Climate change trend-based corn planting risk prediction method and system

Abstract

The invention discloses a corn planting risk prediction method and system based on a climate change trend, and relates to the technical field of environment prediction. The method comprises the following steps: acquiring climate-related data of corn planting; constructing a risk prediction model; the risk prediction model adopts an LSTM-Transform hybrid neural network to process time sequence meteorological data, introduces an SHAP value interpretation model to quantify the contribution degree of each climate factor to the output reduction risk, and couples a DSSAT crop model and a risk probability prediction module; inputting the climate-related data into the risk prediction model, and determining a final corn planting risk result through operation of the risk prediction model; the corn planting risk result comprises a regional risk grade map, planting window period optimization suggestions and extreme weather event probability early warning. According to the invention, the accuracy of corn planting risk prediction can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of environmental prediction, and in particular to a method and system for predicting corn planting risks based on climate change trends. Background Art

[0002] As one of the world's most important food crops, corn's yield stability is directly related to food security and the sustainable development of the agricultural economy. However, in recent years, due to the impact of global climate change, extreme weather events (such as droughts, floods, and high temperatures) have occurred frequently, resulting in a significant increase in the uncertainty faced by corn cultivation. Traditional risk assessment methods mainly rely on historical meteorological statistics or empirical models, which are difficult to accurately capture the nonlinear dynamic characteristics under climate change, and lack in-depth fusion analysis of multi-source climate data, resulting in insufficient prediction accuracy. In addition, existing models often ignore the analysis of the contribution of key climate factors to the risk of yield reduction, limiting the targeted nature of risk prevention and control strategies. Summary of the Invention

[0003] The purpose of the present invention is to provide a corn planting risk prediction method and system based on climate change trends, which can improve the accuracy of corn planting risk prediction.

[0004] To achieve the above object, the present invention provides the following solutions:

[0005] A corn planting risk prediction method based on climate change trends, comprising:

[0006] Obtaining climate-related data for corn planting; the climate-related data includes remote sensing data and field planting data; the remote sensing data includes CMIP6 climate scenario data and satellite remote sensing vegetation index;

[0007] Construct a risk prediction model; the risk prediction model uses an LSTM-Transformer hybrid neural network to process time-series meteorological data, introduces a SHAP value interpretation model to quantify the contribution of various climate factors to yield reduction risk, and couples the DSSAT crop model with a risk probability prediction module;

[0008] The climate-related data is input into the risk prediction model, and the final corn planting risk result is determined through calculation by the risk prediction model; the corn planting risk result includes a regional risk level map, planting window optimization suggestions and extreme weather event probability warning.

[0009] Optionally, the process of acquiring climate-related data for corn planting includes:

[0010] Generate CMIP6 climate scenario data using GCMs model simulation, and generate satellite remote sensing vegetation index through satellite remote sensing inversion, package the CMIP6 climate scenario data and the satellite remote sensing vegetation index into a data file, and name the file remote sensing data;

[0011] Build a field IoT sensor network, use the field IoT sensor network to collect data in the target area, and store the collected data in a file package named field planting data;

[0012] The remote sensing data and the field planting data are hash-encrypted and stored in the cloud blockchain. When a data extraction request is received, identity authentication is performed using the public and private keys of both parties. After the identity authentication is passed, the data is decrypted and extracted.

[0013] Optionally, the risk prediction model construction process includes:

[0014] Obtaining a training data set; the training data set includes training data and corresponding training labels;

[0015] Build a pre-trained network based on a dual-driven framework of machine learning and mechanism models;

[0016] The training data set is input into the pre-trained network, parameter optimization is performed with the goal of minimizing the loss between the network output and the training label, and the pre-trained network using the optimized optimal parameters is determined as the risk prediction model.

[0017] Optionally, in the parameter optimization, a feedback optimization mechanism and a dynamic weight allocation method are adopted; the dynamic weight allocation method adopts dynamic weighted loss.

[0018] Optionally, inputting the climate-related data into the risk prediction model and calculating the risk prediction model to determine the final corn planting risk result specifically includes:

[0019] The climate-related data is input into the risk prediction model, and the abnormal meteorological fluctuations are first processed using a sliding window, and then feature extraction and prediction are performed to determine the final corn planting risk result.

[0020] Optionally, it also includes: building a visual display platform.

[0021] Optionally, the construction process of the visualization display platform includes:

[0022] Construct a sliding dual platform; the sliding dual platform includes a first display platform and a second display platform, and sliding between platform pages is achieved through a touch screen; wherein, the first display platform is used to dynamically display the climate-related data through three interfaces according to different data types; the second display platform is used to display the current forecast results.

[0023] The present invention also provides a corn planting risk prediction system based on climate change trends, comprising:

[0024] A data acquisition unit, configured to acquire climate-related data for corn planting; the climate-related data includes remote sensing data and field planting data; the remote sensing data includes CMIP6 climate scenario data and satellite remote sensing vegetation index;

[0025] A model building unit is used to build a risk prediction model; the risk prediction model uses an LSTM-Transformer hybrid neural network to process time series meteorological data, introduces a SHAP value interpretation model to quantify the contribution of various climate factors to the risk of yield reduction, and couples the DSSAT crop model with a risk probability prediction module;

[0026] The prediction unit is used to input the climate-related data into the risk prediction model, and determine the final corn planting risk result through calculation by the risk prediction model; the corn planting risk result includes a regional risk level map, planting window optimization suggestions and extreme weather event probability warning.

[0027] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0028] The present invention discloses a corn planting risk prediction method and system based on climate change trends. The method comprises obtaining climate-related data for corn planting; the climate-related data comprises remote sensing data and field planting data; the remote sensing data comprises CMIP6 climate scenario data and satellite remote sensing vegetation index; a risk prediction model is constructed; the risk prediction model uses an LSTM-Transformer hybrid neural network to process time series meteorological data, introduces a SHAP value interpretation model to quantify the contribution of each climate factor to the yield reduction risk, and couples a DSSAT crop model with a risk probability prediction module; the climate-related data is input into the risk prediction model, and the risk prediction model is operated to determine a final corn planting risk result; the corn planting risk result comprises a regionalized risk level map, a planting window optimization suggestion, and an extreme weather event probability warning. The present invention can integrate multi-source data and hybrid modeling technology to quantify the risk contribution of climate factors, optimize planting decisions, and enhance extreme weather response capabilities, thereby further improving the accuracy of corn planting risk prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0030] Figure 1 The figure is a flow chart of the corn planting risk prediction method based on climate change trends of the present invention. DETAILED DESCRIPTION

[0031] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0032] The purpose of the present invention is to provide a corn planting risk prediction method and system based on climate change trends, which can improve the accuracy of corn planting risk prediction.

[0033] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0034] like Figure 1 As shown, the present invention provides a corn planting risk prediction method based on climate change trends, comprising:

[0035] Step 100: Acquire climate-related data for corn planting; the climate-related data includes remote sensing data and field planting data; the remote sensing data includes CMIP6 climate scenario data and satellite remote sensing vegetation index.

[0036] Step 200: Construct a risk prediction model; the risk prediction model uses an LSTM-Transformer hybrid neural network to process time series meteorological data, introduces a SHAP value interpretation model to quantify the contribution of each climate factor to the risk of yield reduction, and couples the DSSAT crop model with a risk probability prediction module.

[0037] Step 300: Input the climate-related data into the risk prediction model, and determine the final corn planting risk result through calculation by the risk prediction model; the corn planting risk result includes a regional risk level map, planting window optimization suggestions, and extreme weather event probability warnings.

[0038] As a specific implementation method, the process of obtaining climate-related data for corn planting includes:

[0039] The CMIP6 climate scenario data is generated by simulating the GCMs model, and the satellite remote sensing vegetation index is generated by satellite remote sensing inversion. The CMIP6 climate scenario data and the satellite remote sensing vegetation index are packaged into a data file, and the file is named remote sensing data.

[0040] Build a field Internet of Things sensor network, use the field Internet of Things sensor network to collect data in the target area, and store the collected data in a file package named field planting data.

[0041] The remote sensing data and the field planting data are hash-encrypted and stored in the cloud blockchain. When a data extraction request is received, identity authentication is performed using the public and private keys of both parties. After the identity authentication is passed, the data is decrypted and extracted.

[0042] As a specific implementation, the process of constructing the risk prediction model includes:

[0043] A training dataset is obtained; the training dataset includes training data and corresponding training labels; a pre-trained network is constructed based on a dual-driven framework of machine learning and mechanistic models; the training dataset is input into the pre-trained network, and parameter optimization is performed with the goal of minimizing the loss between the network output and the training labels. The pre-trained network using the optimized parameters is then determined as the risk prediction model. The parameter optimization utilizes a feedback optimization mechanism and a dynamic weight allocation method; the dynamic weight allocation method utilizes a dynamic weighted loss.

[0044] For the core loss function in training, a combination of the main loss function and the auxiliary loss function is used:

[0045] (1) Main loss function (regression task)

[0046] Use weighted mean square error (Weighted-MSE) to train continuous variables (such as yield reduction rate and soil moisture content):

[0047]

[0048] Among them, the dynamic weight w i Adjustments are made based on the degree of climate anomaly of the sample (for example, the weight is increased three times when the drought index is greater than 2σ), and meteorological sensitivity is taken into account, so that higher penalties are imposed on prediction errors during key growth periods of corn (such as the tasting period).

[0049] (2) Auxiliary loss function (classification task)

[0050] Use Focal Loss to train disaster level classification (alleviate category imbalance):

[0051]

[0052] Among them, the parameter γ is set to 2 (focusing on difficult samples), α_t is inversely weighted according to the frequency of disaster level (such as α = 0.7 for level 5 risk) and the threshold is adjusted, and the ROC curve is used to optimize the classification threshold offset (to adapt to the distribution drift caused by climate change).

[0053] Then, multi-task learning integration is performed, and the loss weights of each task are automatically learned using dynamic weighted loss:

[0054]

[0055] Among them, σ1 and σ2 are trainable parameters that reflect the uncertainty of regression and classification tasks. This step can avoid manual parameter adjustment and adapt to the differences in climate prediction difficulty in different regions.

[0056] Finally, for special scene enhancements:

[0057] (1) Extreme event simulation loss: Enhance model robustness by adversarial generation samples (GAN generates rare climate patterns):

[0058]

[0059] Used to generate complex disaster events that have not appeared in history (such as high temperatures, droughts and cyclones).

[0060] (2) Physical constraint loss: Introducing agricultural mechanism model constraints (such as water stress equation):

[0061]

[0062] Where λ is the adaptive coefficient (increasing physical constraints when model confidence is low).

[0063] As a specific implementation manner, inputting the climate-related data into the risk prediction model, and determining the final corn planting risk result through operation of the risk prediction model specifically includes:

[0064] The climate-related data is input into the risk prediction model, and the abnormal meteorological fluctuations are first processed using a sliding window, and then feature extraction and prediction are performed to determine the final corn planting risk result.

[0065] As a specific implementation method, after determining the final corn planting risk result, it also includes: building a visual display platform, and the construction process includes:

[0066] Construct a sliding dual platform; the sliding dual platform includes a first display platform and a second display platform, and sliding between platform pages is achieved through a touch screen; wherein, the first display platform is used to dynamically display the climate-related data through three interfaces according to different data types; the second display platform is used to display the current forecast results.

[0067] Based on the above technical solution, the following embodiments are provided.

[0068] Step 100: Data acquisition.

[0069] (1) Remote sensing data acquisition: Global climate models (GCMs) are used to simulate and generate CMIP6 climate scenario data, including temperature, precipitation, radiation, and other meteorological data under different future emission scenarios (e.g., SSP1-2.6, SSP5-8.5). Vegetation indices (NDVI / EVI) are obtained through satellite remote sensing (e.g., Landsat, MODIS) to monitor corn growth. CMIP6 data and satellite remote sensing vegetation indices are integrated and packaged into remote sensing data files.

[0070] (2) Field planting data acquisition: Deploy a field IoT sensor network (such as soil temperature and humidity sensors, weather stations, and crop growth monitors) in the target corn planting area to collect real-time data on soil moisture, temperature, precipitation, and crop growth. The collected data is stored in a field planting data file package.

[0071] (3) Data storage and secure transmission: SHA-256 hash encryption is used to encrypt remote sensing data and field data and upload them to a cloud blockchain (such as the agricultural data chain based on Hyperledger Fabric). When a user requests data, asymmetric encryption (RSA) is used for authentication (public key / private key matching). After verification, the data is decrypted for model invocation.

[0072] Step 200: Constructing a risk prediction model.

[0073] (1) Preparation of training dataset: Collect historical climate data, corn yield data, and disaster records to construct a labeled training dataset (input: meteorological time series data; label: yield reduction risk level).

[0074] (2) Hybrid neural network architecture: A LSTM-Transformer hybrid model is used, consisting of an LSTM layer and a Transformer layer. The LSTM layer is used to process the long-term dependencies of meteorological data (e.g., a 30-year temperature series); the Transformer layer is used to capture the global characteristics of key climate factors (e.g., the attention weight of extreme high temperature events). SHAP value analysis is introduced to quantify the contribution of various climate factors (e.g., average temperature in July, growing season precipitation) to the risk of yield reduction.

[0075] (3) Coupling DSSAT crop model: The physiological responses of corn (such as photosynthesis and water stress) simulated by DSSAT are used as physical constraints and dynamically weighted fused with the prediction results of LSTM-Transformer (the weights are adjusted according to the season).

[0076] (4) Dynamic optimization training: A feedback optimization mechanism is used to adjust the weights of LSTM and Transformer in real time based on the prediction error. A dynamic weighted loss function is used to assign higher weights to data from high-impact periods (such as the heading period).

[0077] Step 300: Risk prediction and result output.

[0078] (1) Data preprocessing: A sliding window method (window size 30 days, step length 7 days) is used to smooth abnormal fluctuations (such as sudden rainstorms) in the input climate data.

[0079] (2) Model prediction:

[0080] The model takes processed data as input and outputs: regional risk level maps (e.g., high-risk areas: probability of high temperatures >60% in July and August in the Huanghuaihai Plain under the future RCP8.5 scenario); planting window optimization recommendations (e.g., in Northeast China, it is recommended to plant 5-10 days earlier to avoid later drought); and extreme weather probability warnings (e.g., the probability of continuous drought during the corn tasseling period has increased by 25% over the next 10 years).

[0081] (3) Visual display platform:

[0082] Sliding dual platform design:

[0083] First platform (dynamic data display): Screen 1: Real-time meteorological data heat map (such as temperature and precipitation anomalies). Screen 2: Satellite remote sensing vegetation index change animation. Screen 3: Comparative analysis of historical disaster events.

[0084] The second platform (display of prediction results): An interactive map displays the risk level and supports single-click query of optimization measures for specific areas (such as irrigation plans).

[0085] Switch platforms by sliding on the touch screen to improve user experience.

[0086] Taking the Huanghuaihai summer corn region in China as an example, after applying this method, the results are as follows:

[0087] Inputting the 2030 SSP5-8.5 scenario data, the model prediction shows:

[0088] High-risk area: Central Henan (70% probability of high temperature in July), it is recommended to plant heat-resistant varieties.

[0089] Window period adjustment: Shandong recommends postponing sowing until mid-June to avoid drought during flowering.

[0090] Platform warning: The frequency of extreme droughts in 2030-2035 will increase by 15% compared to the 2020s, and drought-resistant facilities need to be planned in advance.

[0091] This example achieves accurate dynamic prediction of corn planting risks under climate change through multi-source data fusion, hybrid model optimization, and visual interaction, providing decision support for agricultural disaster prevention and mitigation.

[0092] In addition, the present invention also provides a corn planting risk prediction system based on climate change trends, comprising:

[0093] A data acquisition unit, configured to acquire climate-related data for corn planting; the climate-related data includes remote sensing data and field planting data; the remote sensing data includes CMIP6 climate scenario data and satellite remote sensing vegetation index;

[0094] A model building unit is used to build a risk prediction model; the risk prediction model uses an LSTM-Transformer hybrid neural network to process time series meteorological data, introduces a SHAP value interpretation model to quantify the contribution of various climate factors to the risk of yield reduction, and couples the DSSAT crop model with a risk probability prediction module;

[0095] The prediction unit is used to input the climate-related data into the risk prediction model, and determine the final corn planting risk result through calculation by the risk prediction model; the corn planting risk result includes a regional risk level map, planting window optimization suggestions and extreme weather event probability warning.

[0096] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0097] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A corn planting risk prediction method based on climate change trends, characterized in that: include: Obtaining climate-related data for corn planting; the climate-related data includes remote sensing data and field planting data; the remote sensing data includes CMIP6 climate scenario data and satellite remote sensing vegetation index; Construct a risk prediction model; the risk prediction model uses an LSTM-Transformer hybrid neural network to process time-series meteorological data, introduces a SHAP value interpretation model to quantify the contribution of various climate factors to yield reduction risk, and couples the DSSAT crop model with a risk probability prediction module; Inputting the climate-related data into the risk prediction model, and determining the final corn planting risk result through calculation by the risk prediction model; The corn planting risk results include regional risk level maps, planting window optimization recommendations, and extreme weather event probability warnings.

2. The corn planting risk prediction method based on climate change trend according to claim 1, characterized in that: The process of obtaining the climate-related data for corn planting includes: Generate CMIP6 climate scenario data using GCMs model simulation, and generate satellite remote sensing vegetation index through satellite remote sensing inversion, package the CMIP6 climate scenario data and the satellite remote sensing vegetation index into a data file, and name the file remote sensing data; Build a field IoT sensor network, use the field IoT sensor network to collect data in the target area, and store the collected data in a file package named field planting data; The remote sensing data and the field planting data are hash-encrypted and stored in the cloud blockchain. When a data extraction request is received, identity authentication is performed using the public and private keys of both parties. After the identity authentication is passed, the data is decrypted and extracted.

3. The corn planting risk prediction method based on climate change trend according to claim 1, characterized in that: The process of constructing the risk prediction model includes: Obtaining a training data set; the training data set includes training data and corresponding training labels; Build a pre-trained network based on a dual-driven framework of machine learning and mechanism models; The training data set is input into the pre-trained network, parameter optimization is performed with the goal of minimizing the loss between the network output and the training label, and the pre-trained network using the optimized optimal parameters is determined as the risk prediction model.

4. The corn planting risk prediction method based on climate change trend according to claim 3, characterized in that: In the parameter optimization, a feedback optimization mechanism and a dynamic weight allocation method are adopted; the dynamic weight allocation method adopts dynamic weighted loss.

5. The corn planting risk prediction method based on climate change trend according to claim 1, characterized in that: Inputting the climate-related data into the risk prediction model and calculating the risk prediction model to determine the final corn planting risk result specifically includes: The climate-related data is input into the risk prediction model, and the abnormal meteorological fluctuations are first processed using a sliding window, and then feature extraction and prediction are performed to determine the final corn planting risk result.

6. The corn planting risk prediction method based on climate change trend according to claim 1, characterized in that: Also includes: Build a visual display platform.

7. The corn planting risk prediction method based on climate change trend according to claim 6, characterized in that: The construction process of the visualization display platform includes: Construct a sliding dual platform; the sliding dual platform includes a first display platform and a second display platform, and sliding between platform pages is achieved through a touch screen; wherein, the first display platform is used to dynamically display the climate-related data through three interfaces according to different data types; the second display platform is used to display the current forecast results.

8. A corn planting risk prediction system based on climate change trends, characterized in that: include: A data acquisition unit, configured to acquire climate-related data for corn planting; the climate-related data includes remote sensing data and field planting data; the remote sensing data includes CMIP6 climate scenario data and satellite remote sensing vegetation index; A model building unit is used to build a risk prediction model; the risk prediction model uses an LSTM-Transformer hybrid neural network to process time series meteorological data, introduces a SHAP value interpretation model to quantify the contribution of various climate factors to the risk of yield reduction, and couples the DSSAT crop model with a risk probability prediction module; A prediction unit, configured to input the climate-related data into the risk prediction model, and determine a final corn planting risk result through calculation by the risk prediction model; The corn planting risk results include regional risk level maps, planting window optimization recommendations, and extreme weather event probability warnings.

Images