam is predicted to obtain the structural damage loss intensity.

2. The method for predicting the structural surface loss state based on dam engineering according to claim 1 is characterized in that: Before performing structural surface hidden danger feature recognition and extraction processing on the comprehensive structural surface image to determine the structural surface hidden danger image, the method further includes: Identify the location of the structural hidden danger area; wherein the location of the structural hidden danger area at least includes: the location of the dam corner area, the location of the expansion joint area, and the location of the dam connection; According to the location of the structural hidden danger area, the plan view in the three-dimensional modeling of the dam section is rasterized to generate a plurality of search grids; According to the search grid center point occupied by the risk area, the risk area detection route is obtained by deeply optimizing the search direction and the cattle plowing method with different starting points; Based on the starting and ending point positions of the risk area detection route, and through the ox-plowing method, a long shoreline embankment detection route is obtained; According to the long shoreline dam detection route, the surface image of the comprehensive structure is collected by using a variety of collection devices mounted on the drone; wherein the various collection devices include: ultra-high-definition camera equipment, infrared thermal imager and total station.

3. The method for predicting the structural surface loss state based on dam engineering according to claim 1 is characterized in that: Before performing structural surface hidden danger feature recognition and extraction processing on the comprehensive structural surface image to determine the structural surface hidden danger image, the method further includes: The initial abnormal heat source area in the infrared heat source distribution map is eliminated, and the remaining abnormal heat source area is matched with the temperature difference gradient map of the water level in the current flowing water area by the fluctuation range threshold of the dynamic temperature difference line to determine the abnormal heat source coordinate position; According to the GIS terrain feature information of the current dam section, the abnormal contour line area in the three-dimensional contour line model map is processed and the abnormal contour line coordinate position is determined; Perform consistency matching processing on the abnormal coordinate position of the heat source in the comprehensive structural surface image and the abnormal coordinate position of the contour line within the threshold of the relevant coordinate area to determine the contour-heat source abnormal area; The image of the contour-heat source anomaly area with structural risk hidden danger characteristics is determined as the initial risk hidden danger image.

4. The method for predicting the structural surface loss state based on dam engineering according to claim 3 is characterized in that: The structure surface hidden danger feature recognition and extraction processing is performed on the comprehensive structure surface image to determine the structure surface hidden danger image, specifically including: According to the grayscale features, color features and texture features in the structural hidden danger detail features in the initial risk hidden danger image, the grayscale risk hidden danger image is segmented into complementary overlapping image areas to obtain structural hidden danger pixel areas; According to the edge pixel distribution positions, the surface size of the pixel area with structural hidden danger is calculated to obtain the surface size parameters of the pixel area with structural hidden danger; Performing color change calculation on the pixel matrix of the central pixel area surrounded by the edge pixel distribution position, and obtaining the spatial depth parameter based on the spatial depth calculation of the pixel area with structural hidden dangers; Determining the extended texture parameters of the extended pixel area according to the extension trend of the structural hidden dangers in the surface size parameters; The surface size parameter, the spatial depth parameter, the extended texture parameter and the structural hidden danger type are determined as structural surface hidden danger features, and the structural surface hidden danger image is optimized and generated.

5. The method for predicting the structural surface loss state based on dam engineering according to claim 1 is characterized in that: Through the GIS information of the structural surface hidden danger image, the corresponding dam section is subjected to the spatiotemporal fusion processing of multi-source external hidden danger factors to determine the hidden danger characteristics of the external environment, including: Based on GIS information, the multi-source external hidden danger factors are collected; wherein the multi-source external hidden danger factors at least include: current flowing water tide factors, water level factors, current dam section maintenance factors, weather forecast factors, current dam section structural strength and flowing water area geostrophic force factors; Clipping the multi-source external hidden danger factors and the GIS information into the same space to obtain actual external hidden danger spatial factors; Performing unified time reference processing on the multi-source external hidden danger factors under multi-source data to obtain the external hidden danger factors at the current time; and performing spatiotemporal fusion of the actual external hidden danger spatial factors and the external hidden danger factors at the current time to obtain fused data; According to the expert experience score of the external hidden danger factors, the attribute data in the fused data are divided into weighted influence degrees to determine the weight coefficient of each attribute data; The fused data and the corresponding weight coefficients are integrated to obtain the external environment hidden danger characteristics.

6. The method for predicting the structural surface loss state based on dam engineering according to claim 1 is characterized in that: Before performing grayscale correlation calculation on the structural surface hidden danger features in the structural surface hidden danger image and the external environment hidden danger features to obtain the hidden danger growth correlation degree, the method further includes: Normalizing the internal index in the structural surface hidden danger feature and the external index in the external environment hidden danger feature under the relevant network time slot to obtain internal index data and external index data; According to the judgment matrix constructed under the subjective factors, the correlation index between the internal index and the external index is calculated based on the subjective factors to obtain the subjective correlation index; and based on the subjective correlation index, the total hierarchical weight of the total index is determined; Based on the total hierarchical weight, the bottom-level indicator weights are calculated for both the internal indicator data and the external indicator data to obtain a subjective grayscale correlation based on the subjective factors.

7. The method for predicting the structural surface loss state based on dam engineering according to claim 1 is characterized in that: After predicting the damage growth of the surface potential hazard area of ​​the dam structure by using the damage accumulation criterion in the structure surface loss prediction model to obtain the structure damage loss intensity, the method further includes: Arrange the structural damage loss intensity of each of the structural surface hidden danger areas in time series to generate a structural damage loss intensity list; According to the preset loss intensity warning threshold, each of the structural damage loss intensity in the structural damage loss intensity list is divided into risk levels to determine the intensity risk level of each of the structural damage loss intensity; Based on the maintenance difficulty under the intensity risk level, determine the ideal maintenance time for each intensity risk level; based on the level order of the intensity risk levels, generate an initial maintenance route; According to the ideal maintenance time, the surface potential hazard areas of the structure in the same strength risk level are prioritized to obtain the same-level priority sequence, and a final maintenance route is generated; and the final maintenance route is integrated with the ideal maintenance time to generate a maintenance inspection plan.

8. A structural surface loss state prediction system based on dam engineering, characterized in that: The system comprises: Image acquisition module, used to perform core node deployment scanning at multiple points in the risk area of ​​the dam to obtain a comprehensive structural surface image; The structural analysis module is used to identify and extract the structural surface hidden danger features of the comprehensive structural surface image of the dam and determine the structural surface hidden danger image; A large database module is used to perform spatiotemporal fusion processing of multi-source external hidden danger factors on the corresponding dam section through the GIS information of the structural surface hidden danger image, and determine the hidden danger characteristics of the external environment; A correlation calculation module is used to perform grayscale correlation calculation on the structural surface hidden danger features in the structural surface hidden danger image and the external environment hidden danger features to obtain a hidden danger growth correlation degree; A model prediction module is used to predict the damage growth of the surface potential hazard area of ​​the dam structure by using the damage accumulation criterion in the structure surface loss prediction model to obtain the structure damage loss intensity; The inspection planning module is used to plan and process the information about the maintenance route and maintenance time of the surface potential hazard area of ​​the structure based on the risk level of the structural damage and loss intensity, and generate a maintenance inspection plan for the dam.

Images