Unmanned aerial vehicle ai image recognition-based water conservancy project risk identification method and system

Abstract

The application discloses a kind of water conservancy project risk identification method and system based on unmanned aerial vehicle AI image recognition, the method includes by obtaining the image of water conservancy project area, pre-processing quality optimization image is obtained;Feature is extracted and divided region generates feature vector, calculates leakage amount, obtains seepage distribution map;Initial three-dimensional data is obtained, and initial structure model is generated, and the amount of structure deformation is calculated, when exceeding preset deformation threshold, it is determined that the region is high-risk area, and structure deformation diagram is generated;Crack width, leakage amount and deformation degree are extracted, and comprehensive risk value is calculated;Image data is preprocessed, and neural network algorithm is used to extract features and classify, and clustering analysis is carried out on classification result, and structured data is generated;Comprehensive risk value and structured data are uploaded to unified platform, data analysis is carried out, future risk change trend is predicted, and risk early warning index is obtained.The method can realize accurate evaluation and intelligent early warning of water conservancy project risk.

Description

Technical Field

[0001] This invention relates to the field of water conservancy project risk identification technology, and in particular to a water conservancy project risk identification method and system based on UAV AI image recognition. Background Technology

[0002] Currently, water conservancy projects are critical infrastructures ensuring important functions such as water resource allocation, flood control, and irrigation. However, these projects face various risks during operation, including structural damage, leakage, cracks, and deformation. These risks severely impact the safe operation of the projects and the surrounding environment. Therefore, timely and accurate identification and assessment of the risks of water conservancy projects are of great significance for ensuring their safe operation and reducing potential losses.

[0003] In one existing technique, a hierarchical model is established to decompose complex decision-making problems into multiple levels and factors, which can be divided into an objective layer, a criterion layer, and an alternative layer. The weight of each factor is determined through expert scoring, and a comprehensive risk value is ultimately calculated. However, this technique relies on expert experience and subjective judgment, lacking objective data support. It is ill-suited to adapting to dynamic changes in risk and lacks the ability to process real-time data.

[0004] Existing technologies cannot accurately assess and predict the risks of water conservancy projects. Summary of the Invention

[0005] This invention provides a method and system for identifying water conservancy project risks based on UAV AI image recognition, so as to achieve accurate assessment and intelligent early warning of water conservancy project risks.

[0006] In a first aspect, to solve the above-mentioned technical problems, the present invention provides a method for risk identification in water conservancy projects, comprising:

[0007] High-resolution images of the water conservancy project area were acquired by drones, and image quality optimization algorithms were used to remove noise, fog, and rain to obtain quality-optimized images.

[0008] Based on the quality-optimized image, crack features are extracted using an edge detection algorithm to obtain a crack distribution map;

[0009] Based on the quality-optimized image, the region is divided and color and texture features are extracted to generate feature vectors. The leakage amount is calculated based on the feature vectors to obtain a leakage distribution map.

[0010] The initial three-dimensional data of the hydraulic engineering structure is obtained, the initial structural model is generated by the three-dimensional reconstruction algorithm, the structural deformation is calculated by the deformation analysis algorithm, and when the structural deformation exceeds the preset deformation threshold, the area is determined to be a high-risk area and a structural deformation map is generated.

[0011] Based on the crack distribution map, the leakage distribution map, and the structural deformation map, the crack width, leakage amount, and deformation degree are extracted, and a weighted algorithm is used to calculate the comprehensive risk value.

[0012] The quality-optimized images are processed in parallel using a distributed computing framework, and features are extracted and classified using a neural network algorithm. The classification results are then subjected to cluster analysis to generate structured data.

[0013] By using data sharing and zero-knowledge proofs, the comprehensive risk value and the structured data are uploaded to a unified platform for data analysis to predict future risk trends and obtain a risk warning index.

[0014] In one optional implementation, the process of acquiring high-resolution images of the water conservancy project area via a drone and then performing noise reduction, dehazing, and deraining processing using image quality optimization algorithms to obtain a quality-optimized image includes:

[0015] High-resolution images of the water conservancy project area were acquired by drones. Denoising was performed using a wavelet transform denoising algorithm, dehazing was performed using a dark channel prior-based dehazing algorithm, and rain removal was performed using a convolutional neural network-based deraining algorithm to obtain the second image.

[0016] Based on the second image, an image quality assessment algorithm is used to calculate the image sharpness and contrast to obtain a quality assessment score. When the quality assessment score is lower than the preset quality threshold, the parameters of the denoising, dehazing, and deraining algorithms are readjusted to generate an optimized image.

[0017] In one optional implementation, the step of extracting crack features using an edge detection algorithm based on the quality-optimized image to obtain a crack distribution map includes:

[0018] Based on the quality-optimized image, crack features are extracted using an edge detection algorithm to obtain a crack feature map;

[0019] Based on the crack feature map, a region endurance algorithm is used to locate and segment the region to obtain a crack region map.

[0020] Based on the crack region map, the crack width parameter is calculated using a morphological algorithm to obtain the width data. When the width data exceeds a preset width threshold, the region is determined to be a high-risk region, and a high-risk region map is obtained.

[0021] The crack area map and the high-risk area map are overlaid to obtain a crack distribution map.

[0022] In one optional implementation, the step of dividing the image into regions and extracting color and texture features based on the quality-optimized image to generate a feature vector, and calculating the leakage amount based on the feature vector to obtain a leakage distribution map, includes:

[0023] The quality-optimized image is divided into regions using an image segmentation algorithm. Color and texture features are extracted from the segmented image to generate a feature vector.

[0024] Based on the feature vector, the leakage amount is calculated. When the leakage amount is greater than the preset leakage threshold, the area is determined to be a high-risk area. The high-risk area and image coordinate information are combined to generate a leakage distribution map.

[0025] In one optional implementation, the steps of acquiring initial three-dimensional data of the hydraulic engineering structure, generating an initial structural model using a three-dimensional reconstruction algorithm, calculating structural deformation using a deformation analysis algorithm, and determining the area as a high-risk area when the structural deformation exceeds a preset deformation threshold, and generating a structural deformation map, include:

[0026] Obtain the original three-dimensional data of key structures in water conservancy projects, and use a three-dimensional reconstruction algorithm to generate an initial structural model;

[0027] Based on the initial structural model, the deformation of each structural unit is calculated using a deformation analysis algorithm to obtain deformation distribution data;

[0028] Based on the deformation distribution data, when the deformation exceeds a preset deformation threshold, the area is determined to be a high-risk area; a spatial interpolation algorithm is used to optimize the boundary of the high-risk area to generate an accurate risk area distribution map;

[0029] The risk area distribution map is overlaid with the initial structural model to generate a three-dimensional structural deformation map containing risk markers;

[0030] Based on the three-dimensional structural deformation diagram, structural parameters of high-risk areas are extracted, and structural safety factors are calculated; when the structural safety factor is lower than the preset safety standard, the final structural deformation diagram is generated.

[0031] In one optional implementation, the step of extracting crack width, leakage amount, and deformation degree based on the crack distribution map, the leakage distribution map, and the structural deformation map, and calculating the comprehensive risk value using a weighted algorithm, includes:

[0032] Based on the crack distribution map, the leakage distribution map, and the structural deformation map, data on crack width, leakage volume, and deformation degree are extracted; the data on crack width, leakage volume, and deformation degree are input into a pre-established risk assessment system, and a weighted algorithm is used to calculate the first risk value;

[0033] Based on the first risk value, the risk level is determined. When the first risk value is greater than the preset risk threshold, it is determined to be a high risk level; otherwise, it is a medium-low risk level. The crack width, leakage volume and deformation degree data are clustered by machine learning algorithm to identify high-risk areas and regression algorithm is used to predict the future trend of crack width, leakage volume and deformation degree to generate trend analysis results.

[0034] The comprehensive risk value is calculated by weighting the risk level, high-risk areas, and trend analysis results.

[0035] In one optional implementation, the process of using a distributed computing framework to process the quality-optimized image in parallel, extracting features and classifying them using a neural network algorithm, performing cluster analysis on the classification results, and generating structured data includes:

[0036] A distributed computing framework is used to process the quality-optimized image in parallel. The image data is divided into multiple subsets, and each subset is assigned to a different computing node. On each computing node, the image data is subjected to denoising, grayscale conversion, and normalization operations to obtain the processed image.

[0037] Based on the processed image, a convolutional neural network algorithm is used to extract edge, texture, and shape features to generate feature vectors; the extracted feature vectors are then input into a support vector machine algorithm for classification to determine the category to which the image data belongs, and the classification result is obtained.

[0038] Based on the classification results, a clustering algorithm is used to perform cluster analysis on the image data, grouping image data with similar features into the same category to generate clustering results;

[0039] The clustering results are matched with preset category labels to determine the category label of each cluster, generating structured data.

[0040] Secondly, the present invention provides a risk identification system for water conservancy projects, comprising:

[0041] The image processing module is used to acquire high-resolution images of water conservancy project areas via UAVs, and to perform noise reduction, defogging, and deraining processing using image quality optimization algorithms to obtain quality-optimized images.

[0042] The crack extraction module is used to extract crack features based on the quality-optimized image using an edge detection algorithm to obtain a crack distribution map;

[0043] The leakage calculation module is used to divide the region and extract color and texture features based on the quality-optimized image, generate feature vectors, calculate the leakage amount based on the feature vectors, and obtain a leakage distribution map.

[0044] The deformation analysis module is used to acquire the initial three-dimensional data of the hydraulic engineering structure, generate the initial structural model using a three-dimensional reconstruction algorithm, calculate the structural deformation using a deformation analysis algorithm, and determine the area as a high-risk area when the structural deformation exceeds the preset deformation threshold, and generate a structural deformation map.

[0045] The risk calculation module is used to extract the crack width, leakage amount and deformation degree based on the crack distribution map, the leakage distribution map and the structural deformation map, and to calculate the comprehensive risk value using a weighted algorithm.

[0046] The structured module is used to perform parallel processing on the quality-optimized image using a distributed computing framework, extract features and classify them using a neural network algorithm, perform cluster analysis on the classification results, and generate structured data.

[0047] The risk warning module is used to upload the comprehensive risk value and the structured data to a unified platform using data sharing and zero-knowledge proofs, perform data analysis, predict future risk trends, and obtain a risk warning index.

[0048] Thirdly, the present invention also provides an electronic device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor executes the computer program to implement the water conservancy project risk identification method described in any one of the above.

[0049] Fourthly, the present invention also provides a computer-readable storage medium comprising a stored computer program, wherein, when the computer program is executed, it controls the device where the computer-readable storage medium is located to perform the water conservancy project risk identification method described in any one of the above-mentioned methods.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] This method includes acquiring high-resolution images of the water conservancy project area using a drone; performing noise reduction, defogging, and rain removal processing using image quality optimization algorithms to obtain a quality-optimized image; extracting crack features using an edge detection algorithm based on the quality-optimized image to obtain a crack distribution map; dividing the area into regions and extracting color and texture features based on the quality-optimized image to generate feature vectors; calculating leakage based on the feature vectors to obtain a leakage distribution map; acquiring initial three-dimensional data of the water conservancy project structure; generating an initial structural model using a three-dimensional reconstruction algorithm; calculating structural deformation using a deformation analysis algorithm; and determining when the structural deformation exceeds a preset threshold. When the deformation threshold is reached, the area is determined to be a high-risk area, and a structural deformation map is generated. Based on the crack distribution map, the leakage distribution map, and the structural deformation map, the crack width, leakage amount, and deformation degree are extracted, and a weighted algorithm is used to calculate the comprehensive risk value. A distributed computing framework is used to process the quality-optimized image in parallel, and a neural network algorithm is used to extract features and classify them. Cluster analysis is performed on the classification results to generate structured data. Data sharing and zero-knowledge proofs are used to upload the comprehensive risk value and the structured data to a unified platform for data analysis to predict future risk trends and obtain a risk warning index.

[0052] This invention acquires high-resolution images using drones, processes them with image quality optimization algorithms, and extracts key features such as cracks, leaks, and structural deformation. Using feature analysis and image segmentation techniques, it accurately locates risk areas and generates distribution maps. Three-dimensional reconstruction technology is employed to analyze the degree of structural deformation. Multiple risk factors are comprehensively assessed to determine the risk level. This invention also utilizes a distributed computing framework to process massive amounts of data and employs zero-knowledge proof technology to securely share assessment results. Through historical data analysis, it predicts future risk trends and generates early warning reports. This achieves accurate assessment, efficient processing, and intelligent early warning of risks in water conservancy projects. Attached Figure Description

[0053] Figure 1 This is a schematic diagram of the process of the water conservancy project risk identification method based on UAV AI image recognition provided in the first embodiment of the present invention;

[0054] Figure 2 This is a schematic diagram of the system structure of the water conservancy project risk identification method based on UAV AI image recognition provided in the second embodiment of the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Reference Figure 1 The first embodiment of the present invention provides a method for risk identification in water conservancy projects, comprising the following steps:

[0057] S11: High-resolution images of the water conservancy project area are acquired by drones, and image quality optimization algorithms are used to denoise, defog, and derain the image to obtain a quality-optimized image.

[0058] S12, Based on the quality-optimized image, an edge detection algorithm is used to extract crack features to obtain a crack distribution map;

[0059] S13, Based on the quality-optimized image, perform region division and extract color and texture features to generate feature vectors, calculate the leakage amount based on the feature vectors, and obtain a leakage distribution map;

[0060] S14. Obtain the initial three-dimensional data of the hydraulic engineering structure, generate the initial structural model using a three-dimensional reconstruction algorithm, calculate the structural deformation using a deformation analysis algorithm, and determine the area as a high-risk area when the structural deformation exceeds the preset deformation threshold, and generate a structural deformation map.

[0061] S15. Based on the crack distribution map, the leakage distribution map, and the structural deformation map, extract the crack width, leakage amount, and deformation degree, and use a weighted algorithm to calculate the comprehensive risk value.

[0062] S16, The quality-optimized image is processed in parallel using a distributed computing framework, features are extracted and classified using a neural network algorithm, and cluster analysis is performed on the classification results to generate structured data;

[0063] S17. Using data sharing and zero-knowledge proof, the comprehensive risk value and the structured data are uploaded to a unified platform for data analysis to predict future risk trends and obtain a risk warning index.

[0064] In step S11, high-resolution images of the water conservancy project area are acquired by a drone, and image quality optimization algorithms are used to perform noise reduction, defogging, and deraining to obtain a quality-optimized image.

[0065] In one implementation, a high-resolution image of the water conservancy project area is acquired by a drone, and then a wavelet transform denoising algorithm is used for denoising, a dark channel prior-based dehazing algorithm is used for dehazing, and a convolutional neural network-based deraining algorithm is used for deraining to obtain a second image.

[0066] Based on the second image, an image quality assessment algorithm is used to calculate the image sharpness and contrast to obtain a quality assessment score. When the quality assessment score is lower than the preset quality threshold, the parameters of the denoising, dehazing, and deraining algorithms are readjusted to generate an optimized image.

[0067] It should be noted that wavelet transform denoising is a signal processing technique based on wavelet transform. It decomposes the signal into sub-signals of different frequencies and uses a thresholding method to denoise these sub-signals. Its basic principle is to use wavelet transform to decompose the noisy signal into different scales, then process the wavelet coefficients, and finally reconstruct the signal through inverse wavelet transform, thereby removing noise. Dark channel prior dehazing is a dehazing method based on image brightness information. Its core idea is to use atmospheric scattering models and dark channel prior knowledge to estimate atmospheric light values ​​and transmittance, thereby achieving dehazing. Image quality assessment algorithms are used to measure image quality. Objective assessment methods evaluate image quality by calculating certain feature values. Image quality is evaluated by comparing the pixel value differences between the original image and the processed image; a higher PSNR value indicates better image quality. Image similarity can also be evaluated based on brightness, contrast, and structural information. The closer the SSIM value is to 1, the better the image quality.

[0068] For example, in a reservoir dam construction project, a hexacopter drone equipped with a 4K camera was used to acquire orthophotos covering an area of ​​20 square kilometers at a speed of 30 m / s and an altitude of 500 meters, achieving a resolution of 2 cm. The sym4 wavelet basis was used for denoising, and the image was decomposed into three layers. A soft thresholding method was applied to shrink coefficients in the high-frequency subband, and finally, wavelet reconstruction was performed to obtain the denoised image. This method preserves the edge details of the image while effectively suppressing noise. In foggy conditions, images are affected by fog. A dehazing algorithm based on dark channel priors can effectively recover scene details obscured by fog by estimating atmospheric illumination and transmittance. By calculating the dark channel map, the atmospheric illumination value was estimated to be 220, and guided filtering was used to optimize the transmittance map, ultimately obtaining a clear dehazed image, making the originally blurred river outline and surrounding vegetation clearly visible. In images taken on rainy days, raindrops severely affect image quality. A deraining algorithm based on convolutional neural networks can automatically learn raindrop features and remove them. For example, a multi-scale residual network with 20 convolutional layers and 64 3x3 convolutional kernels, trained for 100 epochs on 10,000 rainy images, can effectively remove raindrops of different sizes and densities, restoring image details. Image quality assessment is a crucial step in ensuring processing effectiveness. Metrics such as structural similarity and peak signal-to-noise ratio (PSNR) can be used. The SSIM threshold is set to 0.85, and the PSNR threshold to 30 dB. If the assessment result is below the threshold, the aforementioned algorithm parameters need to be adjusted. Increasing the number of wavelet decomposition layers in the denoising algorithm, increasing the minimum filtering window size in the dehazing algorithm, or increasing the number of training epochs in the rain removal network can achieve better processing results. Through this series of processing steps, the final high-quality image can be used for precise analysis in water conservancy projects.

[0069] In step S12, crack features are extracted using an edge detection algorithm based on the quality-optimized image to obtain a crack distribution map.

[0070] In one implementation, crack features are extracted using an edge detection algorithm based on the quality-optimized image to obtain a crack feature map.

[0071] Based on the crack feature map, a region endurance algorithm is used to locate and segment the region to obtain a crack region map.

[0072] Based on the crack region map, the crack width parameter is calculated using a morphological algorithm to obtain the width data. When the width data exceeds a preset width threshold, the region is determined to be a high-risk region, and a high-risk region map is obtained.

[0073] The crack area map and the high-risk area map are overlaid to obtain a crack distribution map.

[0074] It should be noted that edge detection is an important technique in image processing, used to detect regions in an image where grayscale values ​​change significantly. It can be used to extract image contours or features. Edge detection algorithms include the Sobel algorithm, Canny algorithm, Prewitt algorithm, and LoG algorithm. Region growing is an image segmentation method based on pixel similarity. Its basic idea is to start from one or more seed points and gradually merge adjacent pixels with similar features into the same region. The advantages of region growing are its simplicity, intuitiveness, ease of implementation, and robustness to noise.

[0075] For example, after acquiring the quality-optimized image, the Canny algorithm is used. This algorithm effectively extracts edge information from the image through Gaussian filtering, gradient calculation, non-maximum suppression, and double threshold detection, forming a crack feature map. This method accurately captures the crack outline, laying the foundation for subsequent analysis. Next, a region growing algorithm is used to accurately locate and segment the crack region. Starting from a seed point, the region growing algorithm gradually merges similar neighboring pixels, eventually forming a complete crack region. For example, a point with a lower grayscale value in the crack feature map can be selected as a seed point. Then, based on a preset grayscale threshold, surrounding pixels meeting the criteria are continuously incorporated into the region until further expansion is impossible. This yields an accurate crack region map, which is helpful for subsequent width calculation and risk assessment. After extracting crack edge information from the crack region map, morphological algorithms are used to calculate the crack width parameter. Commonly used morphological operations include erosion and dilation. By performing erosion on the crack region, the crack skeleton line can be obtained. Then, by performing dilation, the original width of the crack can be restored. The pixel difference between the two operations is the crack width. For example, assuming a crack is 1 pixel wide after corrosion and recovers to 5 pixels after expansion, the crack width at that location is 4 pixels. This method can quickly and accurately obtain crack width data. Based on the calculated width data and a preset threshold, high-risk areas can be identified. For example, assuming a preset threshold of 5 millimeters, crack areas wider than 5 millimeters are marked as high-risk areas. These areas can be marked in red in the image, forming a high-risk area map. This visualization method intuitively shows areas requiring special attention, helping engineers quickly identify potential hazards. Overlaying the crack area map with the high-risk area map generates a comprehensive crack distribution map.

[0076] In step S13, based on the quality-optimized image, the region is divided and color and texture features are extracted to generate a feature vector. The leakage amount is calculated based on the feature vector to obtain a leakage distribution map.

[0077] In one implementation, the quality-optimized image is divided into regions using an image segmentation algorithm, and color and texture features are extracted from the segmented image to generate a feature vector.

[0078] Based on the feature vector, the leakage amount is calculated. When the leakage amount is greater than the preset leakage threshold, the area is determined to be a high-risk area. The high-risk area and image coordinate information are combined to generate a leakage distribution map.

[0079] It is important to note that leakage image analysis is a crucial step in the maintenance of underground building structures. Image segmentation algorithms, such as the watershed method or region growing, can effectively divide image regions, separating leaking areas from non-leaking areas. Color feature extraction can utilize the HSV color space to analyze the hue, saturation, and brightness values ​​of leaking areas. For example, leaking areas may appear as darker gray or brown with low saturation. Texture features can be extracted using gray-level co-occurrence matrix or wavelet transform, reflecting the roughness and irregularity of the leaking area. Feature vector generation combines color and texture information to form a multi-dimensional data representation. For example, a 5-dimensional feature vector can contain 3 color components and 2 texture descriptors. Using machine learning algorithms such as support vector machines or random forests, leaking areas can be accurately identified based on feature vectors. Leakage calculation considers the area and intensity of the leaking area. Assuming an area of ​​10 square meters and an average leakage intensity of 2 (divided into 0-5 levels), the estimated leakage volume is 20 units. If a preset threshold of 15 units is set, the area will be marked as high-risk. By combining high-risk areas with image coordinate information, a leakage distribution map is generated.

[0080] In step S14, the initial three-dimensional data of the hydraulic engineering structure is obtained, the initial structural model is generated by the three-dimensional reconstruction algorithm, the structural deformation is calculated by the deformation analysis algorithm, and when the structural deformation exceeds the preset deformation threshold, the area is determined to be a high-risk area and a structural deformation map is generated.

[0081] In one implementation, the original three-dimensional data of the key structures of the water conservancy project are obtained, and an initial structural model is generated using a three-dimensional reconstruction algorithm.

[0082] Obtain the original three-dimensional data of key structures in water conservancy projects, and use a three-dimensional reconstruction algorithm to generate an initial structural model;

[0083] Based on the initial structural model, the deformation of each structural unit is calculated using a deformation analysis algorithm to obtain deformation distribution data;

[0084] Based on the deformation distribution data, when the deformation exceeds a preset deformation threshold, the area is determined to be a high-risk area; a spatial interpolation algorithm is used to optimize the boundary of the high-risk area to generate an accurate risk area distribution map;

[0085] The risk area distribution map is overlaid with the initial structural model to generate a three-dimensional structural deformation map containing risk markers;

[0086] Based on the three-dimensional structural deformation diagram, structural parameters of high-risk areas are extracted, and structural safety factors are calculated; when the structural safety factor is lower than the preset safety standard, the final structural deformation diagram is generated.

[0087] It is important to note that the 3D reconstruction of key structures in water conservancy projects is a crucial step in ensuring project safety. Taking a large reservoir dam as an example, the first step is to obtain the original 3D point cloud data of the dam. High-precision point cloud data, containing geometric information about the dam surface, can be obtained through laser scanning technology. Using this data, an octree-based 3D reconstruction algorithm is employed to transform the discrete point cloud into a continuous curved surface model, generating an initial structural model. For this initial structural model, deformation analysis is performed using the finite element method. The dam model is divided into several elements, and factors such as water pressure and temperature changes are considered to calculate the stress and deformation of each element. For example, if the maximum deformation of a certain element is 5mm, exceeding the preset threshold of 3mm, this area is marked as a high-risk area. To more accurately locate high-risk areas, Kriging interpolation can be used to optimize the boundaries of the risk areas. This method considers spatial correlation and can perform reasonable interpolation estimations between known points. Through interpolation optimization, a more continuous and smooth risk area distribution map can be obtained, which is beneficial for subsequent risk assessment and handling. The optimized risk area distribution map is then overlaid with the initial structural model to obtain a 3D structural deformation map. Based on the 3D structural deformation map, structural parameters such as stress distribution and crack width can be extracted from high-risk areas. These parameters can then be used to calculate the structural safety factor. Assuming a safety factor of 1.2 for a certain high-risk area, lower than the preset standard of 1.5, this indicates a potential safety hazard in the area. Finally, the final structural deformation map is generated.

[0088] In step S15, based on the crack distribution map, the leakage distribution map, and the structural deformation map, the crack width, leakage amount, and deformation degree are extracted, and a weighted algorithm is used to calculate the comprehensive risk value.

[0089] In one implementation, based on the crack distribution map, the leakage distribution map, and the structural deformation map, crack width, leakage amount, and deformation degree data are extracted; the crack width, leakage amount, and deformation degree data are input into a pre-established risk assessment system, and a weighted algorithm is used to calculate a first risk value;

[0090] Based on the first risk value, the risk level is determined. When the first risk value is greater than the preset risk threshold, it is determined to be a high risk level; otherwise, it is a medium-low risk level. The crack width, leakage volume and deformation degree data are clustered by machine learning algorithm to identify high-risk areas and regression algorithm is used to predict the future trend of crack width, leakage volume and deformation degree to generate trend analysis results.

[0091] The comprehensive risk value is calculated by weighting the risk level, high-risk areas, and trend analysis results.

[0092] It's important to note that acquiring crack distribution maps, seepage distribution maps, and structural deformation maps provides image data reflecting the actual condition of the engineering structure. For example, a crack distribution map of a dam shows a 5-meter-long, 2-millimeter-wide crack in the middle of the dam body; a seepage distribution map indicates a seepage rate of 100 liters per hour at the dam foundation; and a structural deformation map reveals a settlement deformation of 10 millimeters at the center of the dam crest. Key data, such as crack width, seepage rate, and deformation degree, are extracted from these images. These data are crucial indicators for risk assessment. The extracted data is input into a pre-established risk assessment system, and a weighted algorithm is used to calculate the comprehensive risk value. This system assigns different weights to different factors, such as crack width (0.4), seepage rate (0.3), and deformation degree (0.3). If the comprehensive risk value calculated through weighted calculation exceeds a preset threshold of 0.7, it is classified as a high-risk level; otherwise, it is classified as a medium-to-low-risk level. Next, machine learning algorithms are used to perform cluster analysis on the data to identify high-risk areas. For example, the K-means clustering algorithm is used to divide data points into several clusters, with clusters having higher risk indicators identified as high-risk areas. This method can quickly locate areas requiring focused attention. Simultaneously, regression algorithms are used to predict future trends. For instance, for crack width, a linear regression model can be used to predict its change over the next year. If the prediction shows that the crack width will increase to 3 millimeters, it provides important early warning information. Finally, a weighted average of the risk level, high-risk areas, and trend analysis results is used to calculate the comprehensive risk value.

[0093] In step S16, a distributed computing framework is used to process the quality-optimized image in parallel, a neural network algorithm is used to extract features and classify them, and cluster analysis is performed on the classification results to generate structured data.

[0094] In one implementation, a distributed computing framework is used to process the quality-optimized image in parallel, dividing the image data into multiple subsets, each subset being assigned to a different computing node; on each computing node, the image data is subjected to denoising, grayscale conversion, and normalization operations to obtain the processed image;

[0095] Based on the processed image, a convolutional neural network algorithm is used to extract edge, texture, and shape features to generate feature vectors; the extracted feature vectors are then input into a support vector machine algorithm for classification to determine the category to which the image data belongs, and the classification result is obtained.

[0096] Based on the classification results, a clustering algorithm is used to perform cluster analysis on the image data, grouping image data with similar features into the same category to generate clustering results;

[0097] The clustering results are matched with preset category labels to determine the category label of each cluster, generating structured data.

[0098] It's worth noting that distributed computing frameworks offer significant advantages when processing massive amounts of image data. Taking drone aerial photography of forest fires as an example, the large volume of acquired image data can be divided into multiple subsets and distributed to different computing nodes for parallel processing. Each node is responsible for processing images of a specific area or time period, improving overall processing efficiency. Image preprocessing is a crucial step in improving the accuracy of subsequent analysis. For forest fire images, denoising eliminates interference factors such as smoke, grayscale conversion helps highlight the contrast between flames and the surrounding environment, and normalization ensures comparability of images taken under different lighting conditions. These operations lay the foundation for feature extraction. Convolutional neural networks excel at extracting image features. For forest fire images, neural networks focus on extracting the edge features of flames, the texture features of smoke, and the shape changes of trees after burning. The feature vectors formed by combining these features can effectively describe various aspects of the fire. Support Vector Machine (SVM) algorithms use the extracted feature vectors for classification, dividing images into categories such as fire areas, potential fire hazard areas, and safe areas. Cluster analysis further refines the classification results. For example, fire areas can be clustered into subcategories such as initial fire, full-blown fire, and ember stage. This detailed segmentation helps in developing more precise firefighting strategies. The clustering results are matched with preset labels to generate structured data.

[0099] In summary, this invention discloses a method for risk identification in water conservancy projects based on UAV AI image processing. The method includes acquiring high-resolution images of the water conservancy project area using a UAV; performing noise reduction, defogging, and rain removal using image quality optimization algorithms to obtain a quality-optimized image; extracting crack features using an edge detection algorithm based on the quality-optimized image to obtain a crack distribution map; dividing the area into regions and extracting color and texture features from the quality-optimized image to generate feature vectors; calculating leakage based on the feature vectors to obtain a leakage distribution map; acquiring initial three-dimensional data of the water conservancy project structure; generating an initial structural model using a three-dimensional reconstruction algorithm; and calculating leakage using a deformation analysis algorithm. The structural deformation is assessed. When the structural deformation exceeds a preset threshold, the area is identified as a high-risk area, and a structural deformation map is generated. Based on the crack distribution map, the leakage distribution map, and the structural deformation map, crack width, leakage amount, and deformation degree are extracted, and a weighted algorithm is used to calculate a comprehensive risk value. A distributed computing framework is used to process the quality-optimized image in parallel, and a neural network algorithm is used to extract features and classify them. Cluster analysis is performed on the classification results to generate structured data. Data sharing and zero-knowledge proofs are used to upload the comprehensive risk value and the structured data to a unified platform for data analysis to predict future risk trends and obtain a risk warning index.

[0100] This invention acquires high-resolution images using drones, processes them with image quality optimization algorithms, and extracts key features such as cracks, leaks, and structural deformation. Using feature analysis and image segmentation techniques, it accurately locates risk areas and generates distribution maps. Three-dimensional reconstruction technology is employed to analyze the degree of structural deformation. Multiple risk factors are comprehensively assessed to determine the risk level. This invention also utilizes a distributed computing framework to process massive amounts of data and employs zero-knowledge proof technology to securely share assessment results. Through historical data analysis, it predicts future risk trends and generates early warning reports. This achieves accurate assessment, efficient processing, and intelligent early warning of risks in water conservancy projects.

[0101] Reference Figure 2 The second embodiment of the present invention provides a water conservancy project risk identification system, comprising:

[0102] The image processing module is used to acquire high-resolution images of water conservancy project areas via UAVs, and to perform noise reduction, defogging, and deraining processing using image quality optimization algorithms to obtain quality-optimized images.

[0103] The crack extraction module is used to extract crack features based on the quality-optimized image using an edge detection algorithm to obtain a crack distribution map;

[0104] The leakage calculation module is used to divide the region and extract color and texture features based on the quality-optimized image, generate feature vectors, calculate the leakage amount based on the feature vectors, and obtain a leakage distribution map.

[0105] The deformation analysis module is used to acquire the initial three-dimensional data of the hydraulic engineering structure, generate the initial structural model using a three-dimensional reconstruction algorithm, calculate the structural deformation using a deformation analysis algorithm, and determine the area as a high-risk area when the structural deformation exceeds the preset deformation threshold, and generate a structural deformation map.

[0106] The risk calculation module is used to extract the crack width, leakage amount and deformation degree based on the crack distribution map, the leakage distribution map and the structural deformation map, and to calculate the comprehensive risk value using a weighted algorithm.

[0107] The structured module is used to perform parallel processing on the quality-optimized image using a distributed computing framework, extract features and classify them using a neural network algorithm, perform cluster analysis on the classification results, and generate structured data.

[0108] The risk warning module is used to upload the comprehensive risk value and the structured data to a unified platform using data sharing and zero-knowledge proofs, perform data analysis, predict future risk trends, and obtain a risk warning index.

[0109] It should be noted that the water conservancy project risk identification device provided in this embodiment of the invention is used to execute all the process steps of the water conservancy project risk identification method in the above embodiment. The working principle and beneficial effect of the two are one-to-one, so they will not be described again.

[0110] This invention also provides an electronic device. The electronic device includes a processor, a memory, and a computer program stored in the memory and executable on the processor, such as a water conservancy project risk identification program. When the processor executes the computer program, it implements the steps described in the various water conservancy project risk identification method embodiments above, for example... Figure 1 Step S11 is shown. Alternatively, when the processor executes the computer program, it implements the functions of each module / unit in the above-described device embodiments, such as the water conservancy project risk identification module.

[0111] For example, the computer program may be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the electronic device.

[0112] The electronic device may be a desktop computer, laptop, handheld computer, or smart tablet, etc. The electronic device may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above components are merely examples of electronic devices and do not constitute a limitation on the electronic device. It may include more or fewer components than described above, or combine certain components, or different components. For example, the electronic device may also include input / output devices, network access devices, buses, etc.

[0113] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the electronic device, connecting all parts of the electronic device via various interfaces and lines.

[0114] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the electronic device by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system, at least one application program required for a function (such as sound playback function, image playback function, etc.), etc.; the data storage area may store data created according to the use of the mobile phone (such as audio data, phonebook, etc.). In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disk, memory, plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0115] Wherein, if the modules / units integrated in the electronic device are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signals, telecommunication signals, and software distribution media, etc. It should be noted that the content included in the computer-readable medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electrical carrier signals and telecommunication signals.

[0116] It should be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the device embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0117] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above descriptions are merely specific embodiments of the present invention and are not intended to limit the scope of protection of the present invention. In particular, it should be noted that any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention for those skilled in the art.

Claims

1. A method for risk identification in water conservancy projects, characterized in that, Executed by a computer, including: High-resolution images of the water conservancy project area were acquired by drones, and image quality optimization algorithms were used to remove noise, fog, and rain to obtain quality-optimized images. Based on the quality-optimized image, crack features are extracted using an edge detection algorithm to obtain a crack distribution map; Based on the quality-optimized image, the region is divided and color and texture features are extracted to generate feature vectors. The leakage amount is calculated based on the feature vectors to obtain a leakage distribution map. The initial three-dimensional data of the hydraulic engineering structure is obtained, the initial structural model is generated by the three-dimensional reconstruction algorithm, the structural deformation is calculated by the deformation analysis algorithm, and when the structural deformation exceeds the preset deformation threshold, the area is determined to be a high-risk area and a structural deformation map is generated. Based on the crack distribution map, the leakage distribution map, and the structural deformation map, the crack width, leakage amount, and deformation degree are extracted, and a weighted algorithm is used to calculate the comprehensive risk value. The quality-optimized images are processed in parallel using a distributed computing framework, and features are extracted and classified using a neural network algorithm. The classification results are then subjected to cluster analysis to generate structured data. By using data sharing and zero-knowledge proofs, the comprehensive risk value and the structured data are uploaded to a unified platform for data analysis to predict future risk trends and obtain a risk warning index.

2. The method for risk identification in water conservancy projects according to claim 1, characterized in that, The process involves acquiring high-resolution images of the water conservancy project area using a drone, and then applying image quality optimization algorithms for denoising, defogging, and deraining to obtain an optimized image, including: High-resolution images of the water conservancy project area were acquired by drones. Denoising was performed using a wavelet transform denoising algorithm, dehazing was performed using a dark channel prior-based dehazing algorithm, and rain removal was performed using a convolutional neural network-based deraining algorithm to obtain the second image. Based on the second image, an image quality assessment algorithm is used to calculate the image sharpness and contrast to obtain a quality assessment score. When the quality assessment score is lower than the preset quality threshold, the parameters of the denoising, dehazing, and deraining algorithms are readjusted to generate an optimized image.

3. The method for risk identification in water conservancy projects according to claim 1, characterized in that, The step of extracting crack features using an edge detection algorithm based on the quality-optimized image to obtain a crack distribution map includes: Based on the quality-optimized image, crack features are extracted using an edge detection algorithm to obtain a crack feature map; Based on the crack feature map, a region endurance algorithm is used to locate and segment the region to obtain a crack region map. Based on the crack region map, the crack width parameter is calculated using a morphological algorithm to obtain the width data. When the width data exceeds a preset width threshold, the region is determined to be a high-risk region, and a high-risk region map is obtained. The crack area map and the high-risk area map are overlaid to obtain a crack distribution map.

4. The method for risk identification in water conservancy projects according to claim 1, characterized in that, The process involves dividing the image into regions and extracting color and texture features based on the optimized image, thereby generating features. vector, The leakage rate is calculated based on the eigenvectors, resulting in a leakage distribution map, including: The quality-optimized image is divided into regions using an image segmentation algorithm. Color and texture features are extracted from the segmented image to generate a feature vector. Based on the feature vector, the leakage amount is calculated. When the leakage amount is greater than the preset leakage threshold, the area is determined to be a high-risk area. The high-risk area and image coordinate information are combined to generate a leakage distribution map.

5. The method for risk identification in water conservancy projects according to claim 1, characterized in that, The process involves acquiring initial three-dimensional data of the hydraulic engineering structure, generating an initial structural model using a three-dimensional reconstruction algorithm, calculating structural deformation using a deformation analysis algorithm, and determining the area as a high-risk area when the structural deformation exceeds a preset deformation threshold, and generating a structural deformation map, including: Obtain the original three-dimensional data of key structures in water conservancy projects, and use a three-dimensional reconstruction algorithm to generate an initial structural model; Based on the initial structural model, the deformation of each structural unit is calculated using a deformation analysis algorithm to obtain deformation distribution data; Based on the deformation distribution data, when the deformation exceeds a preset deformation threshold, the area is determined to be a high-risk area; a spatial interpolation algorithm is used to optimize the boundary of the high-risk area to generate an accurate risk area distribution map; The risk area distribution map is overlaid with the initial structural model to generate a three-dimensional structural deformation map containing risk markers; Based on the three-dimensional structural deformation diagram, structural parameters of high-risk areas are extracted, and structural safety factors are calculated; when the structural safety factor is lower than the preset safety standard, the final structural deformation diagram is generated.

6. The method for risk identification in water conservancy projects according to claim 2, characterized in that, The step involves extracting crack width, leakage amount, and deformation degree based on the crack distribution map, leakage distribution map, and structural deformation map, and then calculating a comprehensive risk value using a weighted algorithm, including: Based on the crack distribution map, the leakage distribution map, and the structural deformation map, data on crack width, leakage volume, and deformation degree are extracted; the data on crack width, leakage volume, and deformation degree are input into a pre-established risk assessment system, and a weighted algorithm is used to calculate the first risk value; Based on the first risk value, the risk level is determined. When the first risk value is greater than the preset risk threshold, it is determined to be a high risk level; otherwise, it is a medium-low risk level. The crack width, leakage volume and deformation degree data are clustered by machine learning algorithm to identify high-risk areas and regression algorithm is used to predict the future trend of crack width, leakage volume and deformation degree, and generate trend analysis results. The comprehensive risk value is calculated by weighting the risk level, high-risk areas, and trend analysis results.

7. The method for risk identification in water conservancy projects according to claim 1, characterized in that, The process employs a distributed computing framework to perform parallel processing on the quality-optimized images, uses neural network algorithms to extract features and classify them, performs cluster analysis on the classification results, and generates structured data, including: A distributed computing framework is used to process the quality-optimized image in parallel. The image data is divided into multiple subsets, and each subset is assigned to a different computing node. On each computing node, the image data is subjected to denoising, grayscale conversion, and normalization operations to obtain the processed image. Based on the processed image, a convolutional neural network algorithm is used to extract edge, texture, and shape features to generate feature vectors; the extracted feature vectors are then input into a support vector machine algorithm for classification to determine the category to which the image data belongs, and the classification result is obtained. Based on the classification results, a clustering algorithm is used to perform cluster analysis on the image data, grouping image data with similar features into the same category to generate clustering results; The clustering results are matched with preset category labels to determine the category label of each cluster, generating structured data.

8. A risk identification system for water conservancy projects, characterized in that, include: The image processing module is used to acquire high-resolution images of water conservancy project areas via UAVs, and to perform noise reduction, defogging, and deraining processing using image quality optimization algorithms to obtain quality-optimized images. The crack extraction module is used to extract crack features based on the quality-optimized image using an edge detection algorithm to obtain a crack distribution map; The leakage calculation module is used to divide the region and extract color and texture features based on the quality-optimized image, generate feature vectors, calculate the leakage amount based on the feature vectors, and obtain a leakage distribution map. The deformation analysis module is used to acquire the initial three-dimensional data of the hydraulic engineering structure, generate the initial structural model using a three-dimensional reconstruction algorithm, calculate the structural deformation using a deformation analysis algorithm, and determine the area as a high-risk area when the structural deformation exceeds the preset deformation threshold, and generate a structural deformation map. The risk calculation module is used to extract the crack width, leakage amount and deformation degree based on the crack distribution map, the leakage distribution map and the structural deformation map, and to calculate the comprehensive risk value using a weighted algorithm. The structured module is used to perform parallel processing on the quality-optimized image using a distributed computing framework, extract features and classify them using a neural network algorithm, perform cluster analysis on the classification results, and generate structured data. The risk warning module is used to upload the comprehensive risk value and the structured data to a unified platform using data sharing and zero-knowledge proofs, perform data analysis, predict future risk trends, and obtain a risk warning index.

9. An electronic device, characterized in that, The system includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor, when executing the computer program, implements the water conservancy project risk identification method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a stored computer program, wherein, when the computer program is executed, it controls the device containing the computer-readable storage medium to perform the water conservancy project risk identification method as described in any one of claims 1 to 7.

Images