Regional soil monitoring method

Abstract

The invention discloses a regional soil monitoring method, which comprises the following steps: dividing a planting region and a growth cycle, acquiring remote sensing image data of the planting region, and acquiring real-time soil moisture content data; correcting the soil moisture content data by adopting multiple vegetation indexes, identifying soil texture by adopting a convolutional neural network random detection model, and obtaining soil data features; dividing soil moisture content values of different depths according to the soil data characteristics, and obtaining parameter factors influencing the soil moisture content; according to the parameter factors, a soil texture identification model is adopted to identify soil sample features, a soil monitoring and positioning model is established, real-time soil monitoring is performed on different scenes, a more comprehensive data view angle is provided, real-time monitoring of soil conditions is realized, soil texture feature data can be acquired and processed in real time, and the real-time monitoring of the soil conditions is realized. And dynamic monitoring requirements under different scenes are supported, challenges under complex environmental conditions are dealt with, and continuity and reliability of soil monitoring tasks are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of soil monitoring, and particularly relates to a method for regional soil monitoring. Background Art

[0002] Currently, with the increasing emphasis on ecological environment protection, soil monitoring, as an important part of environmental protection, is gradually attracting more attention, which makes accurate and effective regional soil monitoring methods particularly important. However, existing soil monitoring technologies face multiple challenges and deficiencies.

[0003] Traditional soil monitoring methods mostly rely on laboratory analysis. Although this method is accurate, it is time-consuming and laborious and cannot meet the need for rapid response. Despite the development of modern technologies such as remote sensing RS, geographic information system GIS, and global positioning system GPS providing new possibilities for soil monitoring, in practical applications, the integration and popularity of these technologies are still limited. In addition, monitoring standards and technical specifications for new pollutants have not been fully established, restricting the understanding and treatment of the overall picture of soil pollution. Therefore, it is urgent to develop an efficient, accurate, and easy-to-popularize regional soil monitoring method. Currently, there are still the following problems with regional soil monitoring methods:

[0004] (1) Existing soil monitoring technologies lack unified standards in the process of soil sample processing, resulting in poor consistency and comparability of results, affecting the accuracy of monitoring data. At the same time, although automated monitoring systems are being gradually established, most regions still rely on manual sampling and laboratory analysis, which is not only time-consuming and laborious but also prone to human errors;

[0005] (2) The limitation of real-time monitoring ability is also a prominent problem, and there are not many systems that can achieve real-time monitoring in the true sense and provide timely feedback of data. Summary of the Invention

[0006] The purpose of the present invention is to provide a method for regional soil monitoring to solve the technical problems of low accuracy of soil monitoring data and limitations in automated real-time monitoring in the prior art.

[0007] To solve the above technical problems, the present invention specifically provides the following technical solutions:

[0008] The present invention provides a method for regional soil monitoring, including the following steps:

[0009] By dividing the planting area and growth period, obtaining remote sensing image data of the planting area, and combining with a wireless sensor network node positioning model to monitor the soil environment in real time, real-time soil moisture data is obtained;

[0010] The soil moisture data is corrected using multiple vegetation indices, and a convolutional neural network random detection model is used to identify the soil texture to obtain soil data features;

[0011] The soil moisture values at different depths are divided through the soil data features, and a correlation analysis is carried out with the multiple vegetation indices to obtain the parameter factors affecting the soil moisture;

[0012] Based on the parameter factors, a soil texture recognition model is used to identify the soil sample features, a soil texture recognition platform is established to monitor the soil type in real time, and a soil monitoring positioning model is established to conduct real-time soil monitoring for different scenarios.

[0013] As a preferred solution of the present invention, by dividing the planting area and the growth period, remote sensing image data of the planting area is obtained, including:

[0014] According to the geographical location, topography, soil conditions and climate conditions, the area is divided into different planting areas, and the vegetation types planted in different areas are clarified using historical data;

[0015] The boundary information of specific plots is extracted through high-resolution remote sensing images, and according to the growth characteristics of different crops, the growth period is divided into several stages;

[0016] Combined with the local climate data and the crop growth cycle, the time range of each growth period is determined, and the vegetation indicators are monitored within each growth period;

[0017] The monitored vegetation indicators are set within the time window of the crop growth period, and the growth status of the crops is evaluated through the vegetation indicators.

[0018] As a preferred solution of the present invention, a positioning model is established using the vegetation indicators in combination with wireless sensor network nodes to obtain positioning data, including:

[0019] A number of sensors are randomly set in the vegetation areas divided within the region, the sensor nodes are set as beacon nodes, and the node data is transmitted to the network in a flooding manner;

[0020] The shortest path between the beacon nodes is extracted according to the node data, and the minimum number of hops and the node numbers passed on the shortest path are recorded;

[0021] Based on the shortest path, the average hop distance between the corresponding beacon nodes at both ends of the path is calculated, and the shortest paths between the beacon nodes are sorted in ascending order according to the average hop distance;

[0022] Traverse the unknown nodes on the path in sequence from the sorted shortest paths, obtain the node information corresponding to the unknown nodes, add the unknown nodes to the beacon nodes, plan all the beacon nodes into the positioning sequence, and obtain the fixed number data within the area.

[0023] As a preferred solution of the present invention, the sine-cosine algorithm is used to optimize the positioning data to obtain position information, and the soil environment is monitored in real time through the position information to obtain real-time soil moisture data, including:

[0024] Randomly initialize the positioning data to obtain N position individuals, perform a global optimization search on the N position individuals, and optimize the individual positions using the variation of the sine function. Its expression is:

[0025]

[0026] Among them, represents the value of the i-th position component of the individual in the (t + 1)-th iteration optimization process. r1 and r2 are random numbers, and the range of r2 is [0, 2π]. represents the optimal position in the t-th iteration process. represents the value of the i-th position component of the individual in the t-th iteration optimization process.

[0027] Calculate the fitness value of the individual position using the fitness function and sort it to obtain the optimal position of the current individual and the optimal position of the population.

[0028] Continuously update the individual positions using the periodicity of the sine function, and perform boundary control on the individual positions. Re-initialize the individuals that exceed the boundaries.

[0029] By judging whether it falls into a local optimum during the global optimization search process, calculate the fitness value of each individual position, and update the optimal position of the current iteration individual and the historical optimal position of the population.

[0030] According to the optimal position of the individual and the historical optimal position of the population, monitor the environmental data of the soil in the planning area in real time, and calculate the real-time soil moisture data according to the environmental data of the soil.

[0031] As a preferred solution of the present invention, the multi-vegetation index is used to correct the soil moisture data, and a convolutional neural network random detection model is used to identify the soil texture to obtain soil data characteristics, including:

[0032] Divide the multi-vegetation indices into the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EVI), and Soil Adjusted Vegetation Index (SAVI), establish data characteristic sequences of the multi-vegetation indices with the NDVI, EVI, and SAVI, analyze the correlation of the soil moisture data after being corrected by the multi-vegetation indices, and obtain the corrected soil moisture data.

[0033] Use the corrected soil moisture data as training data, and train the soil moisture data using a convolutional neural network to extract soil moisture characteristic data;

[0034] Classify the soil texture of the soil moisture characteristic data to obtain real-time soil data characteristics.

[0035] As a preferred embodiment of the present invention, divide the soil moisture values at different depths through the soil data characteristics, including:

[0036] Construct soil moisture values at different depths for the soil data characteristics using the NDVI, and use the corrected NDVI data. Take the crop growth period time of the soil area to be measured as the horizontal axis and the pixel-corrected NDVI data as the vertical axis to establish a correlation curve;

[0037] Conduct a regression analysis on the soil moisture at different depths and each vegetation index to obtain the change trends of the soil moisture at different depths and each vegetation index.

[0038] As a preferred embodiment of the present invention, conduct a correlation analysis on the multi-vegetation indices according to the change trend of the soil moisture to obtain parameter factors affecting the soil moisture, including:

[0039] Establish a multiple linear regression equation for the soil moisture at different depths and each vegetation index according to the change trend of the soil moisture;

[0040] Adjust the corresponding depths of the soil moisture to 0 - 10 cm and 10 - 20 cm, and obtain the soil moisture values respectively.

[0041] Analyze the influencing factors under different soil moisture levels according to the regression trend of the soil moisture and each vegetation index to obtain parameter factors affecting the soil moisture.

[0042] As a preferred embodiment of the present invention, use a soil texture recognition model to recognize soil sample characteristics, including:

[0043] The soil texture recognition model uses a convolutional neural network with two convolutional layers and three pooling layers to extract the particle features, color features, and texture features of the soil from the extracted soil images. It transfers the two-dimensional output of the maximum pooling layer to a one-dimensional output through a flattening layer, connects the convolutional layer and the max pooling layer, and passes through the fully connected layer;

[0044] In the fully connected layer, a kernel function is used to classify the extracted features, and image features are extracted by combining the parameter factors with the soil moisture value;

[0045] The RF model is used to predict the soil sample features based on the random combination of the image features, and the soil texture feature data is obtained.

[0046] As a preferred embodiment of the present invention, a soil texture recognition platform is established based on the soil texture feature data to monitor the soil type in real time, including:

[0047] Vue is used to build the user interface of the soil texture recognition platform, and the java language is combined to process the soil texture feature data to obtain the geographical information, data features, and multi-vegetation indices of the corresponding soil area;

[0048] Using the Springboot technology, the soil information is managed and recognized, the type of the soil is recorded, and the differences in the soil of each region and the soil texture are obtained.

[0049] As a preferred embodiment of the present invention, a soil monitoring positioning model is established through the soil texture recognition platform to perform real-time soil monitoring for different scenarios, including:

[0050] On the soil texture recognition platform, the node information based on the sensor is obtained in real time. The beacon node coordinates are selected from the node information, and a non-linear equation set is established for the beacon node coordinates and the distance information. Its expression is:

[0051]

[0052] Among them, (x,y) represents the unknown node coordinates, (x1, y1), (x2, y2), and (x3, y3) respectively represent the coordinates of the three beacon nodes, and d1, d2, and d3 respectively represent the shortest distances from the unknown node to the three beacon nodes;

[0053] By performing iterative optimization and searching on the d1, d2, and d3 using the sine-cosine algorithm, the unknown node coordinates are solved to obtain the positioning coordinates of the soil area, and real-time soil monitoring is performed in different scenarios.

[0054] The present invention has the following beneficial effects compared with the prior art:

[0055] The present invention utilizes multiple vegetation indices to correct soil moisture data, improving the accuracy of soil moisture assessment. By performing multiple linear regression analysis on the relationship between soil moisture at different depths and vegetation indices, the key parameter factors affecting soil moisture are accurately identified. Combining the parameter factors with the soil moisture value to extract image features, this method not only considers the physical appearance characteristics of the soil but also incorporates the influence of environmental factors on soil texture, providing a more comprehensive data perspective and contributing to improving the authenticity and reliability of prediction results.

[0056] Real-time sensor node information is obtained through the soil texture identification platform, and the coordinates of the soil area are quickly calculated in combination with the positioning model, realizing real-time monitoring of the soil condition. It can obtain and process soil texture feature data in real time, support dynamic monitoring requirements in different scenarios, cope with challenges under complex environmental conditions, and ensure the continuity and reliability of soil monitoring tasks. Brief Description of the Drawings

[0057] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings described below are merely exemplary, and for those of ordinary skill in the art, without creative efforts, other implementation drawings can also be obtained based on the provided drawings.

[0058] Figure 1 It is a flowchart of the regional soil monitoring method provided by the embodiment of the present invention. Detailed Embodiments

[0059] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0060] As Figure 1 shown, the present invention provides a regional soil monitoring method, including the following steps:

[0061] By dividing the planting area and growth period, remote sensing image data of the planting area is obtained, and the soil environment is monitored in real time in combination with the wireless sensor network node positioning model to obtain real-time soil moisture data;

[0062] In this embodiment, considering the differences in regional light and heat conditions and crop phenological periods, the planting areas and growth cycles are divided respectively. Within their respective areas and growth periods, factors such as temperature and spectrum are considered to correct the normalized difference vegetation index (NDVI), and the relative soil humidity is monitored. Real-time soil environment information is obtained through a sensor network.

[0063] The soil moisture data is corrected using multiple vegetation indices, and a convolutional neural network random detection model is used to identify the soil texture to obtain soil data characteristics.

[0064] In this embodiment, considering the response relationships between multiple indices and soil moisture at different depths, according to the actual situation of the data source, the normalized difference vegetation index (NDVI), vegetation supply water index (VSWI), and enhanced vegetation index (EVI) are selected to conduct correlation analysis with soil moisture at different depths, regression models are established respectively and the prediction accuracy is analyzed, and different adaptive methods are used for soil moisture monitoring according to the characteristics of different crop periods.

[0065] The soil moisture values at different depths are divided through the soil data characteristics, and correlation analysis is conducted with the multiple vegetation indices to obtain the parameter factors affecting the soil moisture.

[0066] In this embodiment, since the change of soil humidity at different depths may affect the water absorption efficiency of plant roots, by analyzing the soil moisture values at different depths, more accurate soil moisture information can be provided, which helps to achieve refined agricultural irrigation and water resource management.

[0067] In this embodiment, by conducting correlation analysis between the soil moisture values and multiple vegetation indices, the complex relationship between the soil moisture status and vegetation growth can be revealed. This not only helps to understand how vegetation responds to soil moisture changes, but also can identify the key environmental factors affecting the soil moisture, providing a scientific basis for optimizing agricultural production.

[0068] In this embodiment, based on the correlation analysis results between the soil moisture values and multiple vegetation indices, a mathematical model can be established between the soil moisture and other environmental factors such as temperature, precipitation, and vegetation coverage, thereby improving the accuracy and reliability of soil moisture prediction, which is crucial for disaster warning, resource management, and the formulation of ecological restoration plans.

[0069] According to the parameter factors, a soil texture recognition model is used to identify the soil sample characteristics, a soil texture recognition platform is established to monitor the soil type in real time, and a soil monitoring positioning model is established to conduct real-time soil monitoring for different scenarios.

[0070] In this embodiment, the soil texture recognition model can accurately identify soil types such as sandy soil, loam, or clay based on soil sample characteristics, which helps to divide and manage land resources more precisely. Different types of soil have different physical and chemical properties, and have different requirements and impacts on aspects such as agricultural production, soil and water conservation, and ecological environment protection.

[0071] In this embodiment, establishing a soil texture recognition platform can achieve real-time monitoring of soil types, can timely capture changes in soil conditions and make responses, which is particularly useful for situations where management strategies need to be quickly adjusted, such as quickly assessing soil damage after natural disasters, or adjusting irrigation and fertilization plans according to seasonal changes, etc.

[0072] In this embodiment, using the soil monitoring positioning model can conduct targeted soil monitoring in different scenarios, providing scientific data support for formulating more reasonable agricultural practices and environmental protection measures. For example, optimizing water resource allocation in arid regions, implementing restoration projects in polluted areas, etc., so as to improve resource utilization efficiency and environmental quality. By accurately understanding soil characteristics and their spatio-temporal variation laws, long-term land management and ecological protection projects can be better planned and implemented, which not only helps to increase crop yield and quality, but also effectively prevents soil degradation, maintains biodiversity, and promotes the sustainable development of agriculture and social economy.

[0073] By dividing the planting area and growth cycle, remote sensing image data of the planting area is obtained, including:

[0074] According to geographical location, topography, soil conditions, and climate conditions, the region is divided into different planting areas, and the vegetation types planted in different areas are determined using historical data;

[0075] In this embodiment, by subdividing the planting area according to geographical location, topography, soil conditions, and climate conditions, precise agricultural management for different regional characteristics can be achieved, which can better match crops with the growth environment and improve resource utilization efficiency.

[0076] Extract the boundary information of specific plots through high-resolution remote sensing images, and divide the growth cycle into several stages according to the growth characteristics of different crops;

[0077] In this embodiment, using high-resolution remote sensing images to extract the boundary information of specific plots and combining with the division of stages of the crop growth cycle for monitoring can provide real-time and dynamic data on crop growth conditions. This timely information acquisition helps to quickly identify and respond to problems that occur during the crop growth process, so as to take corresponding management measures.

[0078] Combined with local climate data and crop growth cycles, determine the time range of each growth cycle, and monitor vegetation indicators within each of the growth cycles;

[0079] Set the monitored vegetation indicators within the time window of the crop growth cycle, and evaluate the crop growth status through the vegetation indicators.

[0080] In this embodiment, by integrating local climate data with crop growth cycles to determine the time range of each growth cycle and monitoring specific vegetation indicators during these critical time periods, it is possible to provide scientific decision-making support for farmers and managers to maximize economic benefits.

[0081] In this embodiment, by evaluating the crop growth status based on the vegetation index, it is possible to continuously monitor the crop health status and its change trends throughout the entire growth cycle of the crop. This not only improves the accuracy of crop growth status evaluation but also provides a reliable data basis for predicting yields.

[0082] Adopt the vegetation indicators and combine with wireless sensor network nodes to establish a positioning model to obtain positioning data, including:

[0083] Randomly set several sensors in the vegetation areas divided within the region, set the sensor nodes as beacon nodes, and transmit node data to the network in a flooding manner;

[0084] In this embodiment, by randomly setting several sensors in the vegetation areas divided within the region and using these sensor nodes as beacon nodes, a dense monitoring network can be constructed. The flooding algorithm is used to transmit node data in the network, ensuring comprehensive coverage and redundancy of information, which helps to improve the accuracy of positioning.

[0085] Extract the shortest path between the beacon nodes according to the node data, and record the minimum number of hops and the node numbers passed on the shortest path;

[0086] In this embodiment, based on the shortest path between the extracted beacon nodes and the recorded minimum number of hops and the node numbers passed, it is possible to effectively identify the critical paths in the network, calculate the average hop distance corresponding to both ends of the path, and sort the shortest paths. This not only optimizes the path selection but also reduces data transmission delay and improves the working efficiency of the entire network.

[0087] Calculate the average hop distance between the beacon nodes corresponding to both ends of the path according to the shortest path, and sort the shortest paths between the beacon nodes in ascending order of the average hop distance;

[0088] Traverse the unknown nodes on the path in sequence from the sorted shortest path, obtain the node information corresponding to the unknown node, add the unknown node to the beacon node, plan all the beacon nodes into the positioning sequence, and obtain the constant data in the area.

[0089] In this embodiment, unknown nodes are traversed sequentially from the sorted shortest paths and added to the beacon node set. This mechanism allows the system to gradually expand the number of known nodes to form a more complete positioning sequence. This method is particularly suitable for large areas or inaccessible areas, facilitating subsequent data collection and analysis.

[0090] In this embodiment, combining vegetation indicators with wireless sensor networks can not only monitor crop growth conditions in real time, but also accurately locate the specific location of crop growth and its environmental parameters. This is crucial for achieving precision agricultural management and can support farmers in making more scientific and reasonable decisions, thereby improving agricultural production efficiency and resource utilization.

[0091] The positioning data is optimized using a sine-cosine algorithm to obtain position information, and the soil environment is monitored in real time through the position information to obtain real-time soil moisture data, including:

[0092] The positioning data is randomly initialized to obtain N position individuals, and a global optimization search is performed on the N position individuals. The individual positions are optimized using the variation of the sine function, and the expression is:

[0093]

[0094] in, It represents the value of the i-th position component of the individual in the t+1 iteration optimization process, r1 and r2 are random numbers, and the range of r2 is [0,2π]. represents the optimal position during the tth iteration, Represents the value of the i-th position component of the individual in the t-th iterative optimization process;

[0095] In this embodiment, the sine function regards possible positioning data as an "individual", and all individuals together constitute a "population". The position of each individual is represented as a multidimensional vector, whose value represents the coordinates of the solution in the search space. The goal of the algorithm is to continuously update the position of the individual through an iterative process, thereby gradually approaching the optimal solution.

[0096] In this embodiment, the sine-cosine algorithm is used to perform a global optimization search on the positioning data, which can effectively avoid the local optimal problem that may occur in traditional positioning methods. The sine-cosine algorithm dynamically adjusts the position of the individual by simulating the changes of the sine and cosine functions, thereby finding a better solution in a complex environment and improving the accuracy of positioning.

[0097] Calculate the fitness value of the individual position using the fitness function and sort it to obtain the optimal position of the current individual and the optimal position of the population.

[0098] Continuously update the individual position using the periodicity of the sine function, perform boundary control on the individual position, and re-initialize the individuals that exceed the boundary.

[0099] By judging whether the global optimization search process falls into a local optimum, calculate the fitness value of each individual position, and update the optimal position of the current iterative individual and the historical optimal position of the population.

[0100] According to the optimal position of the individual and the historical optimal position of the population, monitor the environmental data of the soil in the planning area in real time, and calculate the real-time soil moisture data according to the environmental data of the soil.

[0101] In this embodiment, the sine-cosine algorithm can quickly converge to the global optimal solution or a position close to the global optimal solution by continuously updating the individual position and using the fitness function to evaluate its performance. At the same time, by re-initializing the individuals that exceed the boundary and implementing the boundary control strategy, the robustness and stability of the algorithm are enhanced, ensuring reliable positioning results even under adverse conditions.

[0102] In this embodiment, based on the optimized position information and the real-time obtained soil environmental data, the manager can obtain a more accurate and comprehensive information basis for making scientific and reasonable agricultural production and environmental protection decisions, which not only helps to improve agricultural production efficiency, but also protects the ecological environment and promotes the coordinated development of the economy and the environment.

[0103] In this embodiment, the monitoring strategy can be dynamically adjusted according to the real-time obtained soil environmental data to timely reflect the changes in soil moisture, which helps to improve the yield and quality of crops.

[0104] Correct the soil moisture data using multiple vegetation indices, use a convolutional neural network random detection model to identify the soil texture, and obtain soil data features, including:

[0105] Divide the multiple vegetation indices into the normalized difference vegetation index (NDVI), the enhanced vegetation index (EVI), and the soil-adjusted vegetation index (SAVI), establish the data feature sequences of the multiple vegetation indices with the normalized difference vegetation index (NDVI), the enhanced vegetation index (EVI), and the soil-adjusted vegetation index (SAVI), analyze the correlation of the soil moisture data after being corrected by the multiple vegetation indices, and obtain the corrected soil moisture data.

[0106] In this embodiment, by combining multiple vegetation indices with the corrected soil moisture data, the relationship between vegetation coverage and soil moisture can be more accurately reflected. Different vegetation indices have different sensitivities to vegetation cover and soil background. Using these indices comprehensively can effectively reduce the errors that may be brought by a single index, thereby providing a more accurate soil moisture assessment.

[0107] Taking the corrected soil moisture data as training data, a convolutional neural network is used to train the soil moisture data to extract soil moisture characteristic data.

[0108] In this embodiment, using the corrected soil moisture data as the training data set can provide a more reliable learning sample for the convolutional neural network, which helps to improve the generalization ability and robustness of the model, enabling it to maintain a high prediction accuracy when facing soil data of different types or conditions.

[0109] Classify the soil texture of the soil moisture characteristic data to obtain real-time soil data characteristics.

[0110] In this embodiment, based on the corrected soil moisture data and the trained CNN model, real-time monitoring of soil texture can be achieved, which means that once any abnormal changes are detected, measures can be taken quickly for adjustment, thereby ensuring the healthy growth of crops. By obtaining detailed soil data characteristics, including but not limited to soil texture information, agricultural managers can also obtain scientific basis to make more reasonable and effective management decisions. For example, according to the specific conditions of different plots, select appropriate crop varieties, fertilization plans, etc., which helps to improve agricultural production efficiency and sustainability.

[0111] Divide the soil moisture values at different depths through the soil data characteristics, including:

[0112] Using the normalized difference vegetation index NDVI for the soil data characteristics to construct soil moisture values at different depths. Using the corrected normalized difference vegetation index NDVI data, with the crop growth period time of the soil area to be measured as the horizontal axis and the pixel-corrected NDVI data as the vertical axis, establish a correlation curve.

[0113] Conduct regression analysis on the soil moisture at different depths and each vegetation index to obtain the change trends of the soil moisture at different depths and each vegetation index.

[0114] In this embodiment, using the corrected NDVI data to establish the correlation curve of soil moisture at different depths can more accurately reflect the change of soil moisture with depth, taking into account the change of crop growth period time, making the soil moisture assessment closer to the actual situation and improving the accuracy of the monitoring results.

[0115] In this embodiment, by performing regression analysis on soil moisture conditions at different depths and various vegetation indices, we can deeply understand how soil moisture affects vegetation growth and how vegetation cover in turn affects soil water dynamics. Based on the change trends between soil moisture conditions at different depths and vegetation indices, agricultural producers can formulate a more scientific and reasonable irrigation plan according to the actual water requirements of crops, ensuring the healthy growth of crops while avoiding water resource waste.

[0116] Perform correlation analysis on the multiple vegetation indices according to the change trend of the soil moisture condition to obtain parameter factors affecting the soil moisture condition, including:

[0117] Establish a multiple linear regression equation for soil moisture conditions at different depths and various vegetation indices according to the change trend of the soil moisture condition;

[0118] Adjust the corresponding depths of the soil moisture condition to 0 - 10 cm and 10 - 20 cm, and obtain soil moisture values respectively.

[0119] In this embodiment, by establishing multiple linear regression equations for different depths, we can accurately identify which vegetation indices and other environmental factors have a significant impact on the soil moisture condition at a specific depth. Based on the regression analysis results between the soil moisture condition and vegetation indices, we can more accurately predict the change trend of the soil moisture condition at different depths.

[0120] Analyze the influencing factors under different soil moisture levels according to the regression trend between the soil moisture condition and various vegetation indices to obtain parameter factors affecting the soil moisture condition.

[0121] In this embodiment, after understanding the soil moisture conditions at different depths and their influencing factors, we can use water resources and fertilizers more targeted, avoiding waste and environmental pollution caused by over-irrigation or over-fertilization. In addition, it can also guide the selection of crop varieties suitable for local soil conditions, further improving the resource utilization efficiency.

[0122] In this embodiment, through multiple linear regression analysis, we can flexibly adjust the model parameters according to the actual situation to better adapt to the characteristics of different regions and different types of soil.

[0123] Use a soil texture recognition model to identify the characteristics of soil samples, including:

[0124] The soil texture recognition model uses a convolutional neural network with two convolutional layers and three pooling layers to extract the particle characteristics, color characteristics, and texture characteristics of the soil from the extracted soil images. It transfers the two-dimensional output of the maximum pooling layer to a one-dimensional output through a flattening layer, connects the convolutional layer and the max pooling layer, and passes through the fully connected layer.

[0125] In this embodiment, two convolutional layers and three pooling layers in the convolutional neural network can effectively extract particle features, color features, and texture features from soil images. This multi-level feature extraction method can capture complex structural information in soil samples, provide rich feature descriptions, and help identify soil texture more accurately.

[0126] In the fully connected layer, a kernel function is used to classify the extracted features, and image features are extracted according to the parameter factor combined with the soil moisture content value.

[0127] In this embodiment, the two-dimensional output of the max pooling layer is converted into a one-dimensional output through a flattening layer and connected to the fully connected layer, enabling the model to comprehensively analyze and classify the extracted features in the final stage. This method combines the advantages of CNN in image feature extraction and the powerful ability of the kernel function in classification, improving the accuracy of soil texture classification.

[0128] The RF model is used to predict the soil sample features based on the random combination of image features, and the soil texture feature data is obtained.

[0129] In this embodiment, image features are extracted according to the parameter factors affecting soil moisture content obtained previously in combination with the soil moisture content value. This method not only considers the physical appearance features of the soil but also incorporates the influence of environmental factors on soil texture, providing a more comprehensive data perspective and helping to improve the authenticity and reliability of the prediction results.

[0130] In this embodiment, the random forest RF model is used to predict the soil sample features based on the random combination of image features, increasing the diversity of the model, reducing the risk of overfitting, and being able to adapt to different types of soil samples. Whether it is sandy soil, loam, or clay, the performance can be optimized by adjusting the model parameters or the training set.

[0131] A soil texture recognition platform is established based on the soil texture feature data to monitor the soil type in real time, including:

[0132] Vue is used to build the user interface of the soil texture recognition platform, and the java language is combined to process the soil texture feature data to obtain the geographical information, data features, and multi-vegetation indices of the corresponding area of the soil.

[0133] Using Springboot technology, the soil information is managed and recognized, the type of the soil is recorded, and the differences between the soils in each region and the soil texture are obtained.

[0134] In this embodiment, the Vue framework is used to build the user interface, which can provide a modern, responsive, and intuitive user experience. The component-based development mode of Vue makes the interface design more flexible and modular, facilitating later maintenance and expansion. At the same time, its lightweight feature ensures the smooth operation of the platform on different devices, improving the user's operation efficiency.

[0135] In this embodiment, the Springboot technology is used to manage and identify soil information, which can quickly build a stable and scalable backend service system, making the work of soil type recording and regional differentiation more efficient.

[0136] In this embodiment, through the comprehensive analysis of soil type, texture characteristics, and multi-vegetation indices, the soil differences in different regions can be accurately identified, revealing the relationship between soil properties and the geographical environment. This not only helps to formulate targeted land management strategies but also provides a scientific basis for agricultural planning, ecological protection, and resource allocation.

[0137] A soil monitoring positioning model is established through the soil texture identification platform to conduct real-time soil monitoring of different scenarios, including:

[0138] On the soil texture identification platform, the node information based on sensors is obtained in real time. The beacon node coordinates are selected from the node information, and a non-linear equation set is established for the beacon node coordinates and distance information. Its expression is:

[0139]

[0140] Where (x, y) represents the unknown node coordinates, (x1, y1), (x2, y2), and (x3, y3) respectively represent the coordinates of three beacon nodes, and d1, d2, and d3 respectively represent the shortest distances from the unknown node to the three beacon nodes;

[0141] By using the sine-cosine algorithm to iteratively optimize and find the optimal solution for d1, d2, and d3, the unknown node coordinates are solved to obtain the positioning coordinates of the soil area, and real-time soil monitoring is carried out in different scenarios.

[0142] In this embodiment, by selecting the beacon node coordinates and using the distance information of the sensor nodes to construct a non-linear equation set to solve the unknown node coordinates, the specific location of the soil area can be accurately located. This method combines the distributed characteristics of the sensor network and the advantages of mathematical modeling, and can provide high positioning accuracy in complex terrains and changing environments.

[0143] In this embodiment, the sine-cosine algorithm is used to iteratively optimize and search for the coordinates of unknown nodes, which can effectively avoid the local optimum problem that may occur in traditional methods. The sine-cosine algorithm dynamically adjusts the individual positions by simulating the changes of sine and cosine functions, ensuring to find the global optimum or a position close to the optimum in the solution space, thereby improving the efficiency and accuracy of positioning calculation.

[0144] In this embodiment, the method is applicable to different soil monitoring scenarios. Whether it is a flat farmland, undulating hills or complex mountainous environments, it can be flexibly deployed and applied. By dynamically adjusting the beacon nodes and distance information, it can adapt to various terrain conditions and meet diverse monitoring requirements.

[0145] The present invention uses a variety of vegetation indices to correct soil moisture data, improving the accuracy of soil moisture assessment. By performing multiple linear regression analysis on the relationship between soil moisture at different depths and vegetation indices, the key parameter factors affecting soil moisture are accurately identified, and the parameter factors are combined with the soil moisture value to extract image features. This method not only considers the physical appearance characteristics of the soil, but also incorporates the influence of environmental factors on soil texture, providing a more comprehensive data perspective and helping to improve the authenticity and reliability of prediction results.

[0146] The present invention obtains sensor node information in real time through the soil texture identification platform and quickly calculates the coordinates of the soil area in combination with the positioning model, realizing real-time monitoring of soil conditions. It can obtain and process soil texture feature data in real time, support dynamic monitoring requirements in different scenarios, cope with challenges under complex environmental conditions, and ensure the continuity and reliability of soil monitoring tasks.

[0147] The above embodiments are only exemplary embodiments of the present application and are not used to limit the present application. The protection scope of the present application is defined by the claims. Those skilled in the art can make various modifications or equivalent replacements within the essence and protection scope of the present application, and such modifications or equivalent replacements should also be regarded as falling within the protection scope of the present application.

Claims

1. A regional soil monitoring method, characterized in that, Including the following steps: By dividing the planting area and growth cycle, obtaining remote sensing image data of the planting area, combining with the wireless sensor network node positioning model to monitor the soil environment in real time, and obtaining real-time soil moisture data; Using multiple vegetation indices to correct the soil moisture data, and using a convolutional neural network random detection model to identify the soil texture to obtain soil data characteristics; Dividing the soil moisture values at different depths through the soil data characteristics, performing a correlation analysis with the multiple vegetation indices, and obtaining parameter factors affecting the soil moisture; Based on the parameter factors, using a soil texture identification model to identify the characteristics of soil samples, establishing a soil texture identification platform to monitor the soil type in real time, establishing a soil monitoring positioning model, and performing real-time soil monitoring for different scenarios.

2. A regional soil monitoring method according to claim 1, wherein: By dividing the planting area and growth cycle, obtaining remote sensing image data of the planting area, including: According to geographical location, topography, soil conditions, and climate conditions, dividing the region into different planting areas, and using historical data to clarify the vegetation types planted in different areas; Extracting the boundary information of specific plots through high-resolution remote sensing images, and dividing the growth cycle into several stages according to the growth characteristics of different crops; Combining local climate data and crop growth cycles to determine the time range of each growth cycle, and monitoring vegetation indicators within each growth cycle; Setting the monitored vegetation indicators within the time window of the crop growth cycle, and evaluating the crop growth status through the vegetation indicators.

3. A regional soil monitoring method according to claim 2, wherein: Using the vegetation indicators to combine with the wireless sensor network nodes to establish a positioning model to obtain positioning data, including: Randomly setting several sensors in the vegetation areas divided within the region, setting the sensor nodes as beacon nodes, and transmitting node data to the network in a flooding manner; Extracting the shortest path between the beacon nodes according to the node data, and recording the minimum hop count and the node numbers passed on the shortest path; Calculating the average hop distance between the corresponding beacon nodes at both ends of the path according to the shortest path, and sorting the shortest paths between the beacon nodes in ascending order of the average hop distance; Successively traversing the unknown nodes on the sorted shortest path, obtaining the node information corresponding to the unknown nodes, adding the unknown nodes to the beacon nodes, and planning all the beacon nodes into the positioning sequence to obtain the positioning data within the region.

4. A regional soil monitoring method according to claim 3, wherein: Using the sine-cosine algorithm to optimize the positioning data to obtain position information, and monitoring the soil environment in real time through the position information to obtain real-time soil moisture data, including: Randomly initializing the positioning data to obtain N position individuals, performing a global optimization search on the N position individuals, and optimizing the individual positions using the change of the sine function. The expression is: Among them, represents the value of the i-th position component of the individual in the (t + 1)-th iteration optimization process. r1 and r2 are random numbers, and the range of r2 is [0, 2π]. represents the optimal position in the t-th iteration process. represents the value of the i-th position component of the individual in the t-th iteration optimization process; The fitness value of the individual position is calculated using the fitness function and sorted to obtain the optimal position of the current individual and the optimal position of the population. The individual position is continuously updated using the periodicity of the sine function, and boundary control is performed on the individual position. Individuals exceeding the boundary are re-initialized. By determining whether the global optimization search process falls into a local optimum, the fitness value of each individual position is calculated, and the optimal position of the current iterative individual and the historical optimal position of the population are updated. According to the optimal position of the individual and the historical optimal position of the population, the environmental data of the soil in the planning area is monitored in real time, and the real-time soil moisture data is calculated based on the environmental data of the soil.

5. A regional soil monitoring method according to claim 4, characterized in that The soil moisture data is corrected using multiple vegetation indices, and a convolutional neural network random detection model is used to identify the soil texture to obtain soil data characteristics, including: The multiple vegetation indices are divided into the normalized difference vegetation index NDVI, the enhanced vegetation index EVI, and the soil-adjusted vegetation index SAVI. A data characteristic sequence of the multiple vegetation indices, the normalized difference vegetation index NDVI, the enhanced vegetation index EVI, and the soil-adjusted vegetation index SAVI is established, and the correlation of the soil moisture data after being corrected by the multiple vegetation indices is analyzed to obtain the corrected soil moisture data. The corrected soil moisture data is used as training data, and a convolutional neural network is used to train the soil moisture data to extract soil moisture characteristic data. The soil texture classification is performed on the soil moisture characteristic data to obtain real-time soil data characteristics.

6. A regional soil monitoring method according to claim 5, characterized in that The soil moisture values at different depths are divided through the soil data characteristics, including: The normalized difference vegetation index NDVI is used to construct the soil moisture values at different depths for the soil data characteristics. Using the corrected normalized difference vegetation index NDVI data, with the crop growth period time of the soil area to be measured as the horizontal axis and the pixel-corrected NDVI data as the vertical axis, a correlation curve is established. Regression analysis is performed on the soil moisture at different depths and each vegetation index to obtain the change trends of the soil moisture at different depths and each vegetation index.

7. A regional soil monitoring method according to claim 6, characterized in that Correlation analysis is performed on the multiple vegetation indices according to the change trend of the soil moisture to obtain parameter factors affecting the soil moisture, including: A multiple linear regression equation is established for the soil moisture at different depths and each vegetation index according to the change trend of the soil moisture. The corresponding depths of the soil moisture are adjusted to 0-10 cm and 10-20 cm, and the soil moisture values are obtained respectively. According to the regression trend of the soil moisture and each vegetation index, the influencing factors under different soil moisture levels are analyzed to obtain the parameter factors affecting the soil moisture.

8. A regional soil monitoring method according to claim 7, characterized in that A soil texture recognition model is used to identify the soil sample characteristics, including: The soil texture recognition model uses a convolutional neural network with two convolutional layers and three pooling layers to extract the particle features, color features, and texture features of the soil from the extracted soil images. It transfers the two-dimensional output of the largest pooling layer to a one-dimensional output through a flattening layer, connects the convolutional layer and the max pooling layer maxpool, and passes through the fully connected layer; In the fully connected layer, a kernel function is used to classify the extracted features, and image features are extracted based on the parameter factor combined with the soil moisture value; The RF model is used to predict the soil sample features according to the random combination of the image features, and the soil texture feature data is obtained.

9. A regional soil monitoring method according to claim 7, characterized in that A soil texture recognition platform is established based on the soil texture feature data to monitor the soil type in real time, including: Vue is used to build the user interface of the soil texture recognition platform, and the Java language is combined to process the soil texture feature data to obtain the geographical information, data features, and multi-vegetation indices of the corresponding soil area; Using Springboot technology to manage and identify soil information, record the type of soil, obtain the differences in soil in each area, and the soil texture.

10. A regional soil monitoring method according to claim 9, characterized in that A soil monitoring positioning model is established through the soil texture recognition platform to perform real-time soil monitoring for different scenarios, including: On the soil texture recognition platform, the node information based on the sensor is obtained in real time. Beacon node coordinates are selected from the node information, and a non-linear equation set is established for the beacon node coordinates and distance information. Its expression is: Where (x,y) represents the unknown node coordinates, (x1, y1), (x2, y2), (x3, y3) respectively represent the coordinates of three beacon nodes, and d1, d2, d3 respectively represent the shortest distances from the unknown node to the three beacon nodes; By using the sine-cosine algorithm to perform iterative optimization and search for d1, d2, d3, the unknown node coordinates are solved to obtain the positioning coordinates of the soil area, and real-time soil monitoring is performed in different scenarios.

Images