Desert photovoltaic sand damage dynamic monitoring method and system based on unmanned aerial vehicle

Data is obtained through the multimodal sensor of the drone, sand and dust feature matrix analysis and three-dimensional dynamic sand and dust concentration field modeling are solved, and sand and dust damage monitoring and prediction are achieved efficient sand and dust damage dynamic monitoring and prediction.

CN120352306AInactive Publication Date: 2025-07-22INNER MONGOLIA AGRICULTURAL UNIVERSITY
View PDF 0 Cites 5 Cited by

Patent Information

Application Number
CN202510487995.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-04-18
Publication Date
2025-07-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing technology is difficult to achieve dynamic monitoring of sand damage in desert photovoltaic power plants, resulting in lag in monitoring data and being unable to deal with sand and dust erosion in a timely manner, affecting power generation efficiency and operation and maintenance costs.

Method used

The drone is equipped with multimodal sensors to obtain dust data in the photovoltaic power station area, perform multimodal data fusion processing, generate dust characteristic matrix, calculate the migration probability distribution of dust particles, and build a three-dimensional dynamic sand and dust concentration field model to output dynamic sand and damage monitoring results.

Benefits of technology

It has realized high-time dynamic monitoring of desert photovoltaic power stations, improved operation and maintenance efficiency and prevention and control capabilities, and optimized operation and maintenance decision-making efficiency and resource scheduling.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120352306A_ABST
    Figure CN120352306A_ABST
Patent Text Reader

Abstract

The invention provides a desert photovoltaic sand damage dynamic monitoring method and system based on an unmanned aerial vehicle, and the method comprises the steps: obtaining the flying dust data of a photovoltaic power station region, generating an initial data set, and enabling the flying dust data to be collected through a multi-mode sensor carried by the unmanned aerial vehicle; performing multi-modal data fusion processing on the initial data set to obtain a dust feature matrix, the dust feature matrix being used for representing dust reflectivity, terrain elevation and wind speed correlation features; calculating migration probability distribution of the flying dust particles based on the sand and dust characteristic matrix, wherein the migration probability distribution is used for representing spatio-temporal evolution laws of the particles with different particle sizes; a three-dimensional dynamic sand and dust concentration field model is generated according to the migration probability distribution, a sand damage dynamic monitoring result is output based on the three-dimensional dynamic sand and dust concentration field model, and the sand damage dynamic monitoring result comprises deposition density distribution and an erosion risk area. By adopting the method, the photovoltaic sand damage of the desert can be accurately and dynamically monitored with high timeliness.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of desert environmental monitoring, and particularly relates to a dynamic monitoring method and system for desert photovoltaic sand hazards based on unmanned aerial vehicles (UAVs). Background Art

[0002] The application of UAVs in the field of sand hazard monitoring in desert photovoltaic power stations is increasingly becoming an important research direction in the new energy industry. With the rapid growth of the global demand for renewable energy, desert areas have become ideal sites for the construction of photovoltaic power stations due to their rich solar energy resources. However, frequent sand and dust weather and severe dust erosion not only threaten the power generation efficiency of photovoltaic panels but also increase the operation and maintenance costs. How to achieve dynamic monitoring and precise response to sand hazards has become a key issue that urgently needs to be broken through in this field. The importance of this research lies in that it is directly related to the long-term stable operation and economic benefit optimization of photovoltaic power stations, and has significant practical significance and application value.

[0003] Currently, traditional sand hazard monitoring methods mostly rely on fixed sensors or manual inspections, and these methods have obvious deficiencies in terms of coverage, real-time performance, and data accuracy. Fixed sensors are limited by the layout density and cannot comprehensively reflect the large-scale sand and dust distribution. Manual inspections are inefficient and difficult to adapt to the harsh desert environment, resulting in lagging monitoring data and making it difficult to support timely decision-making. These limitations make the existing solutions often appear powerless when facing the dynamically changing dust problems.

[0004] The current dynamic monitoring methods for desert photovoltaic sand hazards have the problem of single data type used, making it difficult to accurately and timely conduct dynamic monitoring of sand hazards. Summary of the Invention

[0005] Based on this, it is necessary to provide a dynamic monitoring method and system for desert photovoltaic sand hazards based on UAVs in view of the above technical problems, which can accurately and highly timely conduct dynamic monitoring of sand hazards through UAV remote sensing data, contribute to realizing dynamic and precise monitoring and evolution prediction of desert photovoltaic sand hazards, and effectively improve the operation and maintenance efficiency of photovoltaic power stations and the sand hazard prevention and control ability.

[0006] In the first aspect, the present application provides a dynamic monitoring method for desert photovoltaic sand hazards based on UAVs, including:

[0007] Obtain the dust data of the photovoltaic power station area to generate an initial data set, where the dust data is collected by a multi-modal sensor carried by a UAV;

[0008] Perform multi-modal data fusion processing on the initial data set to obtain a sand and dust feature matrix, which is used to characterize the reflectivity, terrain elevation, and wind speed correlation characteristics of the dust;

[0009] Calculate the migration probability distribution of dust particles based on the dust characteristic matrix, and the migration probability distribution is used to characterize the spatio-temporal evolution law of particles with different particle sizes;

[0010] Generate a three-dimensional dynamic dust concentration field model according to the migration probability distribution, and output the dynamic sand hazard monitoring results based on the three-dimensional dynamic dust concentration field model. The dynamic sand hazard monitoring results include the sediment density distribution and the erosion risk area.

[0011] In a possible embodiment, obtain the dust data in the photovoltaic power station area to generate an initial data set, including:

[0012] Obtain a first instruction, which is used to instruct the unmanned aerial vehicle to fly along a dynamic grid path, and the flight height is adaptively adjusted based on real-time wind speed data;

[0013] Control the unmanned aerial vehicle to collect data according to the first instruction, synchronously collect multi-spectral reflectance data and lidar point cloud data through a multi-modal sensor, and generate an original sensing data set with spatio-temporal marks;

[0014] Perform coordinate alignment and time synchronization processing on the original sensing data set to generate an initial data set.

[0015] In a possible embodiment, perform multi-modal data fusion processing on the initial data set to obtain a dust characteristic matrix, including:

[0016] Perform adaptive median filtering processing on the multi-spectral reflectance data to obtain noise-reduced reflectance data;

[0017] Perform spatial interpolation processing based on the semi-variogram on the lidar point cloud data to generate a terrain point cloud;

[0018] Fuse the noise-reduced reflectance data and the terrain point cloud through an attention mechanism to generate a dust characteristic matrix.

[0019] In a possible embodiment, calculate the migration probability distribution of dust particles based on the dust characteristic matrix, including:

[0020] Based on the dust characteristic matrix, construct a dynamic dust field model, and the dynamic dust field model calculates the particle migration probability density by coupling the wind speed vector and the diffusion coefficient;

[0021] Perform parameter optimization processing on the dynamic dust field model, and use the particle swarm optimization algorithm to iteratively obtain a spatial evolution gradient field, which is used to characterize the diffusion rate and direction of dust particles;

[0022] Perform Monte Carlo trajectory simulation processing based on the spatial evolution gradient field to generate a migration probability distribution, and the migration probability distribution contains a set of spatio-temporal evolution trajectories of particles with different particle sizes.

[0023] In a possible embodiment, a three-dimensional dynamic dust concentration field model is generated according to the migration probability distribution, and dynamic sand hazard monitoring results are output based on the three-dimensional dynamic dust concentration field model, including:

[0024] Performing three-dimensional reconstruction processing on the migration probability distribution based on the marching cubes algorithm to generate a three-dimensional dynamic dust concentration field model;

[0025] Performing vertical direction exponential decay integral processing on the three-dimensional dynamic dust concentration field model to calculate the deposition density distribution;

[0026] Performing binary segmentation processing on the deposition density distribution according to a preset threshold to mark the erosion risk area.

[0027] In a possible embodiment, parameter optimization processing is performed on the dynamic dust field model, and a spatial evolution gradient field is obtained by iterative use of the particle swarm optimization algorithm, including:

[0028] Based on the diffusion coefficient and the wind speed vector, a fitness function of the particle swarm optimization algorithm is defined, and the fitness function is used to quantify the matching degree between the gradient field and the measured data;

[0029] Performing genetic crossover operation on the particle swarm optimization algorithm, setting the crossover probability to a first preset value to generate an initial optimized parameter set;

[0030] Performing mutation operation on the initial optimized parameter set, setting the mutation probability to a second preset value to generate an optimized spatial evolution gradient field;

[0031] Iteratively solve the dynamic dust field model according to the updated optimized parameter set, and output the optimized spatial evolution gradient field.

[0032] In a possible embodiment, performing vertical direction exponential decay integral processing on the three-dimensional dynamic dust concentration field model to calculate the deposition density distribution, including:

[0033] Performing Gaussian smoothing filtering processing on the three-dimensional dynamic dust concentration field model to eliminate high-frequency noise interference and generate a denoised concentration field model;

[0034] Performing adaptive step integral processing on the denoised concentration field model to generate an intermediate deposition density distribution;

[0035] Performing normalization processing on the intermediate deposition density distribution to generate the deposition density distribution.

[0036] In a second aspect, the present application further provides a dynamic sand hazard monitoring system for desert photovoltaic based on an unmanned aerial vehicle, including:

[0037] A data acquisition module for acquiring dust emission data in the photovoltaic power station area to generate an initial data set, and the dust emission data is collected by a multi-modal sensor carried by an unmanned aerial vehicle;

[0038] A data fusion module, configured to perform multi-modal data fusion processing on an initial data set to obtain a dust characteristic matrix, where the dust characteristic matrix is used to characterize the reflectivity, terrain elevation, and wind speed correlation characteristics of dust storms;

[0039] A migration probability calculation module, configured to calculate the migration probability distribution of dust particles based on the dust characteristic matrix, where the migration probability distribution is used to characterize the spatio-temporal evolution law of particles with different particle sizes;

[0040] A dynamic monitoring module, configured to generate a three-dimensional dynamic dust concentration field model according to the migration probability distribution, and output a dynamic sand hazard monitoring result based on the three-dimensional dynamic dust concentration field model, where the dynamic sand hazard monitoring result includes sediment density distribution and erosion risk areas.

[0041] In a third aspect, the present application further provides a computer device, including a memory and a processor, where the memory stores a computer program, and when the processor executes the computer program, the above-mentioned dynamic monitoring method for desert photovoltaic sand hazards based on unmanned aerial vehicles is implemented.

[0042] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the above-mentioned dynamic monitoring method for desert photovoltaic sand hazards based on unmanned aerial vehicles is implemented.

[0043] For the above-mentioned dynamic monitoring method and system for desert photovoltaic sand hazards based on unmanned aerial vehicles, dust storm data in the photovoltaic power station area is obtained through multi-modal sensors carried by the unmanned aerial vehicle to generate an initial data set; multi-modal data fusion processing is performed on the initial data set to obtain a dust characteristic matrix that can characterize the reflectivity, terrain elevation, and wind speed correlation characteristics of dust storms; based on this dust characteristic matrix, the migration probability distribution of dust particles is calculated, so as to characterize the spatio-temporal evolution law of particles with different particle sizes; a three-dimensional dynamic dust concentration field model is generated according to the migration probability distribution, and a dynamic sand hazard monitoring result including sediment density distribution and erosion risk areas is output based on this model. The above technical solution can significantly improve the timeliness while ensuring the monitoring accuracy through unmanned aerial vehicle remote sensing data fusion and dynamic modeling technology, providing a scientific basis for the prediction and active prevention and control of sand hazard evolution in desert photovoltaic power stations, and effectively optimizing the operation and maintenance decision-making efficiency and resource scheduling ability. Description of the Drawings

[0044] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the following will briefly introduce the drawings required for use in the description of the embodiments or related technologies. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0045] Figure 1 Schematic flow chart of a dynamic monitoring method for desert photovoltaic sand hazards based on an unmanned aerial vehicle provided by an embodiment of the present invention;

[0046] Figure 2 Schematic structural diagram of a dynamic monitoring system for desert photovoltaic sand hazards based on an unmanned aerial vehicle provided by an embodiment of the present invention. Specific implementation manners

[0047] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0048] First, a brief introduction is made to the nouns involved in the embodiments of the present application.

[0049] Multispectral reflectance data refers to the quantitative description of the reflection ability of an object to incident light within the spectral ranges of multiple specific bands. By measuring the reflectance of an object at different wavelengths (such as visible light, near-infrared, etc.), its unique spectral characteristics can be obtained, which can be used to distinguish material components, monitor environmental changes, and evaluate ecological conditions, etc. For example, in the agricultural field, multispectral reflectance data can be used to analyze the growth status and health of vegetation; in environmental monitoring, it can be used to identify soil types and detect water pollution, etc. Due to its multi-band and high-resolution characteristics, it can provide rich information and provide important data support for scientific research and practical applications.

[0050] LiDAR point cloud data is a three-dimensional spatial data set obtained through LiDAR (Light Detection and Ranging) technology. It records the time interval or phase difference between the laser transmitter emitting a laser pulse and the receiver receiving the reflected pulse, so as to calculate the accurate distance information of the surface of the target object. Each measurement point returned by a laser pulse contains three-dimensional coordinates (X, Y, Z), and possibly intensity information (related to the material properties of the reflected laser). Point cloud data can present the three-dimensional structure of complex terrains, buildings, vegetation and other objects in a high-precision and high-density form, and is widely used in terrain mapping, urban modeling, autonomous driving, forestry resource surveys and other fields, providing a powerful data basis for accurate spatial analysis and three-dimensional visualization.

[0051] The Particle Swarm Optimization (PSO) is an optimization algorithm based on swarm intelligence, which simulates the behaviors of bird flocks hunting or fish schools foraging in nature. In this algorithm, each particle represents a potential solution in the solution space and has attributes of position and velocity. The particles dynamically adjust their positions and velocities by tracking the individual best solution (the best position experienced by itself) and the global best solution (the best position experienced by all particles in the swarm). Through the cooperation and information sharing among particles, the swarm can gradually search for the global best solution or a solution close to the global best solution. Due to its characteristics such as simple implementation, fast convergence speed, and good adaptability to complex optimization problems, the PSO algorithm is widely applied in many fields such as function optimization, machine learning, and engineering design.

[0052] Monte Carlo trajectory simulation is a numerical calculation method based on random sampling and probability statistics, which is used to simulate the movement trajectories of particles (such as photons, electrons, dust particles, etc.) in a complex environment. By randomly generating the initial states (such as positions, velocities, directions, etc.) of a large number of particles and gradually calculating the movement paths of each particle at different time steps according to physical laws and probability distributions, and finally statistically analyzing the trajectory set of all particles, thus predicting the macroscopic behaviors and distribution laws of the particle swarm. It is applicable to dealing with complex many-body systems and random processes. For example, in atmospheric science, it simulates the transmission paths of dust particles, in optics, it simulates the propagation and scattering of photons in a medium, and in the financial field, it simulates the fluctuations of asset prices, etc. It can effectively solve complex problems that are difficult to handle by traditional analytical methods and provides a powerful tool for scientific research and engineering applications.

[0053] According to the above noun explanations, the implementation environment of the dynamic desert photovoltaic sand hazard monitoring method based on drones provided by the embodiments of this application is described. Schematically, the implementation environment includes: multimodal sensors, terminals, and processors. Among them, the processor is connected to the multimodal sensors and the terminals through network signals; the multimodal sensors include but are not limited to optical sensors, lidars, meteorological sensors, or gas sensors; the processor can be a central processing unit, a multi-core processor, an artificial intelligence chip, etc., which is not limited here.

[0054] Combined with the above noun explanations and the implementation environment, the application scenarios of the embodiments of this application are described. The dynamic desert photovoltaic sand hazard monitoring method based on drones provided in the embodiments of this application can be applied to the following scenarios including but not limited to:

[0055] In a photovoltaic power station in a desert area, sandstorms and sand activities are frequent, and sand hazards have become an important factor affecting the performance of photovoltaic modules and the operation efficiency of the power station. The above method can obtain the dust data in the area of the photovoltaic power station in real time, including information such as the concentration, particle size distribution, and migration path of sand and dust. Through multimodal data fusion and the construction of a dynamic sand and dust field model, this method can accurately predict the deposition area and erosion risk points of sand and dust in the power station, providing a scientific basis for the cleaning and protection measures of photovoltaic modules, thereby effectively extending the service life of photovoltaic modules and improving the power generation efficiency.

[0056] For environmental science research in desert areas, understanding the dynamic change process of sand and dust is an important part of studying the desert ecosystem and climate change. The above method can provide high-precision and high-timeliness sand and dust dynamic data, helping scientists study the origin, transmission path, and settlement law of sand and dust. For example, by analyzing the migration probability distribution of sand and dust particles of different particle sizes, the transmission mechanism of sand and dust in the atmosphere can be revealed, providing basic data for studying the impact of sand and dust on regional climate, air quality, and the ecosystem, and promoting the development of desert environmental science research.

[0057] Schematically, the dynamic monitoring method of desert photovoltaic sand hazards based on drones provided in the embodiments of the present application can also be applied to other application scenarios. Only examples are given here, and the specific application scenarios are not limited.

[0058] In an exemplary embodiment, as Figure 1 shown, a dynamic monitoring method of desert photovoltaic sand hazards based on drones is provided. In this embodiment, taking the application of this method to the terminal in the foregoing implementation environment as an example, it can be understood that this method can also be applied to a server, and can also be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. In this embodiment, the method includes the following steps 101 to 104:

[0059] Step 101, obtain the dust data in the area of the photovoltaic power station and generate an initial data set. The dust data is collected by a multimodal sensor carried by a drone.

[0060] Exemplarily, a drone carries a multimodal sensor into the area of the photovoltaic power station to collect dust data. The sensor can include an optical sensor for obtaining the reflectivity data of the dust, a lidar sensor for measuring the terrain elevation, and a meteorological sensor for recording information such as wind speed. The remote sensing data obtained by the drone is transmitted and stored in real time, and after spatio-temporal alignment processing, an initial data set including spectral bands, elevation coordinates, and wind speed parameters is generated, providing a multi-source data basis for subsequent analysis.

[0061] Step 102, perform multimodal data fusion processing on the initial data set to obtain a dust characteristic matrix, which is used to characterize the reflectivity, terrain elevation, and wind speed correlation characteristics of dust storms.

[0062] Specifically, perform multimodal data fusion processing on the generated initial data set. Integrate the dust storm reflectivity, terrain elevation, and wind speed data collected by different sensors through an algorithm to eliminate redundancy and errors between the data, and generate a dust characteristic matrix. This matrix can comprehensively characterize the multi-dimensional characteristics of dust storms and provide an accurate data model for subsequent analysis.

[0063] Step 103, calculate the migration probability distribution of dust particles based on the dust characteristic matrix, and the migration probability distribution is used to characterize the spatio-temporal evolution law of particles with different particle sizes.

[0064] Specifically, based on the dust characteristic matrix, use a mathematical model to calculate the migration probability distribution of dust particles. This distribution can characterize the spatio-temporal evolution law of dust particles with different particle sizes in the photovoltaic power station area, that is, the migration paths and probabilities of particles with different particle sizes at different positions and times, thus providing a key basis for dynamic dust simulation.

[0065] Step 104, generate a three-dimensional dynamic dust concentration field model according to the migration probability distribution, and output the dynamic sand hazard monitoring results based on the three-dimensional dynamic dust concentration field model. The dynamic sand hazard monitoring results include sediment density distribution and erosion risk areas.

[0066] Specifically, according to the calculated migration probability distribution, construct a three-dimensional dynamic dust concentration field model, which can intuitively display the concentration distribution of dust in the photovoltaic power station area and its dynamic change process. Based on this model, output the dynamic sand hazard monitoring results, including sediment density distribution and erosion risk areas, to provide a scientific basis for the operation and maintenance management of photovoltaic power stations.

[0067] The above dynamic monitoring method and system for desert photovoltaic sand hazards based on unmanned aerial vehicles obtain the dust data of the photovoltaic power station area through the multimodal sensors carried by the unmanned aerial vehicle, and generate an initial data set; perform multimodal data fusion processing on the initial data set to obtain a dust characteristic matrix that can characterize the reflectivity, terrain elevation, and wind speed correlation characteristics of dust storms; calculate the migration probability distribution of dust particles based on this dust characteristic matrix, so as to characterize the spatio-temporal evolution law of particles with different particle sizes; generate a three-dimensional dynamic dust concentration field model according to the migration probability distribution, and output the dynamic sand hazard monitoring results including sediment density distribution and erosion risk areas based on this model. The above technical solution can significantly improve the timeliness while ensuring the monitoring accuracy through unmanned aerial vehicle remote sensing data fusion and dynamic modeling technology, provide a scientific basis for the prediction and active prevention and control of sand hazard evolution in desert photovoltaic power stations, and effectively optimize the operation and maintenance decision-making efficiency and resource scheduling ability.

[0068] In a possible embodiment, obtaining the dust data of the photovoltaic power station area and generating an initial data set may include:

[0069] Step 201, obtaining a first instruction for instructing the drone to fly along a dynamic grid path, and the flight height is adaptively adjusted based on real-time wind speed data.

[0070] Exemplarily, the terminal generates a dynamic grid flight path according to the geographical coordinates of the photovoltaic power station, and the environmental wind speed is monitored in real time by a wind speed sensor. When the wind speed exceeds the preset threshold, the path planning algorithm automatically increases the flight height of the drone to ensure that the sensor can still stably collect data in a strong wind environment; the first instruction may include the coordinates of the dynamically adjusted flight path, height parameters, and the start / stop timing of the sensor, and is sent to the drone flight control system through wireless communication. The flight height of the drone is dynamically adjusted through real-time wind speed feedback to avoid data collection failure caused by strong wind disturbance. The dynamic grid path planning ensures that the sensor covers the entire photovoltaic power station area, eliminates the monitoring blind spots caused by traditional fixed flight routes, and improves data integrity.

[0071] Step 202, controlling the drone to perform data collection according to the first instruction, and synchronously collecting multi-spectral reflectance data and lidar point cloud data through a multi-modal sensor to generate an original sensing data set with spatio-temporal marks.

[0072] Specifically, a hardware-level synchronous triggering mechanism is adopted to ensure that the acquisition timing error between the multi-spectral sensor and the lidar is less than 1 millisecond. The spatio-temporal marking technology provides an accurate alignment basis for subsequent data fusion and reduces the cross-modal registration error.

[0073] Step 203, performing coordinate alignment and time synchronization processing on the original sensing data set to generate an initial data set.

[0074] Specifically, the pixel coordinate system of the multi-spectral data can be aligned with the world coordinate system of the lidar point cloud, and interpolation compensation is performed on the asynchronously acquired data frames based on the time stamps to eliminate the timing deviation caused by the sensor response delay, and an initial data set with strictly synchronized space and time is generated. Each data unit contains reflectance, elevation, and acquisition time information.

[0075] In a possible embodiment, performing multi-modal data fusion processing on the initial data set to obtain a dust feature matrix may include:

[0076] Step 301, performing adaptive median filtering processing on the multi-spectral reflectance data to obtain denoised reflectance data.

[0077] Specifically, adaptive median filtering can be performed on the multispectral reflectance data in the initial dataset, dynamically adjusting the filtering window size according to the noise level of each pixel neighborhood, using a large window to smooth pulse interference in areas with significant dust scattering noise, and using a small window to retain edge information in low-noise areas with rich texture details. After iterative filtering, the denoised reflectance data is output, eliminating abnormal brightness values caused by irregular reflections of dust particles, dynamically adjusting the processing strategy according to the local noise intensity, effectively suppressing dust scattering noise while retaining surface texture details, and avoiding information loss or over-smoothing problems caused by traditional fixed-window filtering.

[0078] Step 302, perform spatial interpolation processing on the lidar point cloud data based on the semivariogram to generate a terrain point cloud.

[0079] Exemplarily, for the void areas formed in the lidar point cloud data due to dust occlusion or other reasons, a spatial interpolation algorithm based on the semivariogram is used for repair. The spatial correlation of the point cloud is analyzed through the semivariogram, and the range and sill value in different directions are calculated; the elevation values of the missing points are estimated by Kriging interpolation according to the spatial correlation weight to generate a complete terrain point cloud. This method can restore the terrain undulation characteristics of the dust-obscured area and ensure the spatial continuity of the elevation data.

[0080] Step 303, fuse the denoised reflectance data and the terrain point cloud through an attention mechanism to generate a dust feature matrix.

[0081] Specifically, the attention mechanism is used to jointly analyze the denoised reflectance data and the complete terrain point cloud; the spectral texture features of the reflectance data and the elevation gradient features of the terrain point cloud are extracted through a convolutional neural network; the attention weights of each feature channel are calculated, and the feature vectors in the dust reflectance mutation area and the terrain steep change area are weighted and fused; a dust feature matrix is generated, where each pixel point contains reflectance intensity, elevation gradient, and the associated weight information of the two.

[0082] In a possible embodiment, calculating the migration probability distribution of dust particles based on the dust feature matrix includes:

[0083] Step 401, based on the dust feature matrix, construct a dynamic dust field model, and the dynamic dust field model calculates the particle migration probability density by coupling the wind speed vector and the diffusion coefficient.

[0084] Specifically, a dynamic dust field model can be constructed based on the wind speed vector and diffusion coefficient in the dust characteristic matrix. By coupling the fluid dynamics equation and the particle diffusion equation, the migration probability density of dust particles with different particle sizes in three-dimensional space can be calculated. Among them, the wind speed vector is extracted from the wind speed correlation characteristics in the characteristic matrix, and the diffusion coefficient is calibrated through historical dust migration data, forming a probability density distribution field describing the combined action of wind-driven and random diffusion of dust, so as to accurately quantify the combined influence of wind speed, terrain and particle size on dust migration and overcome the defect of large prediction deviation of traditional single physical models.

[0085] Step 402: Perform parameter optimization processing on the dynamic dust field model, and use the particle swarm optimization algorithm to iteratively obtain the spatial evolution gradient field, which is used to characterize the diffusion rate and direction of dust particles.

[0086] Specifically, the particle swarm optimization algorithm dynamically adjusts the model parameters by simulating the cooperative search behavior of the particle swarm to optimize the performance of the model. The spatial evolution gradient field is obtained through iterative calculation, and this gradient field can characterize the diffusion rate and direction of dust particles, providing an accurate spatio-temporal evolution law for simulating the migration trajectory of dust particles.

[0087] Step 403: Perform Monte Carlo trajectory simulation processing based on the spatial evolution gradient field to generate a migration probability distribution, which includes a set of spatio-temporal evolution trajectories of particles with different particle sizes.

[0088] Specifically, performing Monte Carlo trajectory simulation processing based on the optimized spatial evolution gradient field can generate a set of spatio-temporal evolution trajectories of particles with different particle sizes, forming a migration probability distribution. The Monte Carlo trajectory simulation comprehensively simulates the migration process of dust particles through a random sampling method, providing rich data support for the dynamic monitoring of sand disasters. The above method can not only reflect the migration path of dust particles, but also provide the spatio-temporal distribution law of particles with different particle sizes, providing a scientific basis for sand disaster prevention and control.

[0089] In a possible embodiment, generating a three-dimensional dynamic dust concentration field model according to the migration probability distribution and outputting the dynamic monitoring result of sand disasters based on the three-dimensional dynamic dust concentration field model may include:

[0090] Step 501: Perform three-dimensional reconstruction processing on the migration probability distribution based on the marching cubes algorithm to generate a three-dimensional dynamic dust concentration field model.

[0091] Specifically, based on the migration probability distribution, the marching cubes algorithm is used to perform three-dimensional reconstruction on the data. This algorithm divides the voxel grid in the three-dimensional space, extracts the isosurface according to the migration probability density value, and constructs a three-dimensional voxel model with a continuous distribution of dust concentration. The three-dimensional dynamic dust concentration field model stores the dust concentration values in units of voxels, supports time series updates, dynamically reflects the spatial evolution characteristics of the dust diffusion and settlement processes, and provides a high-fidelity three-dimensional data basis for predicting the sand hazard diffusion path.

[0092] Step 502: Perform vertical direction exponential decay integral processing on the three-dimensional dynamic dust concentration field model to calculate the deposition density distribution.

[0093] Specifically, perform vertical direction exponential decay integral processing on the generated three-dimensional dynamic dust concentration field model. This processing process calculates the deposition density distribution of dust at different positions by simulating the deposition process of dust particles in the vertical direction and considering the attenuation characteristics of dust concentration with height. The deposition density distribution reflects the accumulation of dust on the ground and provides key data for evaluating the impact of sand hazards.

[0094] Step 503: Perform binary segmentation processing on the deposition density distribution according to a preset threshold to mark the erosion risk areas.

[0095] Specifically, mark the areas with deposition density higher than the preset threshold as erosion risk areas. In this way, high-risk areas where dust deposition may cause a decline or damage to the performance of photovoltaic modules can be quickly identified, avoiding subjective experience biases caused by manual interpretation, and ensuring the objectivity and consistency of risk warnings.

[0096] In a possible embodiment, perform parameter optimization processing on the dynamic dust field model. The particle swarm optimization algorithm is used to iteratively obtain the spatial evolution gradient field, which may include:

[0097] Step 601: Define the fitness function of the particle swarm optimization algorithm based on the diffusion coefficient and wind speed vector. The fitness function is used to quantify the matching degree between the gradient field and the measured data.

[0098] Specifically, define the fitness function of the particle swarm optimization algorithm based on the diffusion coefficient and wind speed vector in the dynamic dust field model. Among them, the fitness function quantifies the matching degree between the two by calculating the mean square error between the predicted gradient field of the model and the measured dust distribution data; the diffusion coefficient is calibrated through historical dust migration data, and the wind speed vector is extracted from real-time wind field monitoring data to ensure that the fitness function accurately reflects the impact of actual environmental conditions on dust migration.

[0099] Step 602: Perform genetic crossover operation on the particle swarm optimization algorithm, and set the crossover probability to the first preset value to generate an initial optimized parameter set.

[0100] Specifically, by randomly selecting a pair of parent particles and exchanging some dimensions of their velocity and position vectors, a diverse set of initial optimization parameters is generated to enhance the search capability in the global solution space and avoid the defect that the traditional particle swarm algorithm is prone to fall into the local optimum. Among them, the first preset value is the crossover probability of the genetic crossover operation in the particle swarm optimization algorithm, which is used to control the exchange frequency of the parent particle parameters. The first preset value can be determined according to the experimental calibration method or based on empirical reference. The crossover probability can also be dynamically adjusted according to the convergence state in the algorithm iteration process, which is not limited here.

[0101] Step 603: performing a mutation operation on the initial optimization parameter set, setting the mutation probability to a second preset value, and generating an optimized spatial evolution gradient field.

[0102] Specifically, tiny perturbations are randomly introduced into the particle position vector to break the symmetry of the parameter search path and generate an updated set of optimization parameters. The mutation operation effectively expands the coverage of the solution space and improves the adaptability of the algorithm to complex nonlinear optimization problems. The setting of the mutation probability aims to maintain parameter diversity and avoid local optimal solutions, which can be determined through small step size trial method, adaptive strategy and / or scenario adaptation method.

[0103] Step 604, iteratively solving the dynamic dust field model according to the updated optimization parameter set, and outputting the optimized spatial evolution gradient field.

[0104] Specifically, based on the updated set of optimized parameters, the dynamic dust field model is solved iteratively multiple times. In each iteration, the particle velocity and position are updated according to the fitness function value, gradually converging to the optimal parameter combination, and outputting the optimized spatial evolution gradient field. The gradient field represents the diffusion rate and dominant migration trend of dust particles in the horizontal and vertical directions in vector form.

[0105] In a possible embodiment, performing vertical exponential decay integral processing on the three-dimensional dynamic dust concentration field model to calculate the deposition density distribution may include:

[0106] Step 701, performing Gaussian smoothing filtering processing based on the three-dimensional dynamic dust concentration field model to eliminate high-frequency noise interference and generate a denoised concentration field model.

[0107] Specifically, Gaussian smoothing filtering is performed on the three-dimensional dynamic dust concentration field model, and a three-dimensional Gaussian kernel is used to perform convolution operations on the concentration field. The smoothing intensity is controlled by adjusting the standard deviation of the kernel function to eliminate high-frequency interference signals caused by sensor noise or simulation errors. After filtering, a denoised concentration field model is generated, which retains the main trend characteristics of dust diffusion while suppressing local abnormal fluctuations and avoiding edge blurring problems caused by traditional mean filtering.

[0108] Step 702: Perform adaptive step-size integration on the denoised concentration field model to generate an intermediate deposition density distribution.

[0109] Specifically, a vertical direction integration operation can be performed on the denoised concentration field model along the normal direction of the photovoltaic panel surface, and the integration step size can be dynamically adjusted according to the change rate of the dust concentration gradient: small step sizes are used for fine sampling in areas with steep concentration gradient changes, and large step sizes are used for fast calculation in flat areas. By adopting a variable step-size strategy to balance calculation accuracy and efficiency, the dust concentration contribution values of each height layer are accumulated layer by layer to generate an intermediate deposition density distribution map.

[0110] Step 703: Perform normalization processing on the intermediate deposition density distribution to generate a deposition density distribution.

[0111] Specifically, linear normalization processing is performed on the intermediate deposition density distribution map. According to the preset maximum deposition density reference value, the deposition mass of each pixel point is mapped to the 0-1 standardization interval to eliminate the influence of dimensional differences on subsequent analysis. The normalized deposition density distribution map intuitively shows the relative severity of dust coverage in different regions in the form of a gray-scale gradient, providing a unified quantitative benchmark for risk grading.

[0112] In summary, the dynamic desert photovoltaic sand hazard monitoring method based on an unmanned aerial vehicle provided in the embodiments of the present application obtains the dust-raising data of the photovoltaic power station area collected by the unmanned aerial vehicle with multi-modal sensors, collects the dust reflectivity, terrain elevation and wind field data according to a dynamic grid path to generate a spatio-temporally aligned initial data set; performs adaptive noise suppression, spatial interpolation and attention mechanism fusion processing on the multi-modal data to construct a dust characteristic matrix representing the optical characteristics of dust-raising, terrain interaction and wind field coupling relationship; constructs a dynamic dust field model based on the principle of fluid dynamics, uses the particle swarm optimization algorithm to iteratively solve the spatial evolution gradient field, and combines Monte Carlo simulation to generate the migration probability distribution of different particle sizes; converts the probability distribution into a dynamic dust concentration field through three-dimensional reconstruction technology, and outputs the deposition density distribution and erosion risk area through vertical direction integration and threshold segmentation. The above technical solution realizes real-time collection of large-scale sand hazard data through the unmanned aerial vehicle platform, breaks through the spatio-temporal resolution limitations of traditional monitoring methods, can implement all-weather dynamic monitoring of the formation, diffusion and deposition processes of sand hazards in desert photovoltaic power stations, provides a scientific decision-making basis for cleaning path planning, power station expansion site selection and ecological governance effect evaluation, and effectively improves the anti-sand damage ability of photovoltaic facilities and the stability of power generation efficiency.

[0113] It should be understood that although the steps in the flowcharts involved in the above-described embodiments are sequentially shown according to the indications of the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps has no strict order limit, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-described embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily executed at the same moment, but can be executed at different moments. The execution order of these steps or stages is not necessarily sequential, but can be executed alternately or in turn with at least a part of other steps or steps or stages in other steps.

[0114] Based on the same inventive concept, an embodiment of the present application also provides a dynamic desert photovoltaic sand hazard monitoring system based on an unmanned aerial vehicle for implementing the above-mentioned dynamic desert photovoltaic sand hazard monitoring method based on an unmanned aerial vehicle. The implementation solutions provided by this system to solve problems are similar to the implementation solutions recorded in the above method. Therefore, the specific limitations in one or more embodiments of the dynamic desert photovoltaic sand hazard monitoring system based on an unmanned aerial vehicle provided below can refer to the limitations on the dynamic desert photovoltaic sand hazard monitoring method based on an unmanned aerial vehicle in the above text, and will not be repeated here.

[0115] In an exemplary embodiment, as Figure 2 shown, a dynamic desert photovoltaic sand hazard monitoring system 10 based on an unmanned aerial vehicle is provided, including:

[0116] A data acquisition module 11, configured to acquire dust data in the photovoltaic power station area and generate an initial data set. The dust data is collected by a multi-modal sensor carried by an unmanned aerial vehicle.

[0117] A data fusion module 12, configured to perform multi-modal data fusion processing on the initial data set to obtain a dust characteristic matrix, and the dust characteristic matrix is used to characterize the reflectivity, terrain elevation, and wind speed correlation characteristics of the dust.

[0118] A migration probability calculation module 13, configured to calculate the migration probability distribution of dust particles based on the dust characteristic matrix, and the migration probability distribution is used to characterize the spatio-temporal evolution law of particles with different particle sizes.

[0119] A dynamic monitoring module 14, configured to generate a three-dimensional dynamic dust concentration field model according to the migration probability distribution and output a dynamic desert photovoltaic sand hazard monitoring result based on the three-dimensional dynamic dust concentration field model. The dynamic desert photovoltaic sand hazard monitoring result includes the deposition density distribution and the erosion risk area.

[0120] In a possible embodiment, the data acquisition module 11 may include:

[0121] A path acquisition unit 111 is configured to obtain a first instruction for instructing a drone to fly along a dynamic grid path, and the flight altitude is adaptively adjusted based on real-time wind speed data.

[0122] A remote sensing aerial survey unit 112 is configured to control the drone to perform data acquisition according to the first instruction, synchronously collect multi-spectral reflectance data and lidar point cloud data through a multi-modal sensor, and generate an original sensing data set with spatio-temporal tags.

[0123] A spatio-temporal synchronization unit 113 is configured to perform coordinate alignment and time synchronization processing on the original sensing data set with spatio-temporal tags to generate an initial data set.

[0124] In a possible embodiment, the data fusion module 12 may include:

[0125] A multi-spectral data denoising unit 121 is configured to perform adaptive median filtering on the multi-spectral reflectance data to obtain denoised reflectance data.

[0126] A point cloud interpolation unit 122 is configured to perform spatial interpolation processing based on a semi-variogram on the lidar point cloud data to generate a terrain point cloud.

[0127] An attention fusion unit 123 is configured to fuse the denoised reflectance data and the terrain point cloud through an attention mechanism to generate a dust characteristic matrix.

[0128] In a possible embodiment, the migration probability calculation module 13 may include:

[0129] A dynamic dust field modeling unit 131 is configured to construct a dynamic dust field model based on the dust characteristic matrix. The dynamic dust field model calculates the particle migration probability density by coupling the wind speed vector and the diffusion coefficient.

[0130] A particle swarm optimization unit 132 is configured to perform parameter optimization on the dynamic dust field model, and iteratively obtain a spatial evolution gradient field using the particle swarm optimization algorithm. The spatial evolution gradient field is used to characterize the diffusion rate and direction of dust particles.

[0131] A Monte Carlo trajectory simulation unit 133 is configured to perform Monte Carlo trajectory simulation processing based on the spatial evolution gradient field to generate a migration probability distribution, which includes a set of spatio-temporal evolution trajectories of particles with different particle sizes.

[0132] In a possible embodiment, the dynamic monitoring module 14 includes:

[0133] A three-dimensional modeling unit 141 is configured to perform three-dimensional reconstruction on the migration probability distribution based on the marching cubes algorithm to generate a three-dimensional dynamic dust concentration field model.

[0134] A physical sedimentation unit 142 is used to perform vertical direction exponential decay integral processing on the three-dimensional dynamic sand dust concentration field model to calculate the sediment density distribution.

[0135] An erosion risk identification unit 143 is used to perform binary segmentation processing on the sediment density distribution according to a preset threshold to mark the erosion risk area.

[0136] In a possible embodiment, the particle swarm optimization unit 132 includes:

[0137] A fitness function definition subunit 1321 is used to define the fitness function of the particle swarm optimization algorithm based on the diffusion coefficient and the wind speed vector. The fitness function is used to quantify the matching degree between the gradient field and the measured data.

[0138] A genetic crossover subunit 1322 is used to perform a genetic crossover operation on the particle swarm optimization algorithm. The crossover probability is set to a first preset value to generate an initial optimized parameter set.

[0139] A mutation optimization subunit 1323 is used to perform a mutation operation on the initial optimized parameter set. The mutation probability is set to a second preset value to generate an optimized spatial evolution gradient field.

[0140] An iterative optimization subunit 1324 is used to iteratively solve the dynamic sand dust field model according to the updated optimized parameter set and output the optimized spatial evolution gradient field.

[0141] In a possible embodiment, the physical sedimentation unit 142 may include:

[0142] A filtering and denoising subunit 1421 is used to perform Gaussian smoothing filtering processing on the three-dimensional dynamic sand dust concentration field model to eliminate high-frequency noise interference and generate a denoised concentration field model.

[0143] An adaptive step size subunit 1422 is used to perform adaptive step size integral processing on the denoised concentration field model to generate an intermediate sediment density distribution.

[0144] A normalization processing subunit 1423 is used to perform normalization processing on the intermediate sediment density distribution to generate the sediment density distribution.

[0145] In an embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of a dynamic monitoring method for desert photovoltaic sand damage based on an unmanned aerial vehicle as described above are implemented.

[0146] In an embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0147] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to the partial descriptions of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the present disclosure solution. A person of ordinary skill in the art can understand and implement it without creative efforts.

[0148] The above embodiments only represent several implementation manners of the embodiments of the present application. The description is relatively specific and detailed, but it should not be construed as a limitation on the patent scope of the embodiments of the application. It should be noted that for those of ordinary skill in the art, without departing from the concept of the embodiments of the present application, several modifications and improvements can still be made, and these all belong to the protection scope of the embodiments of the present application.

Claims

1. A dynamic monitoring method for desert photovoltaic sand hazards based on drones, characterized in that, The method includes: Obtaining the dust data of the photovoltaic power station area to generate an initial data set, where the dust data is collected by a multi-modal sensor carried by a drone; Performing multi-modal data fusion processing on the initial data set to obtain a dust characteristic matrix, which is used to characterize the reflectivity, terrain elevation, and wind speed correlation characteristics of the dust; Calculating the migration probability distribution of the dust particles based on the dust characteristic matrix, where the migration probability distribution is used to characterize the spatio-temporal evolution law of particles with different particle sizes; Generating a three-dimensional dynamic dust concentration field model according to the migration probability distribution, and outputting the dynamic sand hazard monitoring result based on the three-dimensional dynamic dust concentration field model, where the dynamic sand hazard monitoring result includes the deposition density distribution and the erosion risk area.

2. The method according to claim 1, wherein The obtaining the dust data of the photovoltaic power station area to generate an initial data set includes: Obtaining a first instruction, which is used to instruct the drone to fly along a dynamic grid path, and the flight height is adaptively adjusted based on the real-time wind speed data; Controlling the drone to perform data collection according to the first instruction, and synchronously collecting multi-spectral reflectivity data and lidar point cloud data through the multi-modal sensor to generate a raw sensing data set with spatio-temporal marks; Performing coordinate alignment and time synchronization processing on the raw sensing data set to generate the initial data set.

3. The method according to claim 2, wherein The performing multi-modal data fusion processing on the initial data set to obtain a dust characteristic matrix includes: Performing adaptive median filtering processing on the multi-spectral reflectivity data to obtain noise-reduced reflectivity data; Performing spatial interpolation processing based on the semi-variogram on the lidar point cloud data to generate a terrain point cloud; Fusing the noise-reduced reflectivity data and the terrain point cloud through an attention mechanism to generate the dust characteristic matrix.

4. The method according to claim 1, wherein The calculating the migration probability distribution of the dust particles based on the dust characteristic matrix includes: Based on the dust characteristic matrix, constructing a dynamic dust field model, where the dynamic dust field model calculates the particle migration probability density by coupling the wind speed vector and the diffusion coefficient; Performing parameter optimization processing on the dynamic dust field model, and iteratively obtaining a spatial evolution gradient field by using a particle swarm optimization algorithm, where the spatial evolution gradient field is used to characterize the diffusion rate and direction of the dust particles; Performing Monte Carlo trajectory simulation processing based on the spatial evolution gradient field to generate the migration probability distribution, where the migration probability distribution contains a set of spatio-temporal evolution trajectories of particles with different particle sizes.

5. The method according to claim 1, wherein The generating a three-dimensional dynamic dust concentration field model according to the migration probability distribution, and outputting the dynamic sand hazard monitoring result based on the three-dimensional dynamic dust concentration field model includes: Performing three-dimensional reconstruction processing on the migration probability distribution based on the marching cubes algorithm to generate the three-dimensional dynamic dust concentration field model; Performing vertical direction exponential decay integral processing on the three-dimensional dynamic dust concentration field model to calculate the deposition density distribution; Performing binary segmentation processing on the deposition density distribution according to a preset threshold to mark the erosion risk area.

6. The method according to claim 4, characterized in that, The performing parameter optimization processing on the dynamic dust field model, and iteratively obtaining a spatial evolution gradient field by using a particle swarm optimization algorithm includes: Define a fitness function of the particle swarm optimization algorithm based on the diffusion coefficient and the wind speed vector, where the fitness function is used to quantify the matching degree between the gradient field and the measured data; Perform a genetic crossover operation on the particle swarm optimization algorithm with the crossover probability set to a first preset value to generate an initial set of optimization parameters; Perform a mutation operation on the initial set of optimization parameters with the mutation probability set to a second preset value to generate the optimized spatial evolution gradient field; Iteratively solve the dynamic dust field model according to the updated set of optimization parameters and output the optimized spatial evolution gradient field.

7. The method according to claim 5, characterized in that, The performing of the vertical direction exponential decay integral processing on the three-dimensional dynamic dust concentration field model to calculate the deposition density distribution includes: Perform Gaussian smoothing filtering processing on the three-dimensional dynamic dust concentration field model to eliminate high-frequency noise interference and generate a denoised concentration field model; Perform adaptive step-size integral processing on the denoised concentration field model to generate an intermediate deposition density distribution; Perform normalization processing on the intermediate deposition density distribution to generate the deposition density distribution.

8. A dynamic monitoring system for desert photovoltaic sand hazards based on drones, characterized in that, The system includes: A data acquisition module, configured to acquire dust data in the photovoltaic power station area to generate an initial data set, where the dust data is collected by a multi-modal sensor carried by a drone; A data fusion module, configured to perform multi-modal data fusion processing on the initial data set to obtain a dust feature matrix, where the dust feature matrix is used to characterize the reflectivity, terrain elevation, and wind speed correlation features of the dust; A migration probability calculation module, configured to calculate the migration probability distribution of dust particles based on the dust feature matrix, where the migration probability distribution is used to characterize the spatio-temporal evolution law of particles with different particle sizes; A dynamic monitoring module, configured to generate a three-dimensional dynamic dust concentration field model according to the migration probability distribution and output a dynamic sand hazard monitoring result based on the three-dimensional dynamic dust concentration field model, where the dynamic sand hazard monitoring result includes a deposition density distribution and an erosion risk area.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method according to any one of claims 1 to 7.

Citation Information

Cited By

  • Device and method for dynamically regulating and controlling spraying of dust suppression unmanned aerial vehicle for photovoltaic construction of loess

    CN120631033A

  • Ruggedized computer dust particle prediction and self-adaptive dust prevention method and system

    CN120805779A

  • Reinforced computer dust particle prediction and adaptive dust prevention method and system

    CN120805779B

  • Unmanned aerial vehicle anomaly diagnosis method and system based on multi-modal data fusion

    CN121327729A

  • Real-time monitoring method and system for sand flow under regional environment characteristics

    CN121522778A