Black land soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion

Abstract

The invention discloses a dynamic black land soil nutrient prediction system based on satellite remote sensing and multi-source data fusion, and relates to the field of artificial intelligence, and the system comprises a data collection module which obtains multi-source data of black land soil, including satellite remote sensing data, ground sensor data, meteorological data and historical soil nutrient data; the feature extraction module is used for extracting feature information of soil nutrients; the prediction model module is used for constructing a soil nutrient dynamic prediction model and accurately predicting the soil nutrient dynamic state; the feedback optimization module is used for optimizing and updating the prediction model, continuously improving the performance of the model, improving the prediction precision and obtaining an optimized prediction model; the visualization module is used for visually displaying a prediction result, displaying spatial and temporal distribution of soil nutrients in the form of a map and a chart, and providing strategy support of fertilization suggestions and land management; according to the data storage and management module, a database stores different types of data, efficient query, updating and maintenance of the data are achieved, and the safety of the soil nutrient dynamic data is improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion. Background Art

[0002] Black soil, a scarce arable land resource, is one of the world's most fertile soils, known as the "giant panda of arable land." It serves as the ballast and stabilizer for national food security. However, due to factors such as long-term overcultivation, irrational land use, excessive use of fertilizers and pesticides, and climate change, black soil is facing degradation issues such as soil erosion, organic matter depletion, nutrient loss, soil acidification, and salinization. These issues have led to a decline in its fertility, impacting crop growth and yields.

[0003] Traditional soil nutrient analysis relies on ground sampling and laboratory testing, which is time-consuming, labor-intensive, and complex to operate, making it difficult to monitor large areas of land in real time. Prediction models have limited spatiotemporal expression capabilities, making it difficult to capture the dynamic changes in soil nutrients. Data sources are limited, failing to fully leverage the complementarity of multiple data sources, such as satellite remote sensing, ground sensors, and meteorological data. The system lacks real-time feedback and optimization mechanisms, leaving the model's adaptability and prediction accuracy in need of improvement. The system's relatively limited functionality makes it difficult to meet the practical needs of precision agricultural management. Therefore, a dynamic prediction system for black soil nutrients based on satellite remote sensing and multi-source data fusion is urgently needed to achieve high-precision, real-time soil nutrient prediction and provide scientific decision-making support for agricultural production. Summary of the Invention

[0004] In order to solve the above technical problems, a black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion is provided. This technical solution solves the problems that the above-mentioned soil nutrient analysis relies on ground sampling and laboratory testing, which consumes a lot of time and manpower, is complicated to operate, and is difficult to achieve real-time monitoring of large areas of land; the data source is single, and it is impossible to fully utilize the complementarity of multiple data sources such as satellite remote sensing, ground sensors, and meteorological data; the system function is relatively single, and it is difficult to meet the actual needs of agricultural precision management.

[0005] In order to achieve the above objects, the technical solution adopted by the present invention is:

[0006] A black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion, including:

[0007] Data acquisition module: used to obtain multi-source data on black soil, including satellite remote sensing data, ground sensor data, meteorological data, and historical soil nutrient data;

[0008] Feature extraction module: The feature extraction module is electrically connected to the data acquisition module to extract characteristic information related to soil nutrients from the pre-processed data, including organic matter content, aggregate structure, and pH value;

[0009] Prediction model module: The prediction model module and the feature extraction module are electrically connected. Based on the extracted feature information, a soil nutrient dynamic prediction model is constructed to accurately predict soil nutrient dynamics.

[0010] Feedback optimization module: The feedback optimization module is electrically connected to the prediction model module and is used to optimize and update the prediction model according to actual measurement values, continuously improve model performance, enhance prediction accuracy, and obtain an optimized prediction model;

[0011] Visualization module: The visualization interface module and the feedback optimization module are electrically connected to intuitively display the prediction results, showing the spatiotemporal distribution of soil nutrients in the form of maps and charts, and providing fertilization recommendations and land management strategy support;

[0012] Data storage and management module: The data storage and management module is electrically connected to the visualization module. The database stores different types of data to achieve efficient query, update and maintenance of data, thereby improving the security of soil nutrient dynamic data.

[0013] Preferably, the data acquisition module specifically includes:

[0014] Satellite Remote Sensing Technology Unit: Use high-resolution satellites to obtain spectral data and vegetation indices; use thermal infrared sensors to obtain surface temperature data;

[0015] Ground sensor network unit: deploys a distributed sensor network to monitor soil physical and chemical properties and meteorological parameters in real time, and uses wireless sensor networks to achieve wireless transmission of data.

[0016] Preferably, the feature extraction module specifically includes:

[0017] Spectral data analysis unit: Using spectral data analysis technology, a relationship model between spectral characteristics and organic matter content is established to estimate the organic matter content;

[0018] Sensor detection unit: Uses multiple sensors buried in the soil to monitor the soil's organic matter content, pH, structural changes, soil temperature and moisture content in real time.

[0019] Preferably, the spectral data analysis technology specifically includes:

[0020] Using the partial least squares regression statistical learning method, latent variables are extracted through matrix decomposition. The calculation formula is:

[0021] P=Zβ+∈

[0022] Where P is the organic matter content vector, Z is the spectral data matrix, β is the regression coefficient matrix, and ∈ is the residual vector.

[0023] Preferably, the prediction model module specifically includes:

[0024] Hybrid model construction unit: Use convolutional neural networks to extract spatial features from remote sensing images, and use long-short-term memory networks to capture long-term dependencies in time series data. Both are used to process data with spatial and temporal characteristics to achieve dynamic prediction of black soil nutrients.

[0025] Preferably, the convolutional neural network specifically includes:

[0026] The convolutional neural network performs feature extraction and prediction through convolutional layers, pooling layers, and fully connected layers. The calculation formula is:

[0027] y=f(V3·ReLU(V2·ReLU(V1·x+b1)+b2)+b3

[0028] Where V1, V2, and V3 are convolution kernel weights, b1, b2, and b3 are bias terms, and ReLU is the activation function.

[0029] Preferably, the long short-term memory network specifically includes:

[0030] The long short-term memory network processes time series data through a gating mechanism. The calculation formula is:

[0031]

[0032] Where a t is the spatial feature of the tth time step, h t-1 is the hidden state at the t-1th time step, c t-1 is the cell state at the t-1th time step, d t is the forget gate output of the t-th time step, i t is the input gate output at the tth time step, g t is the candidate cell state at the tth time step, c t is the cell state at the tth time step, o t is the output gate output at the tth time step, h t is the hidden state at the tth time step, w d 、w i 、w g 、w o is the weight matrix of the forget gate, input gate, candidate cell state, and output gate, s f 、s i 、s g 、s ois the bias term of the forget gate, input gate, candidate cell state, and output gate;

[0033] At the last time step T, the LSTM network outputs the hidden state h t , and pass it through the fully connected layer to get the prediction result:

[0034] N=W Y ·h t +s y

[0035] Where N is the final predicted soil nutrient value, W Y is the weight matrix of the fully connected layer, s y is the bias term.

[0036] Preferably, the feedback optimization module specifically includes:

[0037] Model Evaluation Unit: Use evaluation indicators to evaluate the prediction model. The mean square error evaluates the difference between the predicted value and the true value. The coefficient of determination measures the model's ability to explain data changes. The mean absolute error evaluates the average deviation between the predicted value and the true value.

[0038] Model Optimization Unit: Use grid search or Bayesian optimization to find the optimal hyperparameter combination and try different model architectures to improve model expressiveness;

[0039] Model update unit: Retrain the model using the latest dataset to ensure that the model can adapt to the new data distribution.

[0040] Preferably, the visualization module specifically includes:

[0041] Map display unit: Display the spatial distribution of soil nutrients through heat maps, and use dynamic maps to show the changing trend of soil nutrients over time;

[0042] Fertilization suggestion unit: Marks the fertilization requirements of different areas on the map and provides interactive functions. Users can click on a specific area to view detailed fertilization suggestions;

[0043] Land Health Assessment Unit: displays soil health status through radar charts and provides early warning functions, issuing alarms when soil health indicators exceed thresholds;

[0044] Land use planning unit: Mark the land use types of different areas on the map. Users can adjust land use planning and view the impact on soil nutrients.

[0045] Preferably, the data storage and management module specifically includes:

[0046] Database classification unit: Relational databases are used to store and manage structured data, including soil nutrient measurements and meteorological data; non-relational databases are used to store semi-structured or unstructured data, including remote sensing images and sensor time series data;

[0047] Data management unit: automatically collects and writes sensor and remote sensing data into the database; supports manual entry and provides an interface for administrators to supplement and correct data; records data modification history and supports backtracking to any version;

[0048] Data query unit: Create composite indexes or full-text indexes for commonly queried fields, partition large tables by time or geographic location, and improve query efficiency.

[0049] Compared with the prior art, the present invention has the following beneficial effects:

[0050] The present invention proposes a black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion, which fully utilizes the complementarity of multiple data sources such as satellite remote sensing, ground sensors, and meteorological data to realize real-time monitoring of large areas of land; realizes high-precision, real-time soil nutrient prediction through prediction models, and captures the dynamic changes of soil nutrients; improves the adaptability and prediction accuracy of the model through real-time feedback and optimization mechanisms; provides map display, fertilization recommendations, land health assessment, and land use planning through visualization modules, and clearly provides scientific decision-making support for agricultural production to meet the actual needs of precise agricultural management. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is the framework diagram of the black soil nutrient dynamic prediction system based on source data fusion;

[0052] Figure 2 This is the internal framework diagram of the data acquisition module;

[0053] Figure 3 This is the internal framework diagram of the feature extraction module;

[0054] Figure 4 This is the internal framework diagram of the feedback optimization module;

[0055] Figure 5 It is the internal framework diagram of the visualization module;

[0056] Figure 6 Internal framework diagram of the data storage and management module. DETAILED DESCRIPTION

[0057] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are merely examples, and those skilled in the art may conceive of other obvious variations.

[0058] Reference Figure 1 As shown in the figure, a black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion includes:

[0059] Data acquisition module: used to obtain multi-source data on black soil, including satellite remote sensing data, ground sensor data, meteorological data, and historical soil nutrient data;

[0060] Feature extraction module: The feature extraction module is electrically connected to the data acquisition module to extract characteristic information related to soil nutrients from the pre-processed data, including organic matter content, aggregate structure, and pH value;

[0061] Prediction model module: The prediction model module and the feature extraction module are electrically connected. Based on the extracted feature information, a soil nutrient dynamic prediction model is constructed to accurately predict soil nutrient dynamics.

[0062] Feedback optimization module: The feedback optimization module is electrically connected to the prediction model module and is used to optimize and update the prediction model according to actual measurement values, continuously improve model performance, enhance prediction accuracy, and obtain an optimized prediction model;

[0063] Visualization module: The visualization interface module and the feedback optimization module are electrically connected to intuitively display the prediction results, showing the spatiotemporal distribution of soil nutrients in the form of maps and charts, and providing fertilization recommendations and land management strategy support;

[0064] Data storage and management module: The data storage and management module is electrically connected to the visualization module. The database stores different types of data to achieve efficient query, update and maintenance of data, thereby improving the security of soil nutrient dynamic data.

[0065] This solution achieves real-time monitoring of large areas of land by fully leveraging the complementarity of multiple data sources such as satellite remote sensing, ground sensors, and meteorological data. It uses prediction models to achieve high-precision, real-time soil nutrient predictions and capture the dynamic changes in soil nutrients. It improves the adaptability and prediction accuracy of the model through real-time feedback and optimization mechanisms. It provides map display, fertilization recommendations, land health assessment, and land use planning through visualization modules, providing clear and scientific decision-making support for agricultural production and meeting the actual needs of precision agricultural management.

[0066] Reference Figure 2 As shown, the data acquisition module includes:

[0067] Satellite Remote Sensing Technology Unit: Use high-resolution satellites to obtain spectral data and vegetation indices; use thermal infrared sensors to obtain surface temperature data;

[0068] Ground sensor network unit: deploys a distributed sensor network to monitor soil physical and chemical properties and meteorological parameters in real time, and uses wireless sensor networks to achieve wireless transmission of data.

[0069] Reference Figure 3 As shown, the feature extraction module includes:

[0070] Spectral data analysis unit: Using spectral data analysis technology, a relationship model between spectral characteristics and organic matter content is established to estimate the organic matter content;

[0071] Sensor detection unit: Uses multiple sensors buried in the soil to monitor the soil's organic matter content, pH, structural changes, soil temperature and moisture content in real time.

[0072] Spectral data analysis technology: Using partial least squares regression statistical learning method, latent variables are extracted through matrix decomposition. The calculation formula is:

[0073] P=Zβ+∈

[0074] Where P is the organic matter content vector, Z is the spectral data matrix, β is the regression coefficient matrix, and ∈ is the residual vector.

[0075] It should be noted that spectral data is used to obtain the reflectance spectrum of the soil surface for preliminary judgment of soil type and health status; the normalized vegetation index and leaf area index are used to reflect vegetation coverage and growth, and indirectly reflect soil health status; satellite remote sensing data and ground sensor data jointly provide data support for the prediction model.

[0076] Reference Figure 4 As shown, the feedback optimization module includes:

[0077] Model Evaluation Unit: Use evaluation indicators to evaluate the prediction model. The mean square error evaluates the difference between the predicted value and the true value. The coefficient of determination measures the model's ability to explain data changes. The mean absolute error evaluates the average deviation between the predicted value and the true value.

[0078] Model Optimization Unit: Use grid search or Bayesian optimization to find the optimal hyperparameter combination and try different model architectures to improve model expressiveness;

[0079] Model update unit: retrains the model using the latest dataset to ensure that the model can adapt to the new data distribution;

[0080] Reference Figure 5 As shown, the visualization module includes:

[0081] Map display unit: Display the spatial distribution of soil nutrients through heat maps, and use dynamic maps to show the changing trend of soil nutrients over time;

[0082] Fertilization suggestion unit: Marks the fertilization requirements of different areas on the map and provides interactive functions. Users can click on a specific area to view detailed fertilization suggestions;

[0083] Land Health Assessment Unit: displays soil health status through radar charts and provides early warning functions, issuing alarms when soil health indicators exceed thresholds;

[0084] Land use planning unit: Mark the land use types of different areas on the map. Users can adjust land use planning and view the impact on soil nutrients.

[0085] It should be noted that the heat map in the visualization module: displays the spatial distribution of soil nutrients through color gradients, such as areas with high and low organic matter content; thematic map: marks the soil nutrient levels of different areas on the map, such as high, medium and low nutrient areas; dynamic map: displays the changing trend of soil nutrients over time through the time axis; uses Three.js or Cesium to realize the three-dimensional spatial distribution display of soil nutrients; historical data review: displays the changing trend of soil nutrients in the past period of time through time series charts, provides historical data comparison function, and helps users analyze the changing patterns of soil nutrients.

[0086] Reference Figure 6 As shown, the data storage and management module includes:

[0087] Database classification unit: Relational databases are used to store and manage structured data, including soil nutrient measurements and meteorological data; non-relational databases are used to store semi-structured or unstructured data, including remote sensing images and sensor time series data;

[0088] Data management unit: automatically collects and writes sensor and remote sensing data into the database; supports manual entry and provides an interface for administrators to supplement and correct data; records data modification history and supports backtracking to any version;

[0089] Data query unit: Create composite indexes or full-text indexes for commonly queried fields, partition large tables by time or geographic location, and improve query efficiency.

[0090] In summary, the advantages of the present invention are: making full use of the complementarity of multiple data sources such as satellite remote sensing, ground sensors, and meteorological data to achieve real-time monitoring of large areas of land; achieving high-precision, real-time soil nutrient prediction through prediction models, and capturing the dynamic changes of soil nutrients; improving the adaptability and prediction accuracy of the model through real-time feedback and optimization mechanisms; providing map display, fertilization recommendations, land health assessment, and land use planning through visualization modules, providing clear and scientific decision-making support for agricultural production, and meeting the actual needs of precise agricultural management.

[0091] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions merely illustrate the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion, characterized in that: include: Data acquisition module: used to obtain multi-source data on black soil, including satellite remote sensing data, ground sensor data, meteorological data, and historical soil nutrient data; Feature extraction module: The feature extraction module is electrically connected to the data acquisition module to extract characteristic information related to soil nutrients from the pre-processed data, including organic matter content, aggregate structure, and pH value; Prediction model module: The prediction model module and the feature extraction module are electrically connected. Based on the extracted feature information, a soil nutrient dynamic prediction model is constructed to accurately predict soil nutrient dynamics. Feedback optimization module: The feedback optimization module is electrically connected to the prediction model module and is used to optimize and update the prediction model according to actual measurement values, continuously improve model performance, enhance prediction accuracy, and obtain an optimized prediction model; Visualization module: The visualization interface module and the feedback optimization module are electrically connected to intuitively display the prediction results, showing the spatiotemporal distribution of soil nutrients in the form of maps and charts, and providing fertilization recommendations and land management strategy support; Data storage and management module: The data storage and management module is electrically connected to the visualization module. The database stores different types of data to achieve efficient query, update and maintenance of data, thereby improving the security of soil nutrient dynamic data.

2. The black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion according to claim 1 is characterized in that: The data acquisition module specifically includes: Satellite Remote Sensing Technology Unit: Use high-resolution satellites to obtain spectral data and vegetation indices; use thermal infrared sensors to obtain surface temperature data; Ground sensor network unit: deploys a distributed sensor network to monitor soil physical and chemical properties and meteorological parameters in real time, and uses wireless sensor networks to achieve wireless transmission of data.

3. The black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion according to claim 2 is characterized in that: The feature extraction module specifically includes: Spectral data analysis unit: Using spectral data analysis technology, a relationship model between spectral characteristics and organic matter content is established to estimate the organic matter content; Sensor detection unit: Uses multiple sensors buried in the soil to monitor the soil's organic matter content, pH, structural changes, soil temperature and moisture content in real time.

4. The black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion according to claim 3 is characterized in that: The spectral data analysis technology specifically includes: Using the partial least squares regression statistical learning method, latent variables are extracted through matrix decomposition. The calculation formula is: P=Zβ+∈ Where P is the organic matter content vector, Z is the spectral data matrix, β is the regression coefficient matrix, and ∈ is the residual vector.

5. The black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion according to claim 3 is characterized in that: The prediction model module specifically includes: Hybrid model construction unit: Use convolutional neural networks to extract spatial features from remote sensing images, and use long-short-term memory networks to capture long-term dependencies in time series data. Both are used to process data with spatial and temporal characteristics to achieve dynamic prediction of black soil nutrients.

6. The black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion according to claim 5 is characterized in that: The convolutional neural network specifically includes: The convolutional neural network performs feature extraction and prediction through convolutional layers, pooling layers, and fully connected layers. The calculation formula is: y=f(V3·ReLU(V2·ReLU(V1·x+b1)+b2)+b3 Among them, V1, V2, V3 are convolution kernel weights, b1, b2, b3 are bias terms, and ReLU is the activation function.

7. The black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion according to claim 5 is characterized in that: The long short-term memory network specifically includes: The long short-term memory network processes time series data through a gating mechanism. The calculation formula is: Where a t is the spatial feature of the tth time step, h t-1 is the hidden state at the t-1th time step, c t-1 is the cell state at the t-1th time step, d t is the forget gate output of the t-th time step, i t is the input gate output at the tth time step, g t is the candidate cell state at the tth time step, c t is the cell state at the tth time step, o t is the output gate output at the tth time step, h t is the hidden state at the tth time step, w d 、w i 、w g 、w o is the weight matrix of the forget gate, input gate, candidate cell state, and output gate, s f 、s i 、s g 、s o is the bias term of the forget gate, input gate, candidate cell state, and output gate; At the last time step T, the LSTM network outputs the hidden state h t , and pass it through the fully connected layer to get the prediction result: N=W Y ·h t +s y N is the final predicted soil nutrient value, W Y is the weight matrix of the fully connected layer, s y is the bias term.

8. The black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion according to claim 5 is characterized in that: The feedback optimization module specifically includes: Model Evaluation Unit: Use evaluation indicators to evaluate the prediction model. The mean square error evaluates the difference between the predicted value and the true value. The coefficient of determination measures the model's ability to explain data changes. The mean absolute error evaluates the average deviation between the predicted value and the true value. Model Optimization Unit: Use grid search or Bayesian optimization to find the optimal hyperparameter combination and try different model architectures to improve model expressiveness; Model update unit: Retrain the model using the latest dataset to ensure that the model can adapt to the new data distribution.

9. The black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion according to claim 8 is characterized in that: The visualization module specifically includes: Map display unit: Display the spatial distribution of soil nutrients through heat maps, and use dynamic maps to show the changing trend of soil nutrients over time; Fertilization suggestion unit: Marks the fertilization requirements of different areas on the map and provides interactive functions. Users can click on a specific area to view detailed fertilization suggestions; Land Health Assessment Unit: displays soil health status through radar charts and provides early warning functions, issuing alarms when soil health indicators exceed thresholds; Land use planning unit: Mark the land use types of different areas on the map. Users can adjust land use planning and view the impact on soil nutrients.

10. The black soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion according to claim 9 is characterized in that: The data storage and management module specifically includes: Database classification unit: Relational databases are used to store and manage structured data, including soil nutrient measurements and meteorological data; non-relational databases are used to store semi-structured or unstructured data, including remote sensing images and sensor time series data; Data management unit: automatically collects and writes sensor and remote sensing data into the database; supports manual entry and provides an interface for administrators to supplement and correct data; records data modification history and supports backtracking to any version; Data query unit: Create composite indexes or full-text indexes for commonly queried fields, partition large tables by time or geographic location, and improve query efficiency.

Images