Lake restoration method and system based on unmanned aerial vehicle

Abstract

The application provides a lake restoration method and system based on a UAV. Through the integration of multi-source heterogeneous data such as images, spectra, water temperature, and water quality, the lake ecological condition is comprehensively perceived, and the limitation of traditional monitoring data being single and one-sided is broken. Relying on deep learning and multi-element data analysis algorithm to deeply analyze the data, combined with intelligent decision algorithm to generate customized maintenance decision, the maintenance measures are changed from experience to science and precision. At the same time, the operation mode of the operation UAV and the multi-functional water operation ship is adopted, which takes into account the needs of small-scale, fine operation and large-scale, large-scale operation. The intelligent positioning of the multi-functional water operation ship and the special operation equipment configuration ensure the accurate landing of the maintenance operation, realize the intelligent and efficient operation of the whole process of lake restoration, greatly improve the efficiency and effect of lake restoration, reduce the maintenance cost, and promote the stable recovery of the lake ecological system.

Description

Technical Field

[0001] This invention relates to the field of lake ecological restoration technology, specifically to a lake restoration method and system based on unmanned aerial vehicles (UAVs). Background Technology

[0002] Lake ecosystems, as vital carriers of water resources, play an irreplaceable role in regulating climate, purifying water quality, and maintaining biodiversity. Submerged plants, as a core component of lake ecosystems, increase dissolved oxygen, absorb nutrients, suppress algal blooms, and provide habitats for aquatic organisms through photosynthesis; their healthy growth is crucial for the stability of lake ecosystems. Therefore, the restoration of submerged plants has become a core technical means for lake ecological restoration.

[0003] However, submerged plants face complex environmental constraints and ecological challenges during the process of artificial planting and the formation of stable communities. In the early stages of restoration, factors such as insufficient water transparency, algal competition, fish grazing, and pollution input can easily lead to low survival rates of submerged plants. After survival, due to differences in the stress resistance of different species, the community tends to become homogenous, resulting in a decline in biodiversity and thus reducing ecosystem stability. Therefore, the full life-cycle maintenance of submerged plants requires precise monitoring of growth status, dynamic regulation of environmental factors, and timely intervention in community structure. Existing maintenance methods have many limitations: traditional manual inspections and regular monitoring are inefficient and have limited coverage, making it difficult to meet the needs of full-area monitoring of large lakes, and manual judgment is easily affected by subjective factors, resulting in large errors; although satellite remote sensing technology can achieve large-scale coverage, its low resolution makes it impossible to accurately identify changes in small-area submerged plant communities and the impact of local environmental heterogeneity, making it difficult to meet the needs of refined maintenance; the existing monitoring and maintenance links are disconnected, and there is a lack of intelligent decision-making systems supported by real-time data, resulting in insufficient targeting and timeliness of maintenance measures, high maintenance costs, and poor results. With its advantages of high mobility, flexible operation, and high monitoring efficiency, drone technology can quickly acquire high-resolution surface information, providing a new technical approach for monitoring submerged plants in lakes.

[0004] In other words, how to provide a drone-based lake restoration method to achieve precise maintenance of submerged plants, so as to improve the efficiency of lake ecological restoration, reduce maintenance costs, and promote the stable recovery of lake ecosystems, is a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] This invention provides a method and system for lake restoration based on unmanned aerial vehicles (UAVs) to solve at least one of the above-mentioned technical problems.

[0006] In a first aspect, this application provides a lake restoration method based on unmanned aerial vehicles (UAVs), the method comprising:

[0007] Acquire multi-source heterogeneous data of the target lake, including image data, spectral data, water temperature data, spatial state information, meteorological information, and water quality information; The multi-source heterogeneous data is preprocessed, and the preprocessed data is analyzed using deep learning image recognition algorithms and multivariate data analysis algorithms. Based on data analysis results and intelligent decision-making algorithms, targeted maintenance decisions are generated, including at least one of fertilization, pesticide application, replanting, harvesting, water purification, algae removal, and water level control. Based on the maintenance decision, the operation drone or watercraft is dispatched to perform the corresponding maintenance operation. The watercraft is equipped with an intelligent positioning system and special operating equipment.

[0008] Optionally, acquiring multi-source heterogeneous data of the target lake includes: A remote sensing monitoring plan is formulated based on the geographic information and the distribution of submerged plants of the target lake. The remote sensing monitoring plan includes the monitoring range, monitoring frequency, preset flight route, flight parameters and shooting parameters. According to the remote sensing monitoring plan, a monitoring drone is scheduled to collect the image data, the spectral data, the water temperature data, and the spatial status information; Acquire meteorological information collected by fixed monitoring stations deployed around the target lake and water quality information collected by water quality sensors deployed in the target lake.

[0009] Optionally, the monitoring drone includes a multi-rotor drone and a fixed-wing drone. The monitoring drone integrates a satellite navigation and positioning system, an inertial measurement unit, and an RTK device for real-time monitoring of the drone's spatial status information. The monitoring drone is also equipped with an optical camera, a multispectral camera, and a thermal imaging camera. The optical camera is used to collect the image data, the multispectral camera is used to collect the spectral data, and the thermal imaging camera is used to collect the water temperature data.

[0010] Optionally, the preprocessing includes: format conversion, spatiotemporal stitching, noise reduction, and calibration.

[0011] Optionally, the analysis of the preprocessed data using deep learning image recognition algorithms and multivariate data analysis algorithms includes: The pre-processed image data is analyzed using a trained image recognition algorithm to identify the types, coverage area, and growth density of submerged plants in the target lake. Based on data analysis algorithms, a comprehensive assessment model for the growth status of submerged plants, a prediction model for algal biomass, and a model for changes in water transparency were constructed. The preprocessed spectral data are subjected to normalized red-edge vegetation index (NDRE) calculation to obtain the vegetation index. The vegetation index NDRE is evaluated based on the comprehensive evaluation model for the growth status of submerged plants to obtain the comprehensive evaluation results for the growth status of submerged plants. The preprocessed spectral data and water temperature data are input into the algal biomass prediction model to obtain the algal biomass prediction results. The preprocessed spectral data and water quality information are input into the water transparency change model to obtain the prediction results of water transparency and dynamic change trend.

[0012] Optionally, the evaluation of the vegetation index NDRE based on the comprehensive evaluation model for the growth status of submerged plants includes: If the growth characteristics of submerged plants show a linear response relationship with the vegetation index NDRE, then a linear evaluation model is selected to evaluate the vegetation index NDRE. If the growth characteristics of submerged plants have a non-linear response relationship with the vegetation index NDRE, the vegetation index NDRE can be evaluated by selecting a power function evaluation model or an exponential evaluation model based on the response trend. The power function evaluation model is suitable for scenarios where the influence of the index changes exponentially, while the exponential evaluation model is suitable for scenarios where the influence of the index increases exponentially or saturates.

[0013] Optionally, the formula for calculating the normalized red-edge vegetation index is: ; The calculation formula for the linear evaluation model is as follows: ; The calculation formula for the power function evaluation model is as follows: ; The formula for calculating the exponential model is as follows: ; Wherein, NIR represents the near-infrared band; RedEdge represents the red-edge band; B represents the biomass or cover of submerged plants; and e represents the base of the natural logarithm. x and y This represents the fitting coefficient.

[0014] Optionally, the calculation formula for the algal biomass prediction model is as follows:

[0015] The calculation formula for the water transparency prediction model is as follows:

[0016] Where Chl-a represents the concentration of chlorophyll a; R 680R represents the surface reflectance in the red light band with a center wavelength of 680 nm; 710 R represents the reflectivity of the red-edge band with a center wavelength of 710 nm; 560 denoted by ρ(λ), which represents the reflectance of the green light band with a center wavelength of 560 nm; t and z represent the fitting coefficients of the algal biomass prediction model; SDD represents the water transparency; ρ(λ) represents the apparent reflectance of the water at wavelength λ; λ1 represents the wavelength of the green light band; λ2 represents the wavelength of the red light or near-infrared band; and m and n represent the fitting coefficients of the water transparency prediction model.

[0017] Secondly, this application provides a lake restoration system based on unmanned aerial vehicles (UAVs), including a ground control station, a monitoring UAV, several fixed monitoring stations, a water quality sensor, an operational UAV, and a waterborne operation vessel. Several fixed monitoring stations are deployed around the target lake to collect meteorological information for the area. The water quality sensor is deployed in the target lake to collect water quality information of the target lake; The monitoring drone is used to collect image data, spectral data, water temperature data, and spatial status information of the target lake; The waterborne operation vessel is equipped with an intelligent positioning system and specialized operating equipment for performing maintenance operations; The ground control station is communicatively connected to the monitoring drone, fixed monitoring station, water quality sensor, operational drone, and waterborne operation vessel. The ground control station is used to acquire multi-source heterogeneous data of the target lake, preprocess the multi-source heterogeneous data, and analyze the preprocessed data using deep learning image recognition algorithms and multivariate data analysis algorithms. Based on the data analysis results and intelligent decision-making algorithms, targeted maintenance decisions are generated, and operational drones and / or waterborne operation vessels are dispatched to perform corresponding maintenance operations based on the maintenance decisions. The multi-source heterogeneous data includes the image data, the spectral data, the water temperature data, the spatial state information, the meteorological information, and the water quality information; the maintenance decision includes at least one of fertilization, pesticide application, replanting, harvesting, water purification, and water level control.

[0018] Optionally, the ground control station is also used for: A remote sensing monitoring plan is formulated based on the geographic information and submerged plant distribution of the target lake. The remote sensing monitoring plan includes the monitoring range, monitoring frequency, preset flight route, flight parameters and shooting parameters. According to the remote sensing monitoring plan, the monitoring drone is controlled to collect image data, spectral data, water temperature data and spatial status information of the target lake.

[0019] This invention integrates multi-source heterogeneous data such as images, spectra, water temperature, and water quality to achieve comprehensive perception of the lake's ecological condition, overcoming the limitations of traditional monitoring data that is singular and one-sided. It relies on deep learning and multivariate data analysis algorithms to deeply analyze the data and combines this with intelligent decision-making algorithms to generate customized maintenance decisions, shifting maintenance measures from experience-based to scientific and precise. Simultaneously, it employs a collaborative operation mode using drones and multi-functional watercraft, balancing the needs of small-scale, precision operations with large-scale, mass operations. The intelligent positioning and specialized equipment configuration of the multi-functional watercraft ensure precise implementation of maintenance operations, achieving intelligent and efficient operation throughout the entire lake restoration process. This significantly improves the efficiency and effectiveness of lake restoration, reduces maintenance costs, and promotes the stable recovery of the lake ecosystem. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this specification or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of a drone-based lake restoration method provided in this application. Detailed Implementation

[0022] This application provides a method and system for lake restoration based on unmanned aerial vehicles (UAVs) to solve at least one of the above-mentioned technical problems.

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in a sequence other than that illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules is not necessarily limited to those explicitly listed, but may include other steps or modules not explicitly listed or inherent to such processes, methods, products, or devices. The naming or numbering of steps appearing in this application does not imply that the steps in the method flow must be performed in the chronological / logical order indicated by the naming or numbering. The execution order of named or numbered process steps can be changed according to the desired technical purpose, as long as the same or similar technical effect is achieved.

[0025] The module division described in this application is a logical division. In practical applications, there may be other division methods. For example, multiple modules may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the shown or discussed mutual coupling, direct coupling, or communication connection may be through some interface, and the indirect coupling or communication connection between modules may be electrical or other similar forms, none of which are limited in this application. Furthermore, the modules or sub-modules described as separate components may or may not be physically separated, may or may not be physical modules, or may be distributed in multiple circuit modules. Some or all of the modules can be selected to achieve the purpose of the solution in this application according to actual needs.

[0026] Next, please refer to Figure 1 , Figure 1 This is a flowchart illustrating a drone-based lake restoration method according to an embodiment of the present invention. As an embodiment of the drone-based lake restoration method provided by the present invention, the drone-based lake restoration method includes the following steps S110 to S140: Step S110: Obtain multi-source heterogeneous data of the target lake. The multi-source heterogeneous data includes image data, spectral data, water temperature data, spatial state information, meteorological information, and water quality information. As one possible approach, the acquisition of multi-source heterogeneous data of the target lake involved in step S110 above specifically includes the following: A remote sensing monitoring plan is developed based on the geographic information of the target lake and the distribution of submerged plants. The remote sensing monitoring plan includes the monitoring range, monitoring frequency, preset flight route, flight parameters and shooting parameters. Based on the remote sensing monitoring plan, monitoring drones are scheduled to collect image data, spectral data, water temperature data, and spatial status information; Meteorological information collected by fixed monitoring stations deployed around the target lake and water quality information collected by water quality sensors deployed in the target lake are obtained. Meteorological information includes wind speed, wind direction, and atmospheric humidity.

[0027] The monitoring frequency can be set according to the growth cycle of submerged plants, such as once a day in the early stage of planting, once a week during the stable growth period, and once every two weeks during routine inspections. Flight parameters are adapted to the importance of the monitoring area; the flight altitude can be set to 30-40 meters in sensitive areas and 80-120 meters in regular areas. Using GIS software combined with satellite imagery and on-site survey data, the drone flight path is precisely planned to ensure comprehensive and complete coverage of the submerged plant planting area and its surrounding 300m radius, while avoiding duplicate monitoring. Multiple parallel flight paths, such as zigzag paths, are planned based on the distribution range of the lake's submerged plants to ensure comprehensive coverage of the target area, with some overlap between adjacent paths to guarantee the integrity of image acquisition. The monitoring plan also includes setting differentiated flight altitude, speed, and shooting parameters according to the monitoring needs and importance of different areas to ensure the acquisition of high-quality, targeted images and data. For areas sensitive to changes in submerged plant communities, the flight altitude should be reduced to 30 meters to increase the image resolution and ensure that even minute changes can be captured; for large areas of routine monitoring, the flight altitude should be appropriately increased to 80 meters to increase the flight speed and improve monitoring efficiency.

[0028] Specifically, by using standardized drones to collect core data such as images and spectra, combined with fixed-point data collection from fixed monitoring stations around the lake and water quality sensors within the lake, a three-dimensional data acquisition network of "drone mobile monitoring + fixed station monitoring" is formed, compensating for the coverage blind spots of single monitoring methods. This data acquisition method not only achieves dynamic monitoring of the entire lake area but also ensures the continuous and stable collection of key data such as meteorological and water quality data, providing a comprehensive, accurate, and high-quality data source for subsequent data analysis, thereby improving the reliability of data support for lake restoration from the source. Water quality information includes key water quality parameters such as pH value, dissolved oxygen (DO), chemical oxygen demand (COD), total nitrogen (TN), and total phosphorus (TP).

[0029] As one feasible approach, monitoring drones include multi-rotor drones and fixed-wing drones. Monitoring drones integrate satellite navigation and positioning systems, inertial measurement units, and RTK devices to monitor the spatial status information of drones in real time. The monitoring drone is also equipped with an optical camera, a multispectral camera, and a thermal imaging camera. The optical camera is used to collect image data, the multispectral camera is used to collect spectral data, and the thermal imaging camera is used to collect water temperature data.

[0030] Among them, monitoring drones can be selected from multi-rotor or fixed-wing drones with excellent stability and long endurance. In complex lake environments, multi-rotor drones can hover and turn flexibly, making them suitable for detailed monitoring of key local areas; while fixed-wing drones can quickly cover large areas, meeting the needs of patrolling the overall condition of the lake.

[0031] The monitoring drone integrates a high-precision satellite navigation and positioning system (GPS) and an inertial measurement unit (IMU), combined with real-time dynamic differential positioning technology (RTK equipment), to achieve centimeter-level precise positioning and autonomous navigation. This ensures that each flight strictly follows a pre-set complex flight path and precise altitude to monitor the target area, and accurately records the geographical location information of the captured images, providing a reliable spatial reference for subsequent data analysis and comparison. Equipped with an ultra-high resolution optical camera, with a resolution of 8000×6000 pixels or even higher, it can clearly capture the subtle features of submerged plants, such as leaf texture and plant morphology, thereby accurately identifying the species, distribution range, and growth status of submerged plants. Even changes in small areas of submerged plant communities can be accurately identified. Simultaneously, it is equipped with a spectral analysis module: using an advanced multispectral camera, it can collect data including red, green, blue, near-infrared, and short-wave infrared, among others. By analyzing spectral information across more than one spectral band and examining the unique reflectance differences of submerged plants in different spectral bands, their health status can be accurately assessed. For example, it can accurately detect whether they are suffering from pests or diseases, or whether they are deficient in nutrients, providing a scientific basis for precise care. Furthermore, installing a high-precision thermal imaging camera with a temperature resolution of up to 0.03℃ allows for the sensitive monitoring of subtle changes in water temperature, enabling precise analysis of the potential impact of temperature variations on the growth of submerged plants. Because even small fluctuations in water temperature can significantly affect the physiological activities of submerged plants, a suitable water temperature is one of the key conditions for their normal growth.

[0032] According to the pre-set monitoring plan, the monitoring drone took off on time at the designated time and flew along the pre-set flight path. During the flight, the optical camera captured visible light images and videos of the submerged plant area from different angles, obtaining image data; the multispectral camera simultaneously acquired multi-band spectral data; and the thermal imaging camera simultaneously acquired water temperature data. At the same time, the GPS, IMU, and RTK devices on the drone recorded high-precision flight trajectory and location information in real time.

[0033] Specifically, by integrating multiple types and devices of monitoring drones, and combining multi-rotor and fixed-wing drones, the system achieves both detailed monitoring of key areas in lakes and rapid coverage monitoring of the entire lake area, adapting to the diverse monitoring needs of complex lake environments. The integration of satellite navigation and positioning systems, IMUs, and RTK devices enables centimeter-level precise positioning and autonomous navigation of the drones, ensuring the accuracy of flight trajectories and shooting locations, and giving the collected data precise spatial attributes, providing a reliable benchmark for subsequent spatiotemporal analysis and comparison. The inclusion of optical cameras, multispectral cameras, and thermal imaging cameras enables the simultaneous acquisition of image, spectral, and water temperature data; a single device can acquire multiple types of core data, significantly improving monitoring efficiency.

[0034] During flight, the monitoring drone transmits massive amounts of collected data to the ground control station at high speed in real time via 5G or satellite communication modules. Encryption technology is used during data transmission to ensure data security and integrity.

[0035] Step S120: Preprocess the multi-source heterogeneous data, and analyze the preprocessed data using deep learning image recognition algorithms and multivariate data analysis algorithms; As one possible approach, the preprocessing involved in step S120 above includes format conversion, spatiotemporal stitching, noise reduction, and calibration to improve data quality.

[0036] This claim clarifies a standardized preprocessing procedure for multi-source heterogeneous data. Format conversion enables unified processing of data of different types and formats, resolving compatibility issues. Spatiotemporal stitching allows dispersed spatial points and time-series data to form a complete, all-time data system for the entire lake area. Noise reduction effectively removes invalid data such as environmental interference and equipment errors during data acquisition, improving data purity. Calibration ensures data accuracy and consistency. Through this series of preprocessing operations, the original collected data is optimized and purified, eliminating redundancy and errors in heterogeneous data. This allows subsequent deep learning image recognition and multivariate data analysis to be conducted based on standardized, high-quality data, avoiding interference from invalid data and improving the accuracy and reliability of data analysis, thus laying a solid data foundation for maintenance decision generation.

[0037] As one possible approach, the analysis of the preprocessed data using deep learning image recognition algorithms and multivariate data analysis algorithms involved in step S120 above specifically includes the following sub-steps (1) to (6): (1) The pre-processed image data is analyzed using a trained image recognition algorithm to identify the types, coverage area and growth density of submerged plants in the target lake; Among them, the image recognition algorithm has been trained on a large number of submerged plant image samples of different types, different growth stages and different health conditions, and can automatically and accurately identify detailed information such as the type, coverage area, growth density and plant height of submerged plants in drone-captured images.

[0038] (2) Construct a comprehensive evaluation model for the growth status of submerged plants, a prediction model for algal biomass, and a model for changes in water transparency based on data analysis algorithms; Among them, the data analysis model integrates multispectral data, environmental monitoring data and historical data, and uses multivariate statistical analysis methods such as principal component analysis (PCA) and partial least squares (PLS) to construct a comprehensive assessment model for the growth status of submerged plants, an algal biomass prediction model and a water transparency change model, etc., to accurately assess the health index, growth trend and community stability of submerged plants, and at the same time accurately grasp the dynamic changes of algal biomass and water transparency. (3) The normalized red-edge vegetation index (NDRE) was calculated from the preprocessed spectral data; (4) The vegetation index NDRE was evaluated based on the comprehensive evaluation model of submerged plant growth status to obtain the comprehensive evaluation results of submerged plant growth status. As one feasible approach, the assessment of the vegetation index NDRE based on the comprehensive assessment model of submerged plant growth status involved in submerged plant submerged plant growth status in the above submerged plant submerged plant submerged plant submerged plant growth status includes the following: If the growth characteristics of submerged plants show a linear response relationship with the vegetation index NDRE, then a linear assessment model should be selected to evaluate the vegetation index NDRE. If the growth characteristics of submerged plants have a non-linear response relationship with the vegetation index NDRE, the power function evaluation model or the exponential evaluation model can be selected to evaluate the vegetation index NDRE based on the response trend. Among them, the power function evaluation model is suitable for scenarios where the influence of the index changes exponentially, while the exponential evaluation model is suitable for scenarios where the influence of the index increases exponentially or saturates.

[0039] This application employs a differentiated model selection strategy for assessing the growth status of submerged plants. Based on the relationship between the growth characteristics of submerged plants and their response to NDRE (Non-Dependent Growth Reduction), linear, power function, or exponential assessment models are matched accordingly. This ensures that the assessment models closely match the actual growth patterns of submerged plants, overcoming the problems of poor adaptability and large deviations in assessment results associated with traditional single assessment models. Linear models adapt to simple linear growth response relationships, ensuring assessment efficiency under normal growth conditions. Power function and exponential models, on the other hand, adapt to non-linear response scenarios involving power-law changes and exponential enhancement / saturation, respectively, accurately capturing the complex growth patterns of submerged plants. This differentiated model selection approach makes the assessment results of submerged plant growth status more closely aligned with reality, significantly improving the accuracy of the assessment and accurately reflecting the true growth status of submerged plants, providing a precise assessment basis for the subsequent development of targeted maintenance measures.

[0040] As an feasible approach, the formula for calculating the normalized red-edge vegetation index is: ; The calculation formula for the linear evaluation model is as follows: ; The calculation formula for the power function evaluation model is as follows: ; The formula for calculating the exponential model is as follows: ; Wherein, NIR represents the near-infrared band; RedEdge represents the red-edge band; B represents the submerged plant biomass (dryweight, g / m²) or coverage (%); and e represents the base of the natural logarithm. x and y This represents the fit coefficient. Model accuracy is evaluated using R², RMSE, and MRE.

[0041] (5) Input the preprocessed spectral data and water temperature data into the algal biomass prediction model to obtain the algal biomass prediction results; (6) Input the preprocessed spectral data and water quality information into the water transparency change model to obtain the prediction results of water transparency and dynamic change trend.

[0042] As one feasible approach, the formula for calculating algal biomass prediction models is as follows:

[0043] The formula for calculating water transparency prediction is as follows:

[0044] Chl-a represents the chlorophyll a concentration, usually expressed in μg / L or mg / m³; R 680R represents the surface reflectance in the red light band with a center wavelength of 680 nm; 710 R represents the reflectivity of the red-edge band with a center wavelength of 710 nm; 560 denoted by ρ(λ), which represents the reflectance of the green light band with a center wavelength of 560 nm; t and z represent the fitting coefficients of the algal biomass prediction model; SDD represents the water transparency; ρ(λ) represents the apparent reflectance of the water at wavelength λ; λ1 represents the wavelength of the green light band; λ2 represents the wavelength of the red light or near-infrared band; and m and n represent the fitting coefficients of the water transparency prediction model.

[0045] Step S130: Based on the data analysis results and combined with the intelligent decision-making algorithm, generate targeted maintenance decisions. The maintenance decisions include at least one of fertilization, pesticide application, replanting, harvesting, water purification, algae removal, and water level control. Specifically, relying on an expert knowledge base built from the experience of lake ecology specialists, numerous maintenance cases, and relevant scientific research findings, the system utilizes intelligent decision-making algorithms such as rule-based reasoning (RBR), case-based reasoning (CBR), and fuzzy logic reasoning to generate precise and personalized maintenance strategies. For example, when reduced water transparency and increased algal biomass are detected, potentially affecting the light conditions for submerged plants, the decision-making system combines real-time and historical data to analyze trends, predict potential impacts on submerged plant growth, and formulate corresponding water level control plans or initiate emergency measures in the restoration area based on the lake's hydrological conditions and the distribution of submerged plants. These measures include administering algaecides, increasing water flow, and lowering water levels. In the later stages of submerged plant restoration, when excessive plant biomass or species homogeneity are detected, potentially affecting the landscape or ecological balance, the system will promptly generate instructions for harvesting, replanting, and other control measures. Simultaneously, the system has an early warning function, capable of issuing alerts based on set thresholds for potential abnormal submerged plant growth, water quality deterioration, and algal blooms, reminding maintenance personnel to pay attention and take timely measures.

[0046] Specifically, in the early stages of submerged plant recovery, when water transparency falls below a set threshold or algal biomass exceeds a set threshold, the decision-making system will quickly formulate targeted maintenance decisions. For example, it might calculate a reasonable water level control plan based on the lake's hydrodynamic model to increase water transparency and improve light conditions for submerged plants; or it might activate emergency measures in the recovery area, such as using drones to deliver microbial agents or coordinating with water purification vessels, or guiding relevant departments to control human activities. If submerged plants in a certain area grow slowly and their health index falls below a threshold, the system will comprehensively analyze multispectral data, water quality data, light intensity data, and historical growth data to determine if it is due to a combination of factors, such as a lack of nutrients like nitrogen and phosphorus, unsuitable water pH, or insufficient light. Based on the specific conditions of the area, the decision-making system will use a multi-objective optimization algorithm to formulate a comprehensive maintenance plan, including appropriate fertilizer types and amounts, measures to adjust water pH, methods to improve light conditions, and specify the implementation time and sequence of each measure. After submerged plants survive and reproduce normally, they enter a community optimization phase. Based on monitoring data of changes in the submerged plant community, if species homogeneity may reduce ecosystem stability, the decision-making system will generate instructions for harvesting, replanting, and other regulatory measures. Once the submerged plant community structure is basically stable, it enters a routine maintenance phase. When excessive plant biomass or the onset of decline may affect the water surface landscape or water quality, the decision-making system will generate pruning or harvesting instructions. Simultaneously, based on daily drone patrols, early warnings are issued for potential abnormal growth of submerged plants, water quality deterioration, algal blooms, etc., reminding maintenance personnel to pay attention and take timely measures.

[0047] Step S140: Based on the maintenance decision, dispatch the operation drone or watercraft to perform the corresponding maintenance operation. The watercraft is equipped with an intelligent positioning system and special operation equipment.

[0048] The decision-making and execution are carried out by a multi-functional surface work vessel equipped with an intelligent positioning system, submerged plant thinning and harvesting equipment, water purification and algae removal equipment, microbial agent dispensing devices, and aquatic animal dispensing devices. Based on the decision-making instructions, the surface work vessel can efficiently perform maintenance operations in areas requiring care, such as using a robotic arm to precisely replant submerged plant seedlings, properly pruning overly dense submerged plants, promptly harvesting aging or dead plants, and trimming submerged plants that have grown above the water surface and are affecting the landscape. Simultaneously, the work vessel can also be equipped with water purification and algae removal equipment to extract water from areas with localized water quality deterioration or algae accumulation, and perform efficient coagulation and sedimentation filtration to quickly restore water quality and transparency.

[0049] Specifically, the drones equipped with a loudspeaker function promptly deter human activities such as sewage discharge and disturbance to the restored water areas. Upon receiving instructions, when the water transparency of open water areas fails to meet the light requirements for submerged plants, an early warning is sent to guide relevant departments to adjust lake levels reasonably using water conservancy facilities according to the water level control plan. For water areas where submerged plants are restored using enclosures, the drones, equipped with algae removal and purification equipment, can precisely treat the restored water, quickly improving water transparency and ensuring sufficient light for the submerged plants. Based on the multi-functional amphibious workboat, and following the decision-making instructions, suitable submerged plant varieties are precisely replanted in areas with monocultures according to the replanting plan to optimize the community structure. In areas where plants are dying or the biomass is excessive, timely harvesting or restoration is carried out. Maintenance execution equipment includes small precision fertilization and pesticide application devices mounted on drones, and specially designed multi-functional amphibious workboats. The drone fertilization and pesticide application devices employ high-precision flow control and positioning technology, enabling precise fertilization and pesticide application operations in designated areas according to the set dosage and method, based on instructions generated by the decision-making system. The multi-functional waterborne operation vessel is equipped with intelligent equipment for replanting, thinning, and harvesting submerged plants, as well as devices for releasing aquatic animals (such as herbivorous fish and snails). Based on decision-making instructions, the vessel can efficiently perform maintenance operations in areas requiring care. For example, it can use robotic arms to precisely replant submerged plant seedlings, properly prune overly dense submerged plants, and promptly harvest aging or dead plants. Simultaneously, by appropriately releasing aquatic animals, it can adjust the submerged plant community structure and maintain ecological balance.

[0050] As one possible approach, the method of this application further includes: after implementing maintenance measures, monitoring and acquiring relevant data again using drones, comparing the data with the data before maintenance to evaluate the maintenance effect; if the expected results are not achieved, optimizing and adjusting the maintenance strategy and repeating the maintenance operation to form a closed-loop feedback.

[0051] Specifically, at specific time intervals after the implementation of maintenance measures, comprehensive monitoring is conducted again using drones. The monitoring data from this second monitoring is then compared and analyzed in detail with the data from before maintenance. Statistical methods and ecological assessment indicators, such as the change rate of submerged plant coverage area, biomass growth rate, changes in community diversity index, increase in water transparency, and decrease in algal biomass, are used to evaluate the effectiveness of the maintenance measures. If the maintenance effect does not meet expectations, the reasons are analyzed in depth based on the assessment results, such as inadequate implementation of maintenance measures or unexpected changes in environmental factors. Machine learning algorithms are then used to optimize and adjust the maintenance strategy, and maintenance operations are repeated, forming a closed-loop feedback mechanism. Through continuous monitoring, evaluation, optimization, and execution, the maintenance effect of submerged plants in the lake is continuously improved, promoting the stable restoration and healthy development of the lake ecosystem. The technical solution provided by this invention has the following beneficial effects: To address the maintenance goals of submerged plants in lake restoration, this application utilizes unmanned aerial vehicle (UAV) remote sensing technology to implement refined management of the entire life cycle of submerged plants, from the initial planting stage to the stabilization period. Based on the growth needs and challenges faced by submerged plants at different stages, UAVs are used to precisely monitor various key indicators and promptly implement corresponding measures to ensure the healthy growth of submerged plants in a suitable environment, promoting the stable restoration and sustainable development of the lake ecosystem. By providing real-time dynamic monitoring and decision support, precise and scientific maintenance can be achieved, significantly improving the efficiency and quality of maintenance work and effectively reducing maintenance costs.

[0052] The following four examples will illustrate this in detail: Example 1 shows the maintenance of submerged plants in a small lake: For a small lake with an area of ​​approximately 1 square kilometer, a monitoring plan was developed based on the characteristic that its submerged plants are mainly distributed in the center of the lake and the shallow water area near the shore. The monitoring cycle was set to once a week, with a flight altitude of 50 meters and a flight speed of 10 meters per second. Professional Geographic Information System (GIS) software was used to plan the UAV flight path to ensure comprehensive coverage of the submerged plant area, with an overlap rate of 30% between adjacent flight paths. A high-resolution optical camera with a shooting resolution of 4000×3000 pixels was set, a multispectral camera to collect spectral data in four bands (red, green, blue, and near-infrared), and a thermal imaging camera with a temperature resolution of 0.1℃.

[0053] According to the monitoring plan, the monitoring drones take off at fixed times each week, flying along planned routes. During the flight, high-resolution optical cameras capture numerous clear visible light images, clearly identifying the types of submerged plants, such as *Myriophyllum spicatum* and *Ceratophyllum demersum*. Multispectral cameras acquire spectral data in different bands, thermal imaging cameras record the water temperature distribution, and GPS and IMU accurately record the flight trajectory and location information. The drones transmit the collected data to the ground control station in real time via wireless communication modules. The data processing server preprocesses the received data, stitching the images into a complete image of the lake's submerged plant area and removing noise. Using a trained image recognition algorithm, the coverage area of ​​submerged plants is identified as approximately 0.6 square kilometers, with varying growth densities in different areas. Analysis of the multispectral data reveals that the health index of submerged plants in areas near the shore is lower, potentially indicating nutrient deficiencies. Combined with thermal imaging data, the water temperature is within the normal range, having no significant adverse impact on the growth of submerged plants.

[0054] Based on data processing results, combined with an expert knowledge base and intelligent decision-making algorithms, the decision support system determined that submerged plant areas near the shore with low health indices needed supplementation with nutrients such as nitrogen and phosphorus. The system calculated that 50 kg of compound fertilizer was required for this area, to be applied via low-altitude, uniform spreading by drone. A specialized fertilization device was mounted on the drone, which, following instructions generated by the decision system, performed low-altitude fertilization in the designated area. During the fertilization process, the drone's flight altitude and speed were strictly controlled to ensure even distribution of fertilizer across the target area.

[0055] Two weeks after fertilization, monitoring was conducted again using drones. Comparing the data before and after fertilization, it was found that the growth of submerged plants in the area had improved, with a higher health index, more vibrant green leaves, and increased growth density. The assessment results indicated that the maintenance measures had achieved some effect, but some areas still required further optimization of the fertilization plan. Based on the assessment results, the decision support system adjusted the dosage and timing of subsequent fertilizations to continue the maintenance of the submerged plants in the lake.

[0056] Example 2 is about the maintenance of submerged plants in a medium-sized lake: For a medium-sized lake with an area of ​​approximately 5 square kilometers, submerged plants are widely distributed and there are several concentrated growth areas. GIS software was used to divide the lake into detailed zones, and satellite imagery and field surveys were combined to determine the species and growth characteristics of submerged plants in different areas. The monitoring cycle was set at once every 10 days during the early growth stage and once every 3 weeks during the stable growth stage. Fixed-wing UAVs were primarily used for rapid patrols of large areas; multi-rotor UAVs were used for detailed monitoring in key areas and areas sensitive to changes in submerged plant communities. The fixed-wing UAVs were set to a flight altitude of 80 meters and a flight speed of 15 m / s; the multi-rotor UAVs were set to a flight altitude of 30 meters and a flight speed of 5 m / s. GIS was used to plan flight routes to ensure comprehensive coverage of the submerged plant areas, with an overlap rate of 30% between adjacent flight routes. A high-resolution optical camera with a resolution of 6000×4000 pixels was used, a multispectral camera to collect spectral data in 8 bands, a thermal imaging camera with a temperature resolution of 0.05℃, and water quality sensors to monitor water quality parameters such as pH, DO, COD, TN, and TP in real time. Five fixed monitoring stations were set up around the lake to monitor meteorological parameters.

[0057] According to the monitoring plan, fixed-wing and multi-rotor UAVs took off on schedule and flew along the planned routes. During the flight, various sensors worked together. High-resolution optical cameras captured a large number of clear visible light images, multispectral cameras acquired rich spectral data, thermal imaging cameras recorded water temperature distribution, water quality sensors collected water quality data in real time, and GPS, IMU, and RTK devices accurately recorded flight trajectories and location information. Fixed monitoring stations simultaneously monitored meteorological parameters and transmitted the data to the ground control station in real time.

[0058] The drone rapidly transmits the collected data to the data processing server cluster at the ground control station via a 5G wireless communication module. The server preprocesses the data, including image stitching, noise reduction, and data calibration. Using trained image recognition algorithms and data analysis models, the submerged plant species were identified as *Myriophyllum spicatum*, *Vallisneria natans*, and *Hydrilla verticillata*, covering an area of ​​approximately 3 square kilometers, with significant variations in growth density across different areas. Multispectral data analysis revealed nitrogen and phosphorus deficiencies in some areas of the submerged plants; combined with water quality data, COD levels exceeded standards in some areas, potentially impacting plant growth. Big data analytics, comparing historical and real-time data, predicted potential pests and diseases in certain areas.

[0059] The decision support system, based on data processing results and combined with an expert knowledge base and intelligent decision-making algorithms, formulates maintenance strategies. For nutrient-deficient areas, the calculated required amount of nitrogen and phosphorus compound fertilizer is 150 kg, to be applied by low-altitude, uniform spreading via drone. For areas with excessive COD, a plan is developed to purify the water by introducing microbial agents and aquatic animals (such as freshwater mussels). For areas prone to pests and diseases, 80 liters of biological control agents are prepared, to be applied via drone spraying. The implementation time and sequence of each measure are determined: fertilization is carried out first, followed by the introduction of microbial agents and aquatic animals one week later, and a decision on whether to apply pesticides two weeks later based on pest and disease monitoring.

[0060] First, drones equipped with fertilization devices conduct low-altitude fertilization operations in nutrient-deficient areas according to instructions, strictly controlling flight altitude and speed to ensure even fertilizer distribution. A week later, a multi-functional watercraft releases microbial agents and freshwater mussels in areas with excessive COD. Two weeks later, based on the drone's monitoring results, drones carrying biological control agents are used to spray areas with pests and diseases. Throughout the maintenance process, drones and fixed monitoring stations are used for real-time tracking and monitoring to ensure that all measures are implemented as planned.

[0061] One month after fertilization, application of microbial agents, and pesticide application, a comprehensive monitoring was conducted again using drones. Comparing data before and after the maintenance, it was found that the growth of submerged plants in nutrient-deficient areas significantly improved, with leaves turning greener and growth density increasing; water quality in areas with excessive COD improved, and the impact on submerged plants lessened; pests and diseases were effectively controlled in pest-affected areas, alleviating damage to submerged plants. Statistical analysis showed that the submerged plant coverage area increased by 0.2 square kilometers, and the biomass growth rate reached 15%. The assessment results indicate that the maintenance measures achieved good results, but some areas still require further optimization, such as appropriately increasing the amount of microbial agents applied in some areas. The decision support system adjusted subsequent maintenance strategies based on the assessment results, continuously maintaining the submerged plants of the lake.

[0062] Example 3 is the maintenance of submerged plants in large lakes: For a large lake with an area of ​​10 square kilometers, considering the wide and complex distribution of submerged plants, the monitoring cycle was set to once every two weeks. A fixed-wing UAV was used, flying at an altitude of 100 meters and a speed of 20 meters per second. GIS software was used to plan complex flight routes, dividing the lake into multiple monitoring areas to ensure comprehensive coverage by the UAV, with an overlap rate of 25% between adjacent flight paths. A high-resolution optical camera with a shooting resolution of 6000×4000 pixels was used, a multispectral camera to collect spectral data in 7 bands, and a thermal imaging camera with a temperature resolution of 0.05℃.

[0063] The fixed-wing UAV took off according to the monitoring plan and flew along the planned route. During the flight, a high-resolution optical camera captured a massive amount of visible light images, a multispectral camera acquired rich spectral data, and a thermal imaging camera recorded the temperature distribution of the water. GPS and IMU accurately recorded the flight trajectory and location information. The UAV transmitted the collected data to the data processing server at the ground control station in real time. After preprocessing the data, the server used advanced image recognition algorithms and data analysis models to identify multiple species of submerged plants covering an area of ​​approximately 5 square kilometers. Through multispectral data analysis, it was found that some areas of submerged plants were affected by pests and diseases. Combined with thermal imaging data and water quality monitoring data, it was discovered that some areas had abnormal water temperatures and poor water quality, which adversely affected the growth of submerged plants.

[0064] Based on the data processing results, the decision support system formulated a comprehensive maintenance strategy. For areas affected by pests and diseases, biological control agents were selected, with a dosage of 100 liters, and the application method was spraying from a floating vessel. For areas with poor water quality and abnormal temperatures, the aquatic environment was improved by adding water purifiers and using artificial aeration.

[0065] Watercraft carrying biological control agents were deployed to spray pesticides in affected areas according to pre-defined routes and parameters. Simultaneously, specialized equipment was used to release water purifiers into areas with poor water quality, and artificial aeration equipment was activated.

[0066] One month after the implementation of maintenance measures, monitoring was conducted again using drones. Comparing the data before and after maintenance, it was found that pests and diseases were effectively controlled, damage to submerged plants was alleviated, water quality in some areas improved, and the growth of submerged plants gradually recovered. Based on the assessment results, the decision support system made appropriate adjustments to subsequent maintenance measures, continuing to provide precise maintenance for submerged plants in large lakes.

[0067] Example 4 illustrates the precise maintenance following the completion of a submerged plant restoration project for a city lake: For an urban lake with an area of ​​approximately 200,000 square meters, five species of submerged plants were planted in the water area within 1.5 meters below the normal water level through water level reduction and artificial planting, resulting in the restoration of about 42% of the lake's surface area. The distribution of submerged plants was complex and influenced by various factors. A combination of fixed-wing and multi-rotor UAVs was used for zoned monitoring according to different ecological areas and the characteristics of the submerged plant communities. The monitoring cycle was set at once every two days in the initial planting stage, once weekly after confirming the survival of the submerged plants and restoration to the normal water level, and once every two weeks after the aquatic plant project was accepted two years later. The fixed-wing UAVs flew at an altitude of 120 meters and a speed of 25 meters per second; the multi-rotor UAVs flew at an altitude of 40 meters and a speed of 6 meters per second. High-precision GIS technology was used to plan reasonable flight routes to ensure comprehensive coverage of the submerged plant area, with an overlap rate of 25% between adjacent flight routes. High-resolution optical cameras with a resolution of 8000×6000 pixels and multispectral cameras collected spectral data in 12 bands were used. A high-precision thermal imaging camera with a temperature resolution of 0.05℃ is installed. A water quality sensor module is added to monitor key water quality parameters in real time, including pH, dissolved oxygen (DO), chemical oxygen demand (COD), total nitrogen (TN), and total phosphorus (TP). Real-time acquisition and analysis of water quality data allows for timely understanding of the impact of aquatic environmental changes on submerged plant growth, providing crucial information for developing scientifically sound conservation strategies.

[0068] Fixed-wing and multi-rotor UAVs took off according to the monitoring plan and flew along the planned routes. During the flight, various sensors worked together. High-resolution optical cameras captured a large number of clear visible light images, multispectral cameras acquired rich spectral data, thermal imaging cameras recorded water temperature distribution, water quality sensors collected water quality data in real time, and GPS, IMU, and RTK devices accurately recorded flight trajectories and location information. The UAVs transmitted the collected data to the data processing server at the ground control station in real time. After preprocessing the data, the server used advanced image recognition algorithms and data analysis models to identify five species of submerged plants covering an area of ​​approximately 8 square kilometers.

[0069] Multispectral data analysis revealed an excessive biomass of submerged plants in an area of ​​approximately 1 square kilometer, with some plants beginning to decline. These plants were primarily *Hydrilla verticillata* and *Myriophyllum spicatum*. Combined with thermal imaging and water quality monitoring data, the area was found to have abnormal water temperature and poor water quality, negatively impacting the aquatic landscape.

[0070] Based on the data processing results, the decision support system generates harvesting instructions. For areas with large submerged plant biomass and areas where submerged plants are dying, harvesting is determined. Harvesting vessels are arranged to operate on sunny mornings, using a strip harvesting method along a pre-defined route. The harvesting width is 30%-50% of the working area width, leaving strips of healthy plants to form an alternating pattern of "harvested strips and preserved strips," with a stubble height of 10-12cm. The harvesting vessels travel at a low, constant speed, collecting all dead plants and debris. After harvesting, the harvested submerged plants are immediately transferred to a temporary storage area on the shore for further processing.

[0071] Within a week of implementing the maintenance measures, the concentrations of dissolved oxygen, ammonia nitrogen, and total phosphorus in the water were monitored daily. When dissolved oxygen fell below 5 mg / L, aeration equipment was activated to increase oxygen levels. When nitrogen and phosphorus concentrations increased, an appropriate amount of slow-release phosphorus removal agent specifically for submerged plants was added. One month after implementing the maintenance measures, monitoring was conducted again using drones. Comparing the data before and after the maintenance, it was found that the growth of submerged plants gradually recovered. Based on the assessment results, the decision support system made appropriate adjustments to subsequent maintenance measures, continuing to provide precise maintenance for submerged plants in large lakes.

[0072] The following describes an embodiment of the lake restoration system based on unmanned aerial vehicles (UAVs) according to the present invention. The system includes: a ground control station, a monitoring UAV, several fixed monitoring stations, water quality sensors, operational UAVs, and a surface vessel for operations. Several fixed monitoring stations are deployed around the target lake to collect meteorological information. Water quality sensors are deployed in the target lake to collect water quality information. Monitoring drones are used to collect image data, spectral data, water temperature data, and spatial status information of the target lake. A surface-mounted work vessel is equipped with an intelligent positioning system and specialized operating equipment to perform maintenance operations. A ground control station communicates with the monitoring drones, fixed monitoring stations, water quality sensors, work drones, and surface-mounted work vessels. The ground control station acquires multi-source heterogeneous data from the target lake, preprocesses the data, and analyzes the preprocessed data using deep learning image recognition algorithms and multivariate data analysis algorithms. Based on the data analysis results and intelligent decision-making algorithms, targeted maintenance decisions are generated, and work drones and / or surface-mounted work vessels are dispatched to perform corresponding maintenance operations based on these decisions. Among them, multi-source heterogeneous data includes image data, spectral data, water temperature data, spatial status information, meteorological information, and water quality information; maintenance decisions include at least one of fertilization, pesticide application, replanting, harvesting, water purification, and water level control.

[0073] This application utilizes several fixed monitoring stations and water quality sensors to achieve continuous, point-to-point collection of meteorological and water quality data, while monitoring drones enable dynamic data collection across the entire lake area, forming a three-dimensional monitoring network of "fixed + mobile" to ensure comprehensive and real-time perception of the lake's ecological condition. The ground control station, as the core of the system, achieves interconnection with various devices, enabling centralized data acquisition, unified processing, and in-depth analysis. It also integrates intelligent decision-making algorithms to generate maintenance decisions and schedule operational equipment, achieving centralized and intelligent management of the system and overcoming the problems of dispersed and poorly coordinated traditional lake restoration equipment. The coordinated configuration of operational drones and multi-functional watercraft balances precise, small-scale maintenance operations with large-scale, professional maintenance work. The intelligent positioning and specialized equipment of the multi-functional watercraft ensure the precise execution of maintenance operations. The entire system achieves a closed-loop operation of data collection, analysis, decision-making, and execution. Each device has a clear division of labor and works efficiently, significantly improving the intelligence level and operational efficiency of lake restoration, and realizing refined and scientific maintenance of the lake ecosystem.

[0074] As one feasible method, ground control stations are also used for: A remote sensing monitoring plan is developed based on the geographic information and submerged plant distribution of the target lake. The remote sensing monitoring plan includes the monitoring range, monitoring frequency, preset flight routes, flight parameters and shooting parameters. According to the remote sensing monitoring plan, the monitoring drone is controlled to collect image data, spectral data, water temperature data and spatial status information of the target lake.

[0075] As one feasible approach, acquiring multi-source heterogeneous data of the target lake specifically includes the following: A remote sensing monitoring plan is formulated based on the geographic information and the distribution of submerged plants of the target lake. The remote sensing monitoring plan includes the monitoring range, monitoring frequency, preset flight route, flight parameters and shooting parameters. According to the remote sensing monitoring plan, a monitoring drone is scheduled to collect the image data, the spectral data, the water temperature data, and the spatial status information; Acquire meteorological information collected by fixed monitoring stations deployed around the target lake and water quality information collected by water quality sensors deployed in the target lake.

[0076] As one possible approach, the monitoring drone includes multi-rotor drones and fixed-wing drones, and the monitoring drone integrates a satellite navigation and positioning system, an inertial measurement unit, and an RTK device for real-time monitoring of the drone's spatial status information. The monitoring drone is also equipped with an optical camera, a multispectral camera, and a thermal imaging camera. The optical camera is used to collect the image data, the multispectral camera is used to collect the spectral data, and the thermal imaging camera is used to collect the water temperature data.

[0077] As one possible approach, the preprocessing includes: format conversion, spatiotemporal stitching, noise reduction, and calibration.

[0078] As one possible approach, the analysis of preprocessed data using deep learning image recognition algorithms and multivariate data analysis algorithms includes: The pre-processed image data is analyzed using a trained image recognition algorithm to identify the types, coverage area, and growth density of submerged plants in the target lake. Based on data analysis algorithms, a comprehensive assessment model for the growth status of submerged plants, a prediction model for algal biomass, and a model for changes in water transparency were constructed. The preprocessed spectral data are subjected to normalized red-edge vegetation index (NDRE) calculation to obtain the vegetation index. The vegetation index NDRE is evaluated based on the comprehensive evaluation model for the growth status of submerged plants to obtain the comprehensive evaluation results for the growth status of submerged plants. The preprocessed spectral data and water temperature data are input into the algal biomass prediction model to obtain the algal biomass prediction results. The preprocessed spectral data and water quality information are input into the water transparency change model to obtain the prediction results of water transparency and dynamic change trend.

[0079] As one possible approach, the evaluation of the vegetation index NDRE based on the comprehensive evaluation model for the growth status of submerged plants includes: If the growth characteristics of submerged plants show a linear response relationship with the vegetation index NDRE, then a linear evaluation model is selected to evaluate the vegetation index NDRE. If the growth characteristics of submerged plants have a non-linear response relationship with the vegetation index NDRE, the vegetation index NDRE can be evaluated by selecting a power function evaluation model or an exponential evaluation model based on the response trend. The power function evaluation model is suitable for scenarios where the influence of the index changes exponentially, while the exponential evaluation model is suitable for scenarios where the influence of the index increases exponentially or saturates.

[0080] As one feasible approach, the formula for calculating the normalized red-edge vegetation index is: ; The calculation formula for the linear evaluation model is as follows: ; The calculation formula for the power function evaluation model is as follows: ; The formula for calculating the exponential model is as follows: ; Wherein, NIR represents the near-infrared band; RedEdge represents the red-edge band; B represents the biomass or cover of submerged plants; and e represents the base of the natural logarithm. x and y This represents the fitting coefficient.

[0081] As one feasible approach, the algal biomass prediction model is calculated using the following formula: ; The calculation formula for the water transparency prediction model is as follows: ; Where Chl-a represents the concentration of chlorophyll a; R 680 R represents the surface reflectance in the red light band with a center wavelength of 680 nm; 710 R represents the reflectivity of the red-edge band with a center wavelength of 710 nm; 560 denoted by ρ(λ), which represents the reflectance of the green light band with a center wavelength of 560 nm; t and z represent the fitting coefficients of the algal biomass prediction model; SDD represents the water transparency; ρ(λ) represents the apparent reflectance of the water at wavelength λ; λ1 represents the wavelength of the green light band; λ2 represents the wavelength of the red light or near-infrared band; and m and n represent the fitting coefficients of the water transparency prediction model.

[0082] Specifically, this application endows the ground control station with the functions of remote sensing monitoring plan formulation and UAV monitoring control, making the ground control station the command center of the entire monitoring process and realizing standardized and intelligent management and control of UAV monitoring work. Based on lake geographic information and the distribution of submerged plants, a remote sensing monitoring plan is formulated, including elements such as monitoring range, frequency, and flight path, making UAV monitoring work more targeted and planned, avoiding resource waste and invalid data caused by blind monitoring. The ground control station directly controls the monitoring UAV to complete data collection according to the plan, ensuring the strict execution of monitoring work, realizing standardized management of the monitoring process, and ensuring the comparability and continuity of monitoring data from different times and batches. This design achieves integrated management and control of monitoring plan formulation and monitoring execution, improving the efficiency and standardization of UAV monitoring work, ensuring the consistency and systematic nature of collected data, and providing a standardized and comparable data source for subsequent data analysis and maintenance decisions, further improving the operational efficiency and accuracy of the entire lake restoration system.

[0083] In the embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0084] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0085] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0086] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0087] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for lake restoration based on unmanned aerial vehicles (UAVs), characterized in that, The method includes: Acquire multi-source heterogeneous data of the target lake, including image data, spectral data, water temperature data, spatial state information, meteorological information, and water quality information; The multi-source heterogeneous data is preprocessed, and the preprocessed data is analyzed using deep learning image recognition algorithms and multivariate data analysis algorithms. Based on data analysis results and intelligent decision-making algorithms, targeted maintenance decisions are generated, including at least one of fertilization, pesticide application, replanting, harvesting, water purification, algae removal, and water level control. Based on the maintenance decision, the operation drone or watercraft is dispatched to perform the corresponding maintenance operation. The watercraft is equipped with an intelligent positioning system and special operating equipment.

2. The lake restoration method based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The acquisition of multi-source heterogeneous data of the target lake includes: A remote sensing monitoring plan is formulated based on the geographic information and the distribution of submerged plants of the target lake. The remote sensing monitoring plan includes the monitoring range, monitoring frequency, preset flight route, flight parameters and shooting parameters. According to the remote sensing monitoring plan, a monitoring drone is scheduled to collect the image data, the spectral data, the water temperature data, and the spatial status information; Acquire meteorological information collected by fixed monitoring stations deployed around the target lake and water quality information collected by water quality sensors deployed in the target lake.

3. The lake restoration method based on unmanned aerial vehicles (UAVs) according to claim 2, characterized in that, The monitoring drones include multi-rotor drones and fixed-wing drones. The monitoring drones are equipped with satellite navigation and positioning systems, inertial measurement units, and RTK devices for real-time monitoring of the drone's spatial status information. The monitoring drone is also equipped with an optical camera, a multispectral camera, and a thermal imaging camera. The optical camera is used to collect the image data, the multispectral camera is used to collect the spectral data, and the thermal imaging camera is used to collect the water temperature data.

4. The lake restoration method based on unmanned aerial vehicles (UAVs) according to claim 1, characterized in that, The preprocessing includes: format conversion, spatiotemporal stitching, noise reduction, and calibration.

5. The lake restoration method based on unmanned aerial vehicles (UAVs) according to claim 3, characterized in that, The analysis of the preprocessed data using deep learning image recognition algorithms and multivariate data analysis algorithms includes: The pre-processed image data is analyzed using a trained image recognition algorithm to identify the types, coverage area, and growth density of submerged plants in the target lake. Based on data analysis algorithms, a comprehensive assessment model for the growth status of submerged plants, a prediction model for algal biomass, and a model for changes in water transparency were constructed. The preprocessed spectral data are subjected to normalized red-edge vegetation index (NDRE) calculation to obtain the vegetation index. The vegetation index NDRE is evaluated based on the comprehensive evaluation model for the growth status of submerged plants to obtain the comprehensive evaluation results for the growth status of submerged plants. The preprocessed spectral data and water temperature data are input into the algal biomass prediction model to obtain the algal biomass prediction results. The preprocessed spectral data and water quality information are input into the water transparency change model to obtain the prediction results of water transparency and dynamic change trend.

6. The lake restoration method based on unmanned aerial vehicles (UAVs) according to claim 5, characterized in that, The evaluation of the vegetation index NDRE based on the comprehensive evaluation model for the growth status of submerged plants includes: If the growth characteristics of submerged plants show a linear response relationship with the vegetation index NDRE, then a linear evaluation model is selected to evaluate the vegetation index NDRE. If the growth characteristics of submerged plants have a non-linear response relationship with the vegetation index NDRE, the vegetation index NDRE is evaluated by selecting a power function evaluation model or an exponential evaluation model based on the response trend; wherein, the power function evaluation model is suitable for scenarios where the influence of the index changes exponentially, and the exponential evaluation model is suitable for scenarios where the influence of the index increases exponentially or saturates.

7. The lake restoration method based on unmanned aerial vehicles (UAVs) according to claim 6, characterized in that, The formula for calculating the normalized red-edge vegetation index is as follows: ； The calculation formula for the linear evaluation model is as follows: ; The calculation formula for the power function evaluation model is as follows: ; The formula for calculating the exponential model is as follows: ; Wherein, NIR represents the near-infrared band; RedEdge represents the red-edge band; B represents the biomass or cover of submerged plants; and e represents the base of the natural logarithm. x and y This represents the fitting coefficient.

8. The lake restoration method based on unmanned aerial vehicles (UAVs) according to claim 7, characterized in that, The calculation formula for the algal biomass prediction model is as follows: The calculation formula for the water transparency prediction model is as follows: Where Chl-a represents the concentration of chlorophyll a; R 680 R represents the surface reflectance in the red light band with a center wavelength of 680 nm. 710 R represents the reflectivity in the red-edge band with a center wavelength of 710 nm; 560 denoted by ρ(λ), which represents the reflectance of the green light band with a center wavelength of 560 nm; t and z represent the fitting coefficients of the algal biomass prediction model; SDD represents the water transparency; ρ(λ) represents the apparent reflectance of the water at wavelength λ; λ1 represents the wavelength of the green light band; λ2 represents the wavelength of the red light or near-infrared band; and m and n represent the fitting coefficients of the water transparency prediction model.

9. A lake restoration system based on unmanned aerial vehicles (UAVs), characterized in that, The system includes: Ground control station, monitoring drones, several fixed monitoring stations, water quality sensors, operational drones, and waterborne operations vessels; Several fixed monitoring stations are deployed around the target lake to collect meteorological information for the area; water quality sensors are deployed in the target lake to collect water quality information; monitoring drones are used to collect image data, spectral data, water temperature data, and spatial status information of the target lake; the surface work vessel is equipped with an intelligent positioning system and dedicated operating equipment to perform maintenance operations; the ground control station is communicatively connected to the monitoring drones, fixed monitoring stations, water quality sensors, operating drones, and surface work vessels; the ground control station is used to acquire multi-source heterogeneous data of the target lake, preprocess the multi-source heterogeneous data, and analyze the preprocessed data using deep learning image recognition algorithms and multivariate data analysis algorithms; based on the data analysis results and combined with intelligent decision-making algorithms, targeted maintenance decisions are generated, and the operating drones and / or surface work vessels are dispatched to perform corresponding maintenance operations based on the maintenance decisions; The multi-source heterogeneous data includes the image data, the spectral data, the water temperature data, the spatial state information, the meteorological information, and the water quality information; the maintenance decision includes at least one of fertilization, pesticide application, replanting, harvesting, water purification, and water level control.

10. The UAV-based lake restoration system according to claim 9, characterized in that, The ground control station is also used for: A remote sensing monitoring plan is formulated based on the geographic information and submerged plant distribution of the target lake. The remote sensing monitoring plan includes the monitoring range, monitoring frequency, preset flight route, flight parameters and shooting parameters. According to the remote sensing monitoring plan, the monitoring drone is controlled to collect image data, spectral data, water temperature data and spatial status information of the target lake.

Images