Weeding monitoring system and method based on unmanned aerial vehicle

Abstract

The invention relates to the technical field of agricultural automation and intelligent control, in particular to a weeding monitoring system and method based on an unmanned aerial vehicle, and the method comprises the steps: collecting spectral reflection data and microwave internal structure data of a target area through a multispectral camera and a microwave sensor carried by the unmanned aerial vehicle; the system generates standardized perception data through a data fusion module, and performs precise classification and recognition of weeds in combination with feature extraction, adaptive fuzzy clustering and multi-model collaborative learning technologies; based on a spatial distribution diagram of weeds, the system predicts a weed diffusion trend through a graph convolutional network, and an optimal weeding strategy is generated in combination with environmental data; and the unmanned aerial vehicle executes the weeding task through the optimized flight path, monitors the weeding effect in real time, continuously optimizes the strategy, and ensures efficient execution of the weeding task.

Description

Technical Field

[0001] The present invention relates to the field of agricultural automation and intelligent control technology, and in particular to a weeding monitoring system and method based on an unmanned aerial vehicle. Background Art

[0002] Drones are increasingly being used in agriculture, especially in weeding monitoring. Traditional weeding methods usually rely on manual or mechanical inspections and operations, which are inefficient and wasteful of manpower and material resources. Drone-based weeding systems can use remote sensing technology and sensors to achieve real-time monitoring, identification, and evaluation of weeds in farmland, thereby effectively improving the accuracy and efficiency of weeding. This automated monitoring method can not only reduce dependence on chemical herbicides, but also reduce operating costs, increase crop yields, and be environmentally friendly.

[0003] The existing drone weeding monitoring technology (Chinese invention patent, publication number: CN118247643B, name: A drone-based weeding monitoring and evaluation method and system) mainly relies on computer vision technology and deep learning models, and adopts a monitoring method based on visual and behavioral trajectory semantics. However, this technical means has many shortcomings in practical applications. The existing technology usually requires multiple data collection and repeated analysis, which makes it difficult to capture the dynamic process of weed spread in real time, resulting in the lag of weed control strategies. In addition, the existing system has limited integration of environmental data and cannot flexibly respond to changes in the spread of weeds under different environmental conditions. Summary of the invention

[0004] In view of the many problems existing in the above-mentioned prior art, the present invention provides a weeding monitoring system and method based on unmanned aerial vehicles. The present invention collects spectral reflection data and microwave internal structure data of the target area by carrying a multispectral camera and a microwave sensor on the unmanned aerial vehicle. Through data fusion, feature extraction, convolutional neural network and recursive neural network and other technologies, the system can accurately identify the spatial distribution and diffusion trend of weeds, generate the optimal weeding strategy and optimize the flight path. The system can monitor the weeding effect in real time and adjust the strategy according to feedback, thereby realizing accurate weeding management of farmland.

[0005] A weeding monitoring system based on a drone includes the following steps:

[0006] The drone platform is equipped with a multispectral camera and a microwave sensor to collect spectral reflectance data and microwave internal structure data of the target area;

[0007] A data fusion module receives the spectral reflection data and the microwave internal structure data, and generates standardized perception data including spectral reflection characteristics and internal structure information;

[0008] A feature extraction module is used to extract features from the standardized perception data, generate spectral feature data and microwave feature data respectively, expand the data dimension through nonlinear transformation, classify the feature data using adaptive fuzzy clustering, and generate preliminary feature classification data;

[0009] A multi-model collaborative learning system receives the preliminary feature classification data, extracts spatial feature data through a convolutional neural network, performs time series feature analysis in combination with a recursive neural network, generates fused feature data, and outputs final weed classification data through feature reconstruction network optimization;

[0010] A spatial distribution map construction module is used to construct a spatial distribution map of weeds based on the final weed classification data, and to analyze the spatial relationship between weed nodes using a graph convolutional network, to predict the weed diffusion trend, to generate diffusion trend data, and to perform density analysis on the weed distribution in the region, to generate density data;

[0011] The weeding strategy generation module generates preliminary weeding strategy data by combining the diffusion trend data, density data and real-time environmental data, optimizes the flight path of the drone, and performs weeding tasks. It also generates weeding effect data through real-time monitoring and optimizes the strategy based on the weeding effect data.

[0012] Preferably, the data fusion module includes:

[0013] A processing unit that receives spectral reflection data and microwave internal structure data generates standardized perception data using a weighted average method, wherein the calculation expression of the standardized perception data is:

[0014]

[0015] Among them, S is the standardized perception data; R is the spectral reflectance data; M is the microwave internal structure data; w1 is the weight coefficient of the spectral data; w2 is the weight coefficient of the microwave data.

[0016] Preferably, the feature extraction module includes:

[0017] A nonlinear transformation unit, used for performing a nonlinear function transformation on the standardized perception data to expand the data dimension and generate expanded feature data;

[0018] The adaptive fuzzy clustering unit is used to perform fuzzy clustering analysis on the expanded feature data to generate preliminary feature classification data. The clustering formula of the fuzzy clustering is:

[0019]

[0020] Among them, μ ij is the membership of the jth sample in the i-th cluster; dij is the distance between the sample and the cluster center; C is the number of clusters; m is the fuzzy index, which is used to control the fuzziness of clustering.

[0021] Preferably, the multi-model collaborative learning system includes:

[0022] A convolutional neural network module, used to process the preliminary feature classification data, and the extracted spatial feature data reflects the shape, color and other visual features of the plant;

[0023] Recursive neural network module, used to analyze time series feature data and identify plant growth trends and cycles.

[0024] Preferably, the fused feature data is generated by weighted fusion of the spatial feature data and the time series feature data, and the fused feature data is used to improve the classification accuracy of the final weed classification data.

[0025] Preferably, the spatial distribution map construction module is used to construct a spatial distribution map of weeds based on the node information in the final weed classification data, and the spatial distribution map is used to visually display the distribution of weeds in the target area.

[0026] Preferably, the graph convolutional network is used to analyze the weed nodes in the spatial distribution map, and the generated diffusion trend data includes the diffusion rate and diffusion direction of each weed node, which is used to guide subsequent weed control strategies.

[0027] Preferably, the weed control strategy generation module generates preliminary weed control strategy data by combining the diffusion trend data, density data and real-time environmental data using a decision tree algorithm, wherein the decision tree algorithm makes weed control decisions based on weed density and diffusion potential.

[0028] Preferably, the weeding strategy generation module further includes a real-time monitoring submodule for monitoring the execution of the weeding task and generating real-time weeding effect data, wherein the real-time weeding effect data is used to evaluate the weeding effect and optimize subsequent weeding strategies.

[0029] A method for executing the drone-based weeding monitoring system comprises the following steps:

[0030] Collect spectral reflection data and microwave internal structure data of the target area;

[0031] receiving the spectral reflection data and the microwave internal structure data, and generating standardized perception data including spectral reflection characteristics and internal structure information;

[0032] Extracting features from the standardized perception data to generate spectral feature data and microwave feature data, respectively, expanding the data dimension through nonlinear transformation, and classifying the feature data using adaptive fuzzy clustering to generate preliminary feature classification data;

[0033] Receiving the preliminary feature classification data, extracting spatial feature data through a convolutional neural network, performing time series feature analysis in combination with a recursive neural network, generating fused feature data, and outputting final weed classification data through feature reconstruction network optimization;

[0034] Based on the final weed classification data, a spatial distribution map of weeds is constructed, and a graph convolutional network is used to analyze the spatial relationship between weed nodes, predict the weed diffusion trend, generate diffusion trend data, and perform density analysis on the weed distribution in the area to generate density data;

[0035] Combined with the diffusion trend data, density data and real-time environmental data, preliminary weeding strategy data is generated, the flight path of the drone is optimized, and the weeding task is performed. Real-time monitoring is performed to generate weeding effect data, and strategy optimization is performed based on the weeding effect data.

[0036] Compared with the prior art, the advantages and beneficial effects of the present invention are:

[0037] The present invention achieves accurate identification and real-time monitoring of weeds by combining multispectral cameras and microwave sensor technology. The system generates standardized perception data including spectral reflectance characteristics and internal structure information through a data fusion module, effectively improving the ability to analyze plant health status and weed spread trends;

[0038] The present invention solves the defect of the prior art that weed spread cannot be monitored dynamically in real time through adaptive fuzzy clustering technology and a multi-model collaborative learning system, thereby improving the accuracy of weed identification and the response speed of weed control strategies.

[0039] By optimizing the flight path of the UAV and combining it with real-time environmental data, the system significantly improves the efficiency and accuracy of weeding operations, reduces the use of pesticides, and protects the ecological environment. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a structural block diagram of the system of the present invention;

[0041] Figure 2 Schematic diagram of feature extraction and learning process in the present invention;

[0042] Figure 3 It is a schematic diagram of the spatial distribution and diffusion analysis process in the present invention;

[0043] Figure 4It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0044] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is obvious that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0045] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0046] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.

[0047] like Figure 1 As shown, a weeding monitoring system based on a drone includes the following steps:

[0048] The drone platform is equipped with a multispectral camera and a microwave sensor to collect spectral reflectance data and microwave internal structure data of the target area;

[0049] In the present invention, the UAV platform is the core component, which is equipped with a multispectral camera and a microwave sensor, and is mainly used to collect spectral reflectance data and microwave internal structure data of the target area. The principle of this configuration is based on remote sensing technology, which captures and analyzes the optical properties of plants and their surroundings to obtain the health status of vegetation and weed distribution in real time.

[0050] Multispectral cameras use light of different wavelengths to capture reflected spectral information, and the information in each band can reveal the physiological state of the plant. The reflectivity of plants to the spectrum varies depending on factors such as their chlorophyll content, water content, and leaf structure. By analyzing these reflection data, an assessment of the health of the plants can be achieved. For example, healthy plants will show higher reflectivity in specific bands (such as red light and near-infrared light), while weeds or damaged plants may exhibit different reflective characteristics. Combined with the generation of standardized perception data, the system can effectively distinguish between different plant types, identify the presence of weeds, and assess their growth status.

[0051] Microwave sensors can deeply detect the internal structure of plants and provide information about soil moisture and root distribution by emitting microwave signals and receiving their echoes. The advantage of microwave technology lies in its ability to penetrate the soil and the internal structure of plants. This feature enables it to work stably under different soil conditions and is not affected by light and climate change. By combining the data obtained by microwave sensors, the system can more comprehensively evaluate the growth environment and potential spread of weeds.

[0052] Combining the data from these two sensors, the drone platform can generate standardized perception data of the target area in real time, forming a high-resolution plant distribution map. This monitoring method significantly improves the efficiency of weed management, so that the growth status of crops and the growth dynamics of weeds can be reflected in real-time data analysis, ultimately helping farmers develop more accurate weed control strategies, reduce the use of pesticides, and improve the economic benefits and ecological sustainability of operations.

[0053] For example, in one specific implementation, drones fly over wheat fields, and the collected spectral reflectance data and microwave data are processed in real time, and the distribution of wheat and weeds is analyzed through image recognition technology. After system analysis, it is found that the weed density in a certain area is high, and the decision support system will recommend farmers to prioritize mechanical or chemical weeding in this area. Through this intelligent monitoring and management, farmers can more effectively control weed growth and ensure the healthy growth of wheat, thereby achieving higher yields and economic benefits.

[0054] A data fusion module receives the spectral reflection data and the microwave internal structure data, and generates standardized perception data including spectral reflection characteristics and internal structure information;

[0055] Preferably, the data fusion module includes:

[0056] A processing unit that receives spectral reflection data and microwave internal structure data generates standardized perception data using a weighted average method, wherein the calculation expression of the standardized perception data is:

[0057]

[0058] Among them, S is the standardized perception data; R is the spectral reflectance data; M is the microwave internal structure data; w1 is the weight coefficient of the spectral data; w2 is the weight coefficient of the microwave data.

[0059] In the present invention, the data fusion module is one of the key components, and its main function is to receive spectral reflectance data and microwave internal structure data, and generate standardized perception data including spectral reflectance characteristics and internal structure information. The principle of this module is based on data fusion technology, which aims to combine data collected by different sensors to improve the overall understanding and analysis capabilities of plant conditions in the target area.

[0060] Specifically, the processing unit of the data fusion module generates standardized perception data through the weighted average method. The use of the weighted average method ensures the comprehensive consideration of different data sources, thereby establishing a reasonable connection between multispectral and microwave data. In this process, the spectral reflectance data (R) provides surface feature information of plants, while the microwave internal structure data (M) provides deep information of plants and soil. The calculation expression of standardized perception data is shown above. Through the above calculation expression, the system can dynamically adjust the influence of each data source in the final output to ensure that the fusion result more accurately reflects the actual situation of the target area.

[0061] For example, in a monitoring of a specific farmland, the spectral reflectance data collected by the drone during flight can effectively identify the growth status of the plants, but this data alone cannot provide in-depth understanding of soil moisture or root conditions. Microwave sensors can obtain this information through the soil. Through the data fusion module, the combination of spectral reflectance data and microwave internal structure data enables the final generated standardized perception data to contain comprehensive information on the health status of vegetation and its growth environment. This integration of information not only improves the accuracy and reliability of the data, but also provides a solid foundation for subsequent weed identification and weed control strategy generation.

[0062] Through such data fusion, the system can achieve more intelligent decision support. For example, when the system detects a high weed density in a certain area and analyzes the soil moisture conditions in the area in combination with microwave data, it can recommend targeted weed control measures to maximize resource utilization efficiency and weed control effect.

[0063] A feature extraction module is used to extract features from the standardized perception data, generate spectral feature data and microwave feature data respectively, expand the data dimension through nonlinear transformation, classify the feature data using adaptive fuzzy clustering, and generate preliminary feature classification data;

[0064] like Figure 2 As shown, preferably, the feature extraction module includes:

[0065] A nonlinear transformation unit, used for performing a nonlinear function transformation on the standardized perception data to expand the data dimension and generate expanded feature data;

[0066] The adaptive fuzzy clustering unit is used to perform fuzzy clustering analysis on the expanded feature data to generate preliminary feature classification data. The clustering formula of the fuzzy clustering is:

[0067]

[0068] Among them, μ ij is the membership of the jth sample in the i-th cluster; d ij is the distance between the sample and the cluster center; C is the number of clusters; m is the fuzzy index, which is used to control the fuzziness of clustering.

[0069] In the present invention, the feature extraction module is a crucial component responsible for extracting useful features from the standardized perception data for subsequent analysis and decision-making. The principle of this module is to accurately extract spectral feature data and microwave feature data through a combination of nonlinear transformation and adaptive fuzzy clustering, thereby achieving effective recognition and classification of vegetation and weeds in the target area.

[0070] The nonlinear transformation unit in the feature extraction module transforms the standardized perception data using nonlinear functions. The purpose of this process is to expand the dimension of the data so that the data appears richer and more complex in multidimensional space. Through this transformation, the potential features in the data can be better captured, especially those that cannot be simply expressed through linear transformation. For example, when processing spectral reflectance data, the reflectance characteristics of some plants at specific wavelengths may change due to lighting conditions and health status. Nonlinear transformation helps to extract the characteristics of these changes, thereby providing a more detailed data basis for subsequent classification.

[0071] The adaptive fuzzy clustering unit performs fuzzy clustering analysis on the extended feature data after nonlinear transformation. The advantage of fuzzy clustering is that it allows a sample to have different degrees of membership in multiple clusters, which is particularly effective for dealing with complex plant species and growth states. In this way, the system can more flexibly identify weeds and crops in mixed vegetation areas.

[0072] For example, suppose weeds and wheat grow intertwined in a field. The feature extraction module can classify wheat and weeds using fuzzy clustering based on the different features of spectral and microwave data. The system may mark the plant feature data in a certain area as belonging to both the "weed" cluster and the "wheat" cluster, but reflecting their main features at different membership levels. This flexibility enables the system to provide farmers with more accurate weed identification information, thereby formulating more effective weed control strategies.

[0073] A multi-model collaborative learning system receives the preliminary feature classification data, extracts spatial feature data through a convolutional neural network, performs time series feature analysis in combination with a recursive neural network, generates fused feature data, and outputs final weed classification data through feature reconstruction network optimization;

[0074] Preferably, the multi-model collaborative learning system includes:

[0075] A convolutional neural network module, used to process the preliminary feature classification data, and the extracted spatial feature data reflects the shape, color and other visual features of the plant;

[0076] Recursive neural network module, used to analyze time series feature data and identify plant growth trends and cycles.

[0077] In the present invention, the multi-model collaborative learning system is a key intelligent analysis component, which is designed to efficiently process preliminary feature classification data and extract valuable information to support decision-making. The system includes a convolutional neural network module and a recursive neural network module, which are used to process spatial feature data and time series feature data respectively.

[0078] The convolutional neural network module specializes in processing the extracted preliminary feature classification data, which includes the shape, color and other visual features of the plants. Through its hierarchical structure, the convolutional neural network can automatically extract important features in the image, for example, identifying information such as the edges, textures and shapes of different plants. For farmland monitored by drones, this means that the system can effectively distinguish between weeds and crops. For example, through training, the convolutional neural network can learn to recognize the characteristics of wheat and common weeds, thereby generating detailed plant distribution maps during the monitoring process. The accuracy of this spatial feature extraction significantly improves the accuracy of weed identification and provides reliable data support for subsequent weed control strategies.

[0079] The recursive neural network module is responsible for analyzing time series feature data to identify trends and cycles of plant growth changes. Since the growth process of plants is dynamic, recursive neural networks can process sequence data and capture their time dependencies, thereby effectively analyzing the growth cycle and change trends of plants. This is crucial for agricultural management, because understanding the changes in plant growth status can help farmers weed or fertilize at the right time. For example, the system can monitor the growth rate of weeds and predict their future growth trends, and combine environmental data with historical data to provide farmers with advice, such as implementing mechanical weeding during the peak period of weed growth.

[0080] Combining the outputs of these two modules, the system can form a comprehensive plant growth analysis model to achieve accurate monitoring and management of weeds and crops. Through this intelligent analysis method, the drone monitoring system not only improves the efficiency of crop management, but also effectively reduces the use of chemical herbicides and promotes the sustainable development of agriculture. In short, the multi-model collaborative learning system provides powerful data processing and analysis capabilities for drone weeding monitoring through the effective combination of convolutional neural networks and recurrent neural networks, helping farmers better deal with weed problems and optimize agricultural production.

[0081] Preferably, the fused feature data is generated by weighted fusion of the spatial feature data and the time series feature data, and the fused feature data is used to improve the classification accuracy of the final weed classification data.

[0082] In the present invention, the generation of fused feature data is a key link, which aims to improve the accuracy of the final weed classification data. The fused feature data is generated by weighted fusion of spatial feature data and time series feature data. This process is based on the idea of ​​multi-source information integration in principle, aiming to make full use of the advantages of different data sources.

[0083] The spatial feature data is extracted by the convolutional neural network, reflecting the shape, color and other visual features of the vegetation, while the time series feature data is analyzed by the recursive neural network to reveal the trend and cycle of plant growth changes. These two types of feature data each provide important information. The spatial feature data can accurately capture the static characteristics of plants, while the time series feature data can reflect the dynamic changes of plants. By weighted fusion of these two types of feature data, a more comprehensive feature set can be generated, thereby improving the classification ability of weeds and crops.

[0084] In practical applications, the process of generating fused feature data usually involves setting weight coefficients to better reflect the importance of each type of feature in the classification task. For example, if the growth status of weeds in a specific area is significantly affected by seasonal changes, then in the fusion process, the weight of the time series feature data can be appropriately increased to ensure that the dynamically changing information is fully utilized in the classification process. Through this weighted fusion, the system can comprehensively consider the current state and historical changes of the plant to generate more representative feature data.

[0085] For example, when monitoring a wheat field, the system first extracts spatial feature data in the area, which may show the distribution of wheat and weeds. Then, combined with time series data, the system can identify the growth trend of weeds in the area and form a deeper understanding of the weed growth cycle. Through weighted fusion, the fused feature data generated by these data not only contains the visual features of the weeds, but also incorporates relevant information about their growth dynamics, thereby significantly improving the accuracy of the final weed classification.

[0086] Fusion feature data plays a vital role in the UAV weeding monitoring system. By effectively integrating different features, the system can make more accurate weed identification and classification decisions in complex environments.

[0087] A spatial distribution map construction module is used to construct a spatial distribution map of weeds based on the final weed classification data, and to analyze the spatial relationship between weed nodes using a graph convolutional network, to predict the weed diffusion trend, to generate diffusion trend data, and to perform density analysis on the weed distribution in the region, to generate density data;

[0088] The spatial distribution map construction module processes the final weed classification data to establish a spatial distribution map of weeds. This process converts the classification data into a graphical representation in a spatial coordinate system, so that the location and category of each weed node can be intuitively displayed on the map. Such spatial graphics can not only clearly show the distribution of weeds, but also help farmers identify the weed growth in different areas and formulate corresponding management strategies.

[0089] The core of this module is to analyze the spatial relationship between weed nodes using graph convolutional networks. Graph convolutional networks are a powerful deep learning tool specifically designed to process graph-structured data. In this application, it can capture the interactions and proximity between weed nodes, thereby effectively identifying the diffusion pattern of weeds. For example, if weeds in a certain area grow rapidly and have a high density, the graph convolutional network can infer the temporal diffusion trend of weeds by analyzing the relationship between the area and the surrounding nodes, and generate diffusion trend data. Such predictive information is crucial for formulating weed control strategies, allowing farmers to take measures in advance to prevent the spread of weeds.

[0090] The module also performs density analysis on the distribution of weeds in the area and generates density data. By combining the number of weed nodes with their coverage area, the system can calculate the density value of the weeds, thereby providing farmers with quantitative information on the intensity of weed growth. For example, in a field, if the density analysis shows that the average density of weeds exceeds a predetermined threshold, farmers can take timely measures, such as mechanical weeding or spraying herbicides, to control the growth of weeds.

[0091] like Figure 3As shown, preferably, the spatial distribution map construction module is used to construct a spatial distribution map of weeds based on the node information in the final weed classification data, and the spatial distribution map is used to visually display the distribution of weeds in the target area.

[0092] In the present invention, the spatial distribution map construction module plays a vital role, and is mainly used to construct the spatial distribution map of weeds according to the node information in the final weed classification data. The core principle of this module is to realize the visualization of weeds in the target area by collecting and processing the weed node information, so as to provide farmers with accurate weed distribution status.

[0093] Specifically, the spatial distribution map construction module first extracts the location information and category of each weed node from the final weed classification data. This node information is usually obtained through the previous feature extraction and classification process. Each node represents the weed information in a certain area, including its type, quantity and growth status. By mapping this node information into a geographic coordinate system, the module can construct a distribution map of weeds in the area. This process usually involves the use of GIS (geographic information system) technology and data visualization tools to present the weed distribution in a graphical way.

[0094] The spatial distribution map not only clearly shows the distribution of weeds, but also indicates the density and type of weeds through different colors and symbols. For example, higher weed density may be marked in red, while lower density may be marked in green. This visual effect allows farmers to identify at a glance the areas that need to be focused on, thereby formulating weed control strategies more effectively.

[0095] For example, when monitoring a piece of farmland, the spatial distribution map can show the growth of weeds in different areas, helping farmers identify which areas are densely weeded and which areas are relatively clean. This information allows farmers to take targeted measures according to the actual situation, such as giving priority to mechanical weeding or spraying herbicides in areas with high weed density to achieve better management results.

[0096] Preferably, the graph convolutional network is used to analyze the weed nodes in the spatial distribution map, and the generated diffusion trend data includes the diffusion rate and diffusion direction of each weed node, which is used to guide subsequent weed control strategies.

[0097] In this invention, the graph convolutional network (GCN) plays an important role, which is specially used to analyze the weed nodes in the spatial distribution map and generate diffusion trend data. This process can effectively capture the spatial relationship between weed nodes by utilizing the powerful characteristics of the graph convolutional network, thereby evaluating the diffusion rate and direction of weeds and providing a scientific basis for subsequent weed control strategies.

[0098] Graph convolutional network is a deep learning model that can process graph structure data. Its principle is to learn the feature representation of nodes by aggregating the information of adjacent nodes. In the present invention, the weed nodes and their connection relationships in the spatial distribution map constitute a graph structure. GCN extracts the comprehensive features of each node in its neighborhood by performing convolution operations on the features of these nodes. These features include not only the basic information of weeds (such as type and density), but also the characteristics of neighboring weed nodes, so that the model can understand the growth environment of weeds more comprehensively.

[0099] In the implementation process, the graph convolutional network first encodes the features of each weed node and propagates information based on the adjacency relationship. For example, if the density of a node is high and the diffusion trend of the adjacent nodes is also strong, the GCN can learn that the node has a high diffusion potential. Through such feature aggregation and propagation, the system can generate diffusion trend data for each weed node, including diffusion rate and diffusion direction. The diffusion rate reflects the growth rate of the weeds, while the diffusion direction indicates the direction in which the weeds may expand outward. This information is crucial for formulating accurate weed control measures.

[0100] For example, when monitoring a corn field, if the graph convolutional network analysis results show that the weeds in a certain area have a high diffusion rate and their diffusion direction points to the adjacent crop rows, farmers can take timely measures based on this information, such as applying targeted herbicides in the area to prevent weeds from further threatening crops. Such applications not only improve the accuracy of weed control, but also reduce the use of chemical agents, thereby achieving more sustainable agricultural management.

[0101] The weeding strategy generation module generates preliminary weeding strategy data by combining the diffusion trend data, density data and real-time environmental data, optimizes the flight path of the drone, and performs weeding tasks. It also generates weeding effect data through real-time monitoring and optimizes the strategy based on the weeding effect data.

[0102] In the present invention, the weed control strategy generation module is a key decision support part, which is responsible for combining diffusion trend data, density data and real-time environmental data to generate preliminary weed control strategy data and optimize the flight path of the drone. The principle of this module is to formulate an effective weed control strategy through data fusion and intelligent algorithm, thereby improving the efficiency and accuracy of weed management.

[0103] Diffusion trend data provides information about the growth dynamics of weeds, including the rate of weed diffusion and possible diffusion directions. Density data shows the concentration of weeds in different areas, which enables the system to accurately identify hot spots of weed growth. At the same time, real-time environmental data (such as climate conditions, soil moisture, etc.) has a direct impact on the growth of weeds, and is therefore an important basis for formulating weed control strategies. By combining these data, the weed control strategy generation module can comprehensively analyze the weed conditions in the target area.

[0104] When generating preliminary weed control strategy data, the system typically uses machine learning algorithms and optimization techniques to take into account multiple factors, including the type, density, spread trend, and environmental conditions of weeds. For example, the system can apply decision trees or other intelligent algorithms to evaluate the effectiveness of different weed control methods in order to recommend the most appropriate treatment for each area. If the weed density in a certain area is high and the growth rate is fast, the system may recommend mechanical weeding, while in areas with fewer weeds, it may recommend chemical treatment.

[0105] Optimizing the flight path of the drone is another important function of this module. By analyzing the generated preliminary weeding strategy, the system is able to calculate the most effective flight route to maximize coverage of the target area while reducing overlapping operations and unnecessary flights. The drone's path optimization not only improves work efficiency, but also reduces energy consumption.

[0106] After executing the weeding task, the system will monitor the weeding effect data in real time, which is used to evaluate the effectiveness and implementation of the weeding strategy. For example, the system can determine the changes in the growth status of weeds by collecting spectral data and density data again. If it is found that weeds are still growing lushly in certain areas, the system will optimize the initial weeding strategy based on these real-time feedbacks, adjust the subsequent weeding methods and flight paths, and ensure the best weeding effect is achieved in the end.

[0107] Through this closed-loop control, the weeding strategy generation module can continuously learn and improve, making the drone more intelligent and efficient when performing weeding tasks.

[0108] Preferably, the weed control strategy generation module generates preliminary weed control strategy data by combining the diffusion trend data, density data and real-time environmental data using a decision tree algorithm, wherein the decision tree algorithm makes weed control decisions based on weed density and diffusion potential.

[0109] In the present invention, the weed control strategy generation module generates preliminary weed control strategy data by combining diffusion trend data, density data and real-time environmental data using a decision tree algorithm. The core of this process is to integrate information from multiple data sources in order to form scientific and reasonable weed control decisions and optimize farmland management.

[0110] The decision tree algorithm is a common machine learning method that can make classifications and decisions by analyzing input features such as weed density, diffusion potential, and environmental conditions. In the present invention, the algorithm first trains the input feature data to establish a relationship model between weed growth and environmental factors. Specifically, diffusion trend data provides information about how weeds spread in time and space, density data describes the concentration of weeds in a specific area, and real-time environmental data (such as temperature, humidity, precipitation, etc.) has a direct impact on weed growth. Through the decision tree algorithm, the system can identify which features are most important for weed growth and formulate corresponding weed control strategies based on these features.

[0111] For example, in a specific monitoring case, suppose that the weed density in a certain area is very high and the weather conditions are conducive to weed growth. The decision tree algorithm will take this information as input and generate a decision node indicating that more intensive mechanical weed control measures should be prioritized in this area, while in areas with relatively few weeds, the system may recommend the use of milder chemical herbicides. In this way, through the structure of the decision tree, farmers can clearly understand the specific actions to be taken in different situations.

[0112] The generated preliminary weed control strategy data can not only guide farmers to choose appropriate weed control methods, but also help optimize the operation plan of drones. After the strategy is executed, the data feedback on the weed control effect will be used to update and improve the decision tree model, further improving the accuracy of subsequent decisions. This intelligent decision-making process makes weed control management more scientific and accurate, which can effectively reduce labor costs and the use of chemical agents, thereby promoting the development of sustainable agriculture.

[0113] Preferably, the weeding strategy generation module further includes a real-time monitoring submodule for monitoring the execution of the weeding task and generating real-time weeding effect data, wherein the real-time weeding effect data is used to evaluate the weeding effect and optimize subsequent weeding strategies.

[0114] In the present invention, the weed control strategy generation module is not only responsible for formulating the preliminary weed control plan, but also includes a real-time monitoring submodule, which is used to monitor the execution of the weed control task and generate real-time weed control effect data. This mechanism is designed to ensure the effectiveness of weed control measures and provide data support for the optimization of subsequent strategies.

[0115] The core principle of the real-time monitoring submodule is to continuously track various indicators during the weeding process and collect relevant data through sensors carried by drones. For example, a multispectral camera may be used during the monitoring process to obtain real-time spectral reflectance data to analyze the health status of plants in the target area and the growth status of weeds. Real-time monitoring can also be combined with microwave sensors to obtain information on the status of soil and plant roots. These data can not only reflect the current weeding effect, but also evaluate whether the weeding measures have achieved the expected results, such as whether the weed density has been significantly reduced and whether crops have been affected.

[0116] The real-time weed control effect data generated provides farmers with a dynamic feedback mechanism, enabling them to quickly evaluate the effectiveness of current strategies. For example, during a certain application process, if the monitoring data shows that the weed growth is not as expected, the system can adjust the weed control strategy based on real-time data, such as increasing the dosage of the pesticide, changing the spraying area, or using other weed control methods. This flexibility enables farmers to respond to changes immediately and improve the success rate and efficiency of weed control operations.

[0117] Through continuous monitoring and feedback, the real-time monitoring submodule can also provide valuable data basis for the subsequent optimization of weed control strategies. By analyzing historical real-time monitoring data, the system can identify which weed control measures are more effective under specific environmental conditions, thereby guiding farmers to formulate more precise strategies in future weed control tasks.

[0118] For example, in a weeding task in a corn field, if real-time monitoring finds that the weed density in a certain area is still high, and environmental data shows that the climate conditions at this time are conducive to the volatilization and diffusion of the pesticide, the system may suggest that the farmer re-apply the pesticide in this area and continue to evaluate the effect of the improvement based on the real-time effect data. This dynamic adjustment mechanism enables the drone weeding monitoring system to not only perform preset tasks, but also be able to perform intelligent management according to actual conditions, significantly improving the scientificity and efficiency of agricultural management.

[0119] like Figure 4 As shown, a method for executing the drone-based weeding monitoring system comprises the following steps:

[0120] Collect spectral reflectance data and microwave internal structure data of the target area; the UAV weeding monitoring system is mainly based on multi-spectral and microwave sensors, and collects spectral reflectance data and microwave internal structure data of the target area through the UAV platform.

[0121] The spectral reflectance data and microwave internal structure data are received, and standardized perception data including spectral reflectance characteristics and internal structure information is generated; spectral reflectance data can reveal the surface characteristics of plants, while microwave internal structure data deeply analyzes the internal growth of plants, such as root structure and soil moisture. These two types of data are integrated through a data fusion module to generate standardized perception data including spectral reflectance characteristics and internal structure information. This step improves the overall analyzability and accuracy of the data.

[0122] The standardized perception data is feature extracted to generate spectral feature data and microwave feature data respectively, the data dimension is expanded through nonlinear transformation, and the feature data is classified by adaptive fuzzy clustering to generate preliminary feature classification data; the feature extraction module processes these standardized data to generate spectral feature data and microwave feature data respectively, and expands the data dimension through nonlinear transformation to ensure that the system can capture more detailed vegetation features. These features are classified through adaptive fuzzy clustering to generate preliminary feature classification data, which enables the system to distinguish between crops and weeds.

[0123] The preliminary feature classification data is received, spatial feature data is extracted through a convolutional neural network, time series feature analysis is performed in combination with a recursive neural network, fused feature data is generated, and the final weed classification data is output through feature reconstruction network optimization. After the system receives the preliminary feature classification data, the convolutional neural network in the multi-model collaborative learning system is used to extract the spatial features of the plant, while the recursive neural network processes the time series features. The advantage of the convolutional neural network is that it can accurately extract spatial features such as plant shape and color, while the recursive neural network is used to analyze plant growth trends. This information is fused to generate more accurate weed classification data, which is further optimized through the feature reconstruction network.

[0124] Based on the final weed classification data, a spatial distribution map of weeds is constructed, and the spatial relationship between weed nodes is analyzed using a graph convolutional network to predict the weed diffusion trend, generate diffusion trend data, and perform density analysis on the weed distribution in the region to generate density data; based on these weed classification data, the system constructs a spatial distribution map of weeds, which not only shows the current distribution of weeds, but also analyzes the spatial relationship between weed nodes through a graph convolutional network to predict the diffusion trend of weeds. These diffusion trend data are used to generate density data of weeds in the region, providing a basis for subsequent weed control strategies.

[0125] Combined with the diffusion trend data, density data and real-time environmental data, the system generates preliminary weeding strategy data, optimizes the flight path of the drone, and executes the weeding task. The system generates weeding effect data in real time and optimizes the strategy based on the weeding effect data. The system combines the diffusion trend, density data and real-time environmental data to generate an optimized weeding strategy, and performs path optimization and task execution through the drone. During the execution process, the system continuously adjusts and optimizes the strategy by real-time monitoring of the weeding effect to ensure that the final weeding task achieves the best effect.

[0126] The present invention can monitor the spread trend of weeds in real time, accurately analyze their growth dynamics, and provide efficient weed control strategies through multi-source data fusion and intelligent learning systems. This greatly reduces the use of chemical agents, improves operating efficiency, and reduces damage to crops, with significant agricultural management and ecological benefits.

[0127] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware.

[0128] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A weeding monitoring system based on drones, characterized in that: The following steps are involved: The drone platform is equipped with a multispectral camera and a microwave sensor to collect spectral reflectance data and microwave internal structure data of the target area; A data fusion module receives the spectral reflection data and the microwave internal structure data, and generates standardized perception data including spectral reflection characteristics and internal structure information; A feature extraction module is used to extract features from the standardized perception data, generate spectral feature data and microwave feature data respectively, expand the data dimension through nonlinear transformation, classify the feature data using adaptive fuzzy clustering, and generate preliminary feature classification data; A multi-model collaborative learning system receives the preliminary feature classification data, extracts spatial feature data through a convolutional neural network, performs time series feature analysis in combination with a recursive neural network, generates fused feature data, and outputs final weed classification data through feature reconstruction network optimization; A spatial distribution map construction module is used to construct a spatial distribution map of weeds based on the final weed classification data, and to analyze the spatial relationship between weed nodes using a graph convolutional network, to predict the weed diffusion trend, to generate diffusion trend data, and to perform density analysis on the weed distribution in the area, to generate density data; The weeding strategy generation module generates preliminary weeding strategy data by combining the diffusion trend data, density data and real-time environmental data, optimizes the flight path of the drone, and performs weeding tasks. It also generates weeding effect data through real-time monitoring and optimizes the strategy based on the weeding effect data.

2. The drone-based weeding monitoring system according to claim 1, characterized in that: The data fusion module comprises: A processing unit that receives spectral reflection data and microwave internal structure data generates standardized perception data using a weighted average method, wherein the calculation expression of the standardized perception data is: Among them, S is the standardized perception data; R is the spectral reflectance data; M is the microwave internal structure data; w1 is the weight coefficient of the spectral data; w2 is the weight coefficient of the microwave data.

3. The drone-based weeding monitoring system according to claim 1, characterized in that: The feature extraction module comprises: A nonlinear transformation unit, used for performing a nonlinear function transformation on the standardized perception data to expand the data dimension and generate expanded feature data; The adaptive fuzzy clustering unit is used to perform fuzzy clustering analysis on the expanded feature data to generate preliminary feature classification data. The clustering formula of the fuzzy clustering is: Among them, μ ij is the membership of the jth sample in the i-th cluster; d ij is the distance between the sample and the cluster center; C is the number of clusters; m is the fuzzy index, which is used to control the fuzziness of clustering.

4. The drone-based weeding monitoring system according to claim 1, characterized in that: The multi-model collaborative learning system comprises: A convolutional neural network module, used to process the preliminary feature classification data, and the extracted spatial feature data reflects the shape, color and other visual features of the plant; Recursive neural network module, used to analyze time series feature data and identify plant growth trends and cycles.

5. The drone-based weeding monitoring system according to claim 1, characterized in that: The fused feature data is generated by weighted fusion of the spatial feature data and the time series feature data, and the fused feature data is used to improve the classification accuracy of the final weed classification data.

6. The drone-based weeding monitoring system according to claim 1, characterized in that: The spatial distribution map construction module is used to construct a spatial distribution map of weeds based on the node information in the final weed classification data, and the spatial distribution map is used to visually display the distribution of weeds in the target area.

7. The drone-based weeding monitoring system according to claim 1, characterized in that: The graph convolutional network is used to analyze the weed nodes in the spatial distribution map, and the generated diffusion trend data includes the diffusion rate and diffusion direction of each weed node, which is used to guide subsequent weed control strategies.

8. The drone-based weeding monitoring system according to claim 1, characterized in that: The weed control strategy generation module generates preliminary weed control strategy data by combining the diffusion trend data, density data and real-time environmental data using a decision tree algorithm, wherein the decision tree algorithm makes weed control decisions based on weed density and diffusion potential.

9. The drone-based weeding monitoring system according to claim 1, characterized in that: The weeding strategy generation module also includes a real-time monitoring submodule for monitoring the execution of the weeding task and generating real-time weeding effect data, wherein the real-time weeding effect data is used to evaluate the weeding effect and optimize the subsequent weeding strategy.

10. A method for executing the drone-based weeding monitoring system according to any one of claims 1 to 9, characterized in that: The following steps are involved: Collect spectral reflection data and microwave internal structure data of the target area; receiving the spectral reflection data and the microwave internal structure data, and generating standardized perception data including spectral reflection characteristics and internal structure information; Extracting features from the standardized perception data to generate spectral feature data and microwave feature data, respectively, expanding the data dimension through nonlinear transformation, and classifying the feature data using adaptive fuzzy clustering to generate preliminary feature classification data; Receiving the preliminary feature classification data, extracting spatial feature data through a convolutional neural network, performing time series feature analysis in combination with a recursive neural network, generating fused feature data, and outputting final weed classification data through feature reconstruction network optimization; Based on the final weed classification data, a spatial distribution map of weeds is constructed, and a graph convolutional network is used to analyze the spatial relationship between weed nodes, predict the weed diffusion trend, generate diffusion trend data, and perform density analysis on the weed distribution in the area to generate density data; Combined with the diffusion trend data, density data and real-time environmental data, preliminary weeding strategy data is generated, the flight path of the drone is optimized, and the weeding task is performed. Real-time monitoring is performed to generate weeding effect data, and strategy optimization is performed based on the weeding effect data.

Images