A method, system and device for researching and judging water conservancy river and lake four disorder problems

Abstract

The application discloses a kind of water conservancy river and lake four disorder problems research and judgment method, system and equipment, is specifically relevant to water conservancy river and lake research determination field, including constructing multi-source heterogeneous data set and extracting dynamic environment feature generates the feature data set of target water area;Lightweight recognition engine is deployed to feature dimension reduction in edge computing node and generates four light weight feature packages;Through space-time correlation analysis, generate multidimensional decision graph;Finally drive visual early warning platform realizes four disorder problems dynamic visualization.A kind of water conservancy river and lake four disorder problems research and judgment method, system and equipment by controlling space-time deviation within threshold, realized the automatic alignment and efficient fusion of multi-source data, significantly reduce the need for manual intervention and pre-processing computing resource consumption;Through radiation correction coefficient and dynamic weight distribution strategy, reduce the model parameter adjustment frequency;Through space-time correlation analysis, provide multidimensional scientific basis for decision-making, optimize the management efficiency.

Description

Technical Field

[0001] This invention relates to the field of water conservancy and river / lake research and assessment technology, and more specifically, to a method, system, and equipment for assessing the four types of disorder in water conservancy and river / lake systems. Background Technology

[0002] With the acceleration of urbanization and the increasing pressure on the ecological environment in my country, the problems of "four irregularities" in rivers and lakes have developed into a key challenge threatening water resource security and ecological balance. Traditional manual inspection methods are generally limited by low efficiency and narrow coverage. Although drone aerial photography and satellite remote sensing technologies have been gradually applied to the field of river and lake monitoring, there are still obvious shortcomings in the intelligent identification, accurate judgment and rapid handling of the "four irregularities".

[0003] Currently, at the data collection level, high-precision aerial photography monitoring of key areas is carried out using drones, combined with computer vision algorithms for automated screening of "four irregularities" (illegal construction, illegal dumping, illegal land use, and illegal land use), effectively improving monitoring efficiency and coverage. At the data analysis level, basic geographic information of rivers and lakes, real-time monitoring data, and historical governance records are integrated, and machine learning algorithms are used to achieve intelligent classification of problems and risk prediction. At the decision support level, early warning models are established based on spatiotemporal big data analysis technology, providing a scientific basis for the dynamic optimization of governance solutions.

[0004] However, in practical use, it still has some drawbacks. For example, due to differences in spatiotemporal resolution and format standards, data from UAV aerial photography, satellite remote sensing, and geographic information systems require manual intervention for alignment and correlation, leading to excessive computational resource consumption in the data preprocessing stage. It also has weak adaptability to dynamic environments; existing algorithms are prone to feature misjudgment in scenarios with complex lighting, vegetation obstruction, or seasonal hydrological changes, requiring repeated adjustments to model parameters to maintain recognition accuracy. Furthermore, its real-time processing capabilities at the edge are limited; high-resolution image data transmission to the cloud suffers from high bandwidth consumption and response latency, hindering rapid on-site assessment. Therefore, a method, system, and equipment for assessing the four major problems of water conservancy, river, and lake management are needed to improve the timeliness of monitoring and early warning and the scientific nature of decision optimization. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art, the present invention provides a method, system and equipment for judging the four problems of water conservancy, river and lake disorder, and solves the problems mentioned in the background art through the following solutions.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A method for assessing the four irregularities in water conservancy, rivers, and lakes includes:

[0008] S1: Acquire real-time images of the target water area by using a drone equipped with a multispectral sensor, and simultaneously receive satellite remote sensing data, geographic information system vector data, and real-time environmental perception data including light intensity, vegetation cover index, and hydrological parameters to construct a multi-source heterogeneous dataset of the target water area.

[0009] S2: Dynamically extract environmental features from the multi-source heterogeneous dataset to generate a feature dataset of the target water area;

[0010] S3: Pre-deploy a lightweight recognition engine on the edge computing node to perform region segmentation and feature dimensionality reduction on the feature dataset, generating four types of lightweight feature packages;

[0011] S4: Perform spatiotemporal correlation analysis between the four types of lightweight feature packages and the historical governance database to generate a multidimensional decision map of the target water area;

[0012] S5: Based on the multi-dimensional decision map of the target water area, a visualization and early warning platform is driven to realize the dynamic visualization of the four disorder problems in the target water area.

[0013] Preferably, step S1, constructing the multi-source heterogeneous dataset of the target water area, specifically includes:

[0014] By using the time synchronization protocol between the drone flight control software and the satellite data receiver, the time deviation of multi-source data acquisition is controlled within a preset threshold range;

[0015] Based on the WGS84 geographic coordinate system, coordinate transformation is performed on multi-source data. When the difference in grid scale between GIS vector data and satellite remote sensing data exceeds a preset threshold, the spatial resampling function is invoked to perform interpolation calculation. In the spatial resampling function, the interpolation algorithm type is dynamically selected according to the spectral characteristics of real-time environmental perception data.

[0016] Generate a multi-source heterogeneous dataset represented by a multi-source data fusion matrix.

[0017] Preferably, step S2, which involves dynamic environmental feature extraction from the multi-source heterogeneous dataset, specifically includes:

[0018] Based on real-time environmental perception data of light intensity and vegetation cover index, an adaptive spatial registration weight matrix is ​​constructed to register multi-source data.

[0019] A Gaussian-Laplace pyramid fusion model was used to perform multi-scale denoising on the registered multi-source data.

[0020] An anti-interference environment feature map is generated by a dynamic environment feature extraction model. The anti-interference environment feature map includes a radiometrically corrected multispectral feature layer, a dynamic mask feature layer, and a time-series change feature layer.

[0021] The data from each layer of the anti-interference environment feature map are fused to generate a feature dataset.

[0022] Preferably, the dynamic environment feature extraction model in S2 is a composite neural network architecture comprising a light intensity compensation layer, a vegetation occlusion attention mechanism layer, and a hydrological periodic feature embedding layer, wherein:

[0023] The illumination intensity compensation layer calculates the radiation correction coefficient based on the illumination intensity and performs adaptive brightness compensation for multispectral images.

[0024] The vegetation occlusion attention mechanism layer generates an occlusion area mask based on the vegetation coverage index and dynamically allocates attention weights to the feature extraction channels.

[0025] The hydrological periodic feature embedding layer encodes hydrological parameters into periodic feature vectors and performs feature-level fusion with satellite remote sensing data.

[0026] Preferably, in step S2, the reflectance of the multispectral image is used in the illumination intensity compensation layer. Calculate the radiation correction factor Specifically, it is expressed as:

[0027] ,

[0028] in, This is expressed as the calibrated light intensity. This represents the current light intensity. This is expressed as an environmental sensitivity coefficient. This is expressed as the average reflectance of the current scene;

[0029] In the vegetation occlusion attention mechanism layer, a binary mask is generated based on the vegetation cover index. ; Calculate the dynamic weights of the convolution channels in the mask region. Specifically, it is expressed as:

[0030] ,

[0031] in, Represented as the first Channel feature map, Represented as the baseline weight, It is represented as a weighting adjustment coefficient.

[0032] Preferably, step S3, which involves region segmentation and feature dimensionality reduction of the feature dataset, specifically includes:

[0033] Based on the geographic grid coding rules of the target water area, the high-resolution image is divided into block units of a preset size through a dynamic block algorithm;

[0034] Perform a sparse representation on each block unit to generate an optimized sparse basis for the discrete cosine transform;

[0035] An adaptive measurement matrix is ​​constructed based on real-time environmental perception data;

[0036] Four lightweight feature packages are generated through compressed sampling.

[0037] Preferably, in S4, the multidimensional decision map includes the risk level, disposal priority, and associated geographic grid code and quarterly timestamp corresponding to the types of problems such as illegal occupation, illegal mining, illegal dumping, and illegal construction.

[0038] Preferably, the dynamic visualization method for the four disorder problems in S5 includes a flood control capacity influence coefficient matrix, a gridded risk probability, and a two-dimensional gradient color scale.

[0039] To achieve the above objectives, the present invention also provides the following technical solution: a system for assessing the four types of disorder in water conservancy, rivers, and lakes, applicable to a method for assessing the four types of disorder in water conservancy, rivers, and lakes, comprising:

[0040] Water area data acquisition module: Real-time images of the target water area are acquired by using a drone equipped with a multispectral sensor, and satellite remote sensing data, geographic information system vector data, and real-time environmental perception data including light intensity, vegetation cover index and hydrological parameters are received simultaneously to construct a multi-source heterogeneous dataset of the target water area.

[0041] Dynamic environment feature extraction module: extracts dynamic environment features from the multi-source heterogeneous dataset to generate a feature dataset of the target water area;

[0042] Four-class feature segmentation module: A lightweight recognition engine is pre-deployed on the edge computing node to perform region segmentation and feature dimensionality reduction on the feature dataset, generating four types of lightweight feature packages;

[0043] Spatiotemporal analysis module: Performs spatiotemporal correlation analysis between the four types of lightweight feature packages and the historical governance database to generate a multidimensional decision map of the target water area;

[0044] Visualization and Early Warning Module: Based on the multi-dimensional decision map of the target water area, a visualization and early warning platform is driven to realize the dynamic visualization of the four disorder problems in the target water area.

[0045] To achieve the above objectives, the present invention also provides an electronic device, including a system central processing unit, a user information terminal, and a system operation database, characterized in that the system operation database is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the system central processing unit to implement a method for judging the four irregularities in water conservancy, rivers and lakes.

[0046] The technical effects and advantages of this invention are as follows:

[0047] 1. This invention controls the spatiotemporal deviation of UAV imagery, satellite remote sensing, and GIS data within a threshold range through a time synchronization protocol and a dynamic interpolation algorithm. It also dynamically selects the interpolation algorithm type based on real-time environmental data, thereby achieving automatic alignment and efficient fusion of multi-source data and significantly reducing the need for manual intervention and the consumption of preprocessing computing resources.

[0048] 2. This invention improves the robustness of feature extraction under complex lighting, vegetation shading, and seasonal hydrological changes by using a radiation correction coefficient and a dynamic weight allocation strategy, thereby reducing the frequency of model parameter adjustment.

[0049] 3. This invention generates risk levels, handling priorities, and geocoding for the four types of disorder by using spatiotemporal correlation analysis, providing multi-dimensional scientific basis for decision-making and optimizing governance efficiency. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating the steps of a method for assessing the four types of disorder in water conservancy, rivers, and lakes, according to an embodiment of this application.

[0051] Figure 2 This is a decision tree diagram for dynamically selecting an interpolation algorithm in a method for assessing the four types of disorder in water conservancy, rivers, and lakes, as provided in an embodiment of this application.

[0052] Figure 3 This is a structural block diagram of a system for assessing the four types of disorder in water conservancy, rivers, and lakes, according to an embodiment of this application.

[0053] Figure 4 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application.

[0054] Explanation of reference numerals in the attached drawings: 400, a schematic diagram of the structure of an electronic device; 401, a system central processing unit; 402, a communication bus; 403, a system operating database; 404, a user information terminal. 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] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to and includes any or all possible combinations of one or more of the listed items; and in the description of the embodiments of this application, unless otherwise stated, “a plurality” means two or more.

[0057] In the first aspect, please refer to Figure 1 The method shown is for assessing the four types of disorder in water conservancy, rivers, and lakes. It is applicable to the assessment system for these four types of disorder. The method includes:

[0058] S1: Acquire real-time images of the target water area by using a drone equipped with a multispectral sensor, and simultaneously receive satellite remote sensing data, geographic information system vector data, and real-time environmental perception data including light intensity, vegetation cover index, and hydrological parameters to construct a multi-source heterogeneous dataset of the target water area.

[0059] S2: Dynamically extract environmental features from the multi-source heterogeneous dataset to generate a feature dataset of the target water area;

[0060] S3: Pre-deploy a lightweight recognition engine on the edge computing node to perform region segmentation and feature dimensionality reduction on the feature dataset, generating four types of lightweight feature packages;

[0061] S4: Perform spatiotemporal correlation analysis between the four types of lightweight feature packages and the historical governance database to generate a multidimensional decision map of the target water area;

[0062] S5: Based on the multi-dimensional decision map of the target water area, a visualization and early warning platform is driven to realize the dynamic visualization of the four disorder problems in the target water area.

[0063] It should be noted that the assessment system for the four irregularities in water conservancy, rivers, and lakes can be the assessment system for the four irregularities in water conservancy, rivers, and lakes described below, or it can be an existing assessment system for the four irregularities in water conservancy, rivers, and lakes. Optionally, the assessment system for the four irregularities in water conservancy, rivers, and lakes can be an existing assessment system for the four irregularities in water conservancy, rivers, and lakes that is already built and in operation, or it can be model data of an existing assessment system for the four irregularities in water conservancy, rivers, and lakes. The assessment method shown in this embodiment can analyze the application of the assessment system for the four irregularities in water conservancy, rivers, and lakes during experiments or simulations, thereby improving the structure and operation mode of the existing assessment system for the four irregularities in water conservancy, rivers, and lakes.

[0064] Specifically, before implementing the aforementioned method, corresponding sensors, including light sensors and flow velocity sensors, can be installed in the assessment system for the four irregularities in water conservancy, rivers, and lakes. The specific installation locations can be set according to actual needs. In this embodiment, the sensors can be distributed in a grid-like manner in the assessment system for the four irregularities in water conservancy, rivers, and lakes. That is, multiple sensors are arranged at equal intervals in the target water area where the assessment system for the four irregularities in water conservancy, rivers, and lakes is applied, so as to comprehensively collect multiple structural data of water flow in space in the assessment system for the four irregularities in water conservancy, rivers, and lakes. The specific sensor distribution density can be set according to actual needs.

[0065] Specifically, in S1, the construction of the multi-source heterogeneous dataset of the target water area includes:

[0066] S101: By using the time synchronization protocol between the UAV flight control software and the satellite data receiver, the time deviation of multi-source data acquisition is controlled within a preset threshold range. The multi-source data includes real-time image data acquired by the UAV, satellite remote sensing data, geographic information system vector data, and real-time environmental perception data.

[0067] In this embodiment, a clock synchronization module based on an improved PTP is embedded in the UAV flight control software. The clock synchronization module obtains GNSS timing signals through the API interface of the satellite data receiver and establishes a hardware and software collaborative time reference system. In specific implementation, the UAV sends a time calibration request packet to the satellite data receiver every 50ms. The receiver returns a response data packet with a timestamp. By calculating the bidirectional transmission delay and dynamically compensating for network jitter error, the time deviation between the acquisition of UAV aerial images and satellite remote sensing data is stably controlled within ±30ms.

[0068] Furthermore, the bidirectional transmission delay dynamic compensation network jitter error is calculated. Specifically, it is expressed as:

[0069] ,

[0070] in, This represents the time when the drone sent the request. This represents the time the request was received by the satellite. This represents the satellite's response time. This is represented as the drone's response time.

[0071] S102: Based on the WGS84 geographic coordinate system, perform coordinate transformation on multi-source data, and when the difference in grid scale between GIS vector data and satellite remote sensing data exceeds a preset threshold, call the spatial resampling function to perform interpolation calculation;

[0072] This embodiment is based on the WGS84 geographic coordinate system. It converts the local coordinate system of UAV aerial imagery, the UTM projection coordinate system of satellite remote sensing data, and the CGCS2000 coordinate system of GIS vector data into latitude and longitude coordinates under the WGS84 ellipsoid. The spatial resolution difference detection algorithm is used to calculate the grid scale difference rate between satellite remote sensing data and GIS vector data. When the difference rate exceeds 8%, the spatial resampling function is triggered.

[0073] Furthermore, the grid scale difference rate between satellite remote sensing data and GIS vector data is calculated using a spatial resolution difference detection algorithm. Specifically, it is expressed as:

[0074] ,

[0075] in, Represented as satellite data grid size, Represented as GIS vector data grid size;

[0076] Furthermore, see Figure 2 In the spatial resampling function, the interpolation algorithm type is dynamically selected based on the spectral characteristics of the real-time environmental perception data. The selection strategy for the interpolation algorithm includes, but is not limited to, a water body area priority strategy, a vegetation cover area strategy, and a bare land area strategy. Specifically, the water body area priority strategy uses a bilinear interpolation algorithm to reduce the jagged effect at water body boundaries when the normalized water index is >0.2; the vegetation cover area strategy switches to a cubic convolution interpolation algorithm to preserve vegetation texture details when the normalized vegetation index is >0.3; and the bare land area strategy enables the nearest neighbor interpolation algorithm to reduce computational complexity when the average reflectance in the visible light band is <0.15. In this embodiment, the default strategy is bilinear interpolation, which ensures a balance between computational quality and computational efficiency.

[0077] S103: Generate a multi-source heterogeneous dataset represented by a multi-source data fusion matrix;

[0078] This embodiment uses the time of UAV image acquisition as a benchmark and applies a 5-second time sliding window compensation to the satellite remote sensing data to eliminate temporal misalignment caused by transmission delay. Simultaneously, multi-source data is encoded using 1m×1m geographic grid units, with each grid unit associated with corresponding metadata. This metadata includes RGB-IR four-channel pixel values ​​from the UAV imagery, thermal infrared radiance values ​​from the satellite remote sensing data, land use type codes from the GIS vector data, and light intensity, vegetation cover index, and hydrological parameters from the real-time environmental perception data. Finally, a spatial-spectral consistency verification algorithm is used to eliminate fusion anomalies, generating a spatiotemporally consistent multi-source data fusion matrix.

[0079] Specifically, in S2, data cleaning and spatiotemporal registration operations are performed on the multi-source heterogeneous dataset, and dynamic environmental feature extraction and fusion are realized to generate a feature dataset including;

[0080] S201: Based on the light intensity and vegetation cover index of real-time environmental perception data, an adaptive spatial registration weight matrix is ​​constructed, and a non-rigid registration of multi-source heterogeneous datasets is performed by an improved SURF algorithm, wherein the feature point matching threshold is dynamically adjusted according to the light intensity.

[0081] This embodiment is based on the light intensity from real-time environmental perception data. With vegetation cover index Construct a spatial registration weight matrix Specifically, it is expressed as:

[0082] ,

[0083] in, This represents the current light intensity. This is represented as the baseline illumination threshold. This is expressed as an adjustment coefficient. These represent the light intensity in the real-time environmental perception data. With vegetation cover index The index is set in this embodiment. , ;when At the same time, dynamically reduce the feature point matching threshold. To compensate for feature loss under low light conditions;

[0084] Furthermore, the feature point matching threshold The formula for calculating the reduction is as follows:

[0085] ,

[0086] in, This is represented as the baseline threshold. It is represented as the attenuation coefficient. This represents the current light intensity. This is represented as the baseline illumination threshold; it should be noted that the value of the baseline threshold is determined within... To achieve optimal matching accuracy; the attenuation coefficient is the rate at which the control threshold changes with light intensity;

[0087] This embodiment uses the CUDA parallel computing architecture to accelerate the matching of SURF descriptors of multi-source data, so that the average error of non-rigid registration is controlled within 1.2 pixels;

[0088] S202: The Gaussian-Laplace pyramid fusion model is used to perform multi-scale denoising on the registered multi-source data. The number of pyramid layers is adaptively selected according to the image resolution, and the fusion coefficient is generated by jointly calculating the vegetation cover index and hydrological parameters.

[0089] In this embodiment, the rule for adaptively selecting the number of pyramid layers based on image resolution is as follows: when the image resolution... At that time, a 5-layer pyramid was constructed; when At that time, a seven-layer pyramid was constructed; when At that time, a nine-layer pyramid was constructed;

[0090] In this embodiment, the hydrological parameters are calculated using flow velocity data;

[0091] Furthermore, based on the vegetation cover index With flow velocity in hydrological parameters Joint calculation to generate fusion coefficients Specifically, it is expressed as:

[0092] ,

[0093] in, and These represent the weights of vegetation cover index and flow velocity in hydrological parameters, respectively. This represents the preset maximum flow rate. This is expressed as the ratio of the current flow rate to the preset maximum flow rate, calculated using the activation function tanh.

[0094] S203: Generate an anti-interference environment feature map from the denoised data using a dynamic environment feature extraction model;

[0095] It should be noted that the dynamic environment feature extraction model is a composite neural network architecture comprising a light intensity compensation layer, a vegetation occlusion attention mechanism layer, and a hydrological periodic feature embedding layer. The light intensity compensation layer calculates the radiometric correction coefficient based on light intensity and performs adaptive brightness compensation on the multispectral image. The vegetation occlusion attention mechanism layer generates an occlusion area mask based on the vegetation cover index and dynamically allocates attention weights to the feature extraction channels. The hydrological periodic feature embedding layer encodes hydrological parameters into periodic feature vectors and performs feature-level fusion with satellite remote sensing data. The anti-interference environment feature map includes a radiometric correction multispectral feature layer, a dynamic mask feature layer, and a time-series change feature layer.

[0096] Furthermore, in the illumination intensity compensation layer, the band reflectance based on multispectral imagery... Calculate the radiation correction factor Specifically, it is expressed as:

[0097] ,

[0098] in, This is expressed as the calibrated light intensity. This represents the current light intensity. This is expressed as an environmental sensitivity coefficient. This is expressed as the average reflectance of the current scene; it should be noted that the setting... , ;

[0099] Furthermore, in the vegetation occlusion attention mechanism layer, a binary mask is generated based on the vegetation cover index. A vegetation zone is defined as having a vegetation cover index ≥ 0.3, with a value of 1; otherwise, it is defined as 0. The dynamic weights of the convolution channels in the mask region are calculated. Specifically, it is expressed as:

[0100] ,

[0101] in, Represented as the first Channel feature map, Represented as the baseline weight, This is expressed as a weighting adjustment coefficient; it should be noted that the benchmark weight is... ,in, The average vegetation cover index for the current scene is calculated from the vegetation mask region of the multi-source data fusion matrix in S1; the weight adjustment coefficient is set to... , Expressed as the number of pixels, This represents the maximum activation value of the current feature channel;

[0102] Furthermore, in the hydrological periodic feature embedding layer, hydrological parameters are encoded as periodic feature vectors. In this embodiment, the flow velocity among the hydrological parameters is used as an example. With water depth Encoding is performed, specifically as follows:

[0103] ,

[0104] in, Represented as hydrological cycle, Represented as the current timestamp; through a fully connected layer... Feature-level fusion is performed after mapping to the same dimension as the satellite remote sensing features;

[0105] S204: The data of each layer of the anti-interference environment feature map are fused in the frequency domain space through a cross-modal feature alignment algorithm to generate a feature dataset;

[0106] Specifically, feature alignment is performed in the frequency domain space by performing two-dimensional fast Fourier transforms on the radiometrically corrected multispectral feature layer, the dynamic mask feature layer, and the time-varying feature layer to obtain the amplitude spectrum and phase spectrum of each feature layer. Using the phase spectrum of the radiometrically corrected multispectral feature layer as a reference, phase alignment is performed on the dynamic mask feature layer and the time-varying feature layer. The aligned phase spectrum is combined with the original amplitude spectrum, and an inverse fast Fourier transform is performed to generate a fused feature map, i.e., the feature dataset.

[0107] Specifically, in S3, a lightweight recognition engine is pre-deployed on the edge computing node of the target water area to perform regional segmentation and feature dimensionality reduction on the feature dataset, generating a lightweight feature package containing four types of problem identifiers: illegal occupation, illegal mining, illegal dumping, and illegal construction.

[0108] In some embodiments, the steps of performing region segmentation and feature dimensionality reduction on the feature dataset transmitted in S2 to generate four types of lightweight feature packages are as follows:

[0109] S301: Based on the geographic grid coding rules of the target water area, the high-resolution image is divided into block units of a preset size through a dynamic block segmentation algorithm, and a 10% overlap area is set at the block boundary to avoid feature truncation.

[0110] It should be noted that the segmentation rules for the preset-sized blocks include: when the image resolution... At that time, the preset size of the block unit is Pixel; when At that time, the preset size of the block unit is Pixel; when At that time, the preset size of the block unit is Pixel;

[0111] Furthermore, a 10% overlap area is set at the block boundary, and the pixels in the overlap area are smoothly fused by a cosine weighting function to eliminate the block edge feature truncation effect.

[0112] S302: Perform sparse representation based on K-SVD dictionary learning on each block unit to generate an optimized sparse basis for discrete cosine transform;

[0113] Specifically, a training set is constructed by extracting 100,000 block samples from historical data on the four types of disorder stored in the system's operational database; the discrete cosine transform dictionary is initialized, and the dictionary atoms are updated through an iterative optimization algorithm to minimize the sparse representation error, while constraining the number of non-zero elements in the sparse coefficients to not exceed a preset threshold; the optimized discrete cosine transform sparse basis is output; sparse coding is performed on the real-time block data to solve for the coefficient vector that satisfies the minimum reconstruction error and sparsity constraints, and the orthogonal matching pursuit algorithm is used to accelerate the calculation process.

[0114] S303: Construct an adaptive measurement matrix based on the vegetation cover index and hydrological parameters in the real-time environmental perception data;

[0115] It should be noted that the adaptive measurement matrix is ​​dynamically generated based on real-time environmental perception data. Specifically, when the vegetation cover index is not lower than 0.3, the sampling rate is increased from the baseline of 20% to 30% by increasing the number of rows in the adaptive measurement matrix, and a partially orthogonal matrix structure is used to ensure sampling stability. When the flow velocity reaches or exceeds 0.5 m / s, the row vectors of the adaptive measurement matrix are orthogonalized to reduce the matrix condition number and enhance the reconstruction quality. Based on the spatial distribution ratio of vegetation and water areas, the measurement matrices corresponding to different areas are dynamically fused to generate the final environmental adaptive measurement matrix.

[0116] S304: Generate four types of lightweight feature packages using a compressed sampling formula. The four types of lightweight feature packages include lightweight feature packages for the problems of haphazard occupation, haphazard sampling, haphazard stacking, and haphazard construction. Each lightweight feature package is associated with the probability distribution vectors of the problems of haphazard occupation, haphazard sampling, haphazard stacking, and haphazard construction.

[0117] Specifically, in S4, the steps of spatiotemporal correlation analysis and multidimensional decision graph generation include:

[0118] S401: Divide the historical governance data of different watersheds stored in the system operation database into source domain and target domain. The source domain data contains complete annotations of the four types of disorder, while the target domain data is a lightweight feature package of the current watershed.

[0119] The source domain data in this embodiment comes from three adjacent watersheds that have completed remediation, including historical annotations of the four types of pollution problems and remediation effectiveness scores, spanning from 2015 to 2024; the target domain data consists of four lightweight feature packages of the target water area.

[0120] S402: The historical scores of source domain governance are mapped to the probability space of the target domain through a linear weighting strategy, and the distributions of the source domain and the target domain in the probability space are aligned through the KL divergence loss function.

[0121] In this embodiment, the difference between the feature distributions of the source domain and the target domain is calculated, and the probability density curves of the features of the two domains are fitted by kernel density estimation to minimize the difference in information entropy between the two curves.

[0122] S403: Obtain the risk weight matrix by passing the probability space of the target domain through a temporal convolutional network;

[0123] This embodiment receives the aligned feature vector and the encoded hydrological parameters, and generates a 4×4 risk weight matrix by mapping it in the fully connected layer through a temporal convolutional network. The elements of the risk weight matrix represent the risk level of the four types of disorder in the four quarters.

[0124] S404: Weight the probability distribution vector of the four types of illegal occupation, illegal mining, illegal dumping and illegal construction in the target water area with the risk weight matrix according to the quarterly dimension to generate a multi-dimensional decision map. The multi-dimensional decision map includes the risk level, disposal priority and associated geographic grid code and quarterly timestamp corresponding to the types of illegal occupation, illegal mining, illegal dumping and illegal construction.

[0125] Specifically, in S5, the dynamic visualization method for the four types of disorder includes a flood control capacity influence coefficient matrix, a gridded risk probability, and a two-dimensional gradient color scale. The implementation steps for the dynamic visualization of the four types of disorder are as follows:

[0126] S501: Calculate the flow velocity distribution and water level changes in the cross section of the river based on the spatial distribution of the problems of random dumping and construction, and generate the flood discharge capacity influence coefficient matrix;

[0127] This embodiment uses a two-dimensional shallow water equation to describe the river flow movement, and models the problem of haphazard dumping and construction as the cross-sectional blockage rate. The haphazard dumping area is calculated by the haphazard dumping mask coverage rate of UAV imagery, and the haphazard construction area is determined based on the proportion of the projected area of ​​the buildings. The flood discharge capacity influence coefficient is calculated based on the shallow water equation solver of the finite volume method, and then the flood discharge capacity influence coefficient matrix is ​​obtained. The value range of the elements of the flood discharge capacity influence coefficient matrix is ​​[0, 1], where 1 represents complete blockage.

[0128] S502: Construct a spatiotemporal heat map to simulate the risk diffusion path of the four disorder problems within a preset governance cycle and calculate the gridded risk probability value;

[0129] In some embodiments, an improved random walk algorithm is used to simulate the diffusion path of the four disorder problems, and 1000 random walk paths are iteratively calculated to statistically analyze the gridded risk probability value; the risk probability value is mapped to a 256-level grayscale gradient to generate a spatiotemporal heatmap; when a new four disorder problem detection result is input, the heatmap is updated every 5 minutes.

[0130] S503: Based on the flood discharge capacity impact coefficient and risk probability value, a two-dimensional gradient color mark is generated by mapping the HSV color space, where red corresponds to high-risk areas;

[0131] In some embodiments, the flood discharge capacity impact dimension is represented by a gradient from blue to purple, and the risk probability dimension is represented by a gradient from green to red. The two-dimensional data are fused using the HSV color space to display the two-dimensional gradient color scale.

[0132] Secondly, this embodiment also discloses a system for assessing the problems of disorder in water conservancy, rivers, and lakes, referring to... Figure 3The system, applicable to the method described in the first aspect, comprises:

[0133] Water area data acquisition module: Real-time images of the target water area are acquired by using a drone equipped with a multispectral sensor, and satellite remote sensing data, geographic information system vector data, and real-time environmental perception data including light intensity, vegetation cover index and hydrological parameters are received simultaneously to construct a multi-source heterogeneous dataset of the target water area.

[0134] Dynamic environment feature extraction module: extracts dynamic environment features from the multi-source heterogeneous dataset to generate a feature dataset of the target water area;

[0135] Four-class feature segmentation module: A lightweight recognition engine is pre-deployed on the edge computing node to perform region segmentation and feature dimensionality reduction on the feature dataset, generating four types of lightweight feature packages;

[0136] Spatiotemporal analysis module: Performs spatiotemporal correlation analysis between the four types of lightweight feature packages and the historical governance database to generate a multidimensional decision map of the target water area;

[0137] Visualization and Early Warning Module: Based on the multi-dimensional decision map of the target water area, a visualization and early warning platform is driven to realize the dynamic visualization of the four disorder problems in the target water area.

[0138] In a third aspect, this embodiment also discloses an electronic device, referring to... Figure 4 The electronic device may include: at least one system central processing unit 401, at least one communication bus 402, at least one system operation database 403, and user information terminal 404; the system operation database is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the system central processing unit to implement the method described in the first aspect.

[0139] It should be noted that the system's central processing unit 401 is the core computing and control unit of the entire system for assessing the four irregularities in water conservancy, rivers, and lakes. It includes one or more processing cores, which are connected to various parts of the system through various interfaces and lines. By running or executing instructions, programs, code sets, or instruction sets stored in the system's operating database, and by calling the data stored therein, it performs various functions for assessing the four irregularities in water conservancy, rivers, and lakes, including real-time monitoring of river and lake shorelines, intelligent identification of illegal and irregular activities, and analysis and statistics of relevant data.

[0140] The communication bus 402 is used to realize the connection and communication between components.

[0141] The system's operational database 403 is used to store a large amount of data related to the assessment of the four irregularities in water conservancy, rivers, and lakes, including geographical information of rivers and lakes, water level and flow data, and a large amount of historical operational data, including monitoring data and processing results data for different time periods. When the system's central processing unit performs various functions, it will frequently call these data from the system's operational database for operations such as data comparison, model training, and decision making, thereby achieving precise control and efficient management of the assessment of the four irregularities in water conservancy, rivers, and lakes.

[0142] Among them, the user information terminal 404 is an information output device for receiving information from the water conservancy, river and lake four-fold disorder problem analysis system. It provides users with an interface to interact with the system by connecting to external devices such as displays and cameras through standard wired or wireless interfaces.

[0143] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0144] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for assessing the four irregularities in water conservancy, rivers, and lakes, including: S1: Acquire real-time images of the target water area by using a drone equipped with a multispectral sensor, and simultaneously receive satellite remote sensing data, geographic information system vector data, and real-time environmental perception data including light intensity, vegetation cover index, and hydrological parameters to construct a multi-source heterogeneous dataset of the target water area. S2: Dynamic environmental feature extraction is performed on the multi-source heterogeneous dataset to generate a feature dataset of the target water area: an anti-interference environmental feature map is generated through a dynamic environmental feature extraction model. The anti-interference environmental feature map includes a radiometrically corrected multispectral feature layer, a dynamic mask feature layer, and a time-series change feature layer. The data from each layer of the anti-interference environment feature map are fused to generate a feature dataset; The dynamic environmental feature extraction model is a composite neural network architecture comprising a light intensity compensation layer, a vegetation occlusion attention mechanism layer, and a hydrological periodic feature embedding layer. Specifically: the light intensity compensation layer calculates radiometric correction coefficients based on light intensity to adaptively compensate for the brightness of multispectral images; the vegetation occlusion attention mechanism layer generates occlusion area masks based on vegetation cover index and dynamically allocates attention weights to feature extraction channels; and the hydrological periodic feature embedding layer encodes hydrological parameters into periodic feature vectors and performs feature-level fusion with satellite remote sensing data. S3: Pre-deploy a lightweight recognition engine on the edge computing node to perform region segmentation and feature dimensionality reduction on the feature dataset, generating four types of lightweight feature packages; S4: Perform spatiotemporal correlation analysis between the four types of lightweight feature packages and the historical governance database to generate a multidimensional decision map of the target water area; wherein, the spatiotemporal correlation analysis includes: S401: Divide the historical governance data of different watersheds stored in the system operation database into source domain and target domain. The source domain data contains complete annotations of the four types of disorder, while the target domain data is a lightweight feature package of the current watershed. S402: The historical scores of source domain governance are mapped to the probability space of the target domain through a linear weighting strategy, and the distributions of the source domain and the target domain in the probability space are aligned through the KL divergence loss function. S403: Obtain the risk weight matrix by passing the probability space of the target domain through a temporal convolutional network; S404: The probability distribution vector of the four types of illegal occupation, illegal mining, illegal dumping and illegal construction in the target water area is weighted by the risk weight matrix according to the quarterly dimension to generate a multi-dimensional decision map. The multi-dimensional decision map includes the risk level, disposal priority and associated geographic grid code and quarterly timestamp corresponding to the types of illegal occupation, illegal mining, illegal dumping and illegal construction. S5: Based on the multi-dimensional decision map of the target water area, a visualization and early warning platform is driven to realize the dynamic visualization of the four disorder problems in the target water area.

2. The method for assessing the four irregularities in water conservancy, rivers, and lakes according to claim 1, characterized in that: S1, constructing the multi-source heterogeneous dataset of the target water area, specifically includes: By using the time synchronization protocol between the drone flight control software and the satellite data receiver, the time deviation of multi-source data acquisition is controlled within a preset threshold range; Based on the WGS84 geographic coordinate system, coordinate transformation is performed on multi-source data. When the difference in grid scale between GIS vector data and satellite remote sensing data exceeds a preset threshold, the spatial resampling function is invoked to perform interpolation calculation. In the spatial resampling function, the interpolation algorithm type is dynamically selected according to the spectral characteristics of real-time environmental perception data. Generate a multi-source heterogeneous dataset represented by a multi-source data fusion matrix.

3. The method for assessing the four irregularities in water conservancy, rivers, and lakes according to claim 1, characterized in that: S2, which involves dynamic environmental feature extraction from the multi-source heterogeneous dataset, specifically includes: Based on real-time environmental perception data of light intensity and vegetation cover index, an adaptive spatial registration weight matrix is ​​constructed to register multi-source data. A Gaussian-Laplace pyramid fusion model is used to perform multi-scale denoising on the registered multi-source data.

4. The method for assessing the four irregularities in water conservancy, rivers, and lakes according to claim 3, characterized in that: S2, in the illumination intensity compensation layer, is based on the band reflectance of the multispectral image. Calculate the radiation correction factor Specifically, it is expressed as: ， in, This is expressed as the calibrated light intensity. This represents the current light intensity. This is expressed as an environmental sensitivity coefficient. This is expressed as the average reflectance of the current scene; In the vegetation occlusion attention mechanism layer, a binary mask is generated based on the vegetation cover index. ; Calculate the dynamic weights of the convolution channels in the mask region. Specifically, it is expressed as: ， in, Represented as the first Channel feature map, Represented as the baseline weight, It is represented as a weighting adjustment coefficient.

5. The method for assessing the four irregularities in water conservancy, rivers, and lakes according to claim 1, characterized in that: S3, which performs region segmentation and feature dimensionality reduction on the feature dataset, specifically includes: Based on the geographic grid coding rules of the target water area, the high-resolution image is divided into block units of a preset size through a dynamic block algorithm; Perform a sparse representation on each block unit to generate an optimized sparse basis for the discrete cosine transform; An adaptive measurement matrix is ​​constructed based on real-time environmental perception data; Four lightweight feature packages are generated through compressed sampling.

6. The method for assessing the four irregularities in water conservancy, rivers, and lakes according to claim 1, characterized in that: The S5, the dynamic visualization method for the four disorder problems, includes the flood control capacity influence coefficient matrix, gridded risk probability, and two-dimensional gradient color scale.

7. A system for assessing the four irregularities in water conservancy, rivers, and lakes, characterized in that: A method for assessing the four types of disorder in water conservancy, rivers, and lakes, applicable to any one of claims 1-6, includes: Water area data acquisition module: Real-time images of the target water area are acquired by using a drone equipped with a multispectral sensor, and satellite remote sensing data, geographic information system vector data, and real-time environmental perception data including light intensity, vegetation cover index and hydrological parameters are received simultaneously to construct a multi-source heterogeneous dataset of the target water area. Dynamic environment feature extraction module: extracts dynamic environment features from the multi-source heterogeneous dataset to generate a feature dataset of the target water area; Four-class feature segmentation module: A lightweight recognition engine is pre-deployed on the edge computing node to perform region segmentation and feature dimensionality reduction on the feature dataset, generating four types of lightweight feature packages; Spatiotemporal analysis module: Performs spatiotemporal correlation analysis between the four types of lightweight feature packages and the historical governance database to generate a multidimensional decision map of the target water area; Visualization and Early Warning Module: Based on the multi-dimensional decision map of the target water area, a visualization and early warning platform is driven to realize the dynamic visualization of the four disorder problems in the target water area.

8. An electronic device, comprising a system central processing unit, a user information terminal, and a system operating database, characterized in that, The system runtime database is used to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the system central processing unit to implement the method as described in any one of claims 1 to 6.

Images