Agricultural disaster dynamic prediction and emergency decision-making method and system based on multi-source data

Abstract

The application discloses a method and system for dynamic prediction and emergency decision of agricultural disasters based on multi-source data, and relates to the technical field of agricultural disaster prevention. The method comprises the following steps: collecting multi-source historical data and real-time data; dividing the grades of disasters; taking crop data as output, other multi-source data and prediction time as input, and establishing a crop prediction model; calculating a super prediction factor for reflecting the self-recovery effect of the agricultural ecological environment; combining the crop prediction model and the super prediction factor to establish a disaster prediction model; obtaining a prediction result, combining the disaster grade, and performing a preset emergency decision scheme. The application uses the super prediction factor to represent the subsequent disaster dynamic situation of the to-be-predicted period, quantifies the self-recovery effect of the agricultural ecological environment, combines the crop prediction model, analyzes the agricultural disaster dynamic situation after the to-be-predicted period, and improves the prediction accuracy and the effectiveness of the emergency decision.

Description

Technical Field

[0001] This invention relates to the field of agricultural disaster prevention technology, specifically to a method and system for dynamic prediction and emergency decision-making of agricultural disasters based on multi-source data. Background Technology

[0002] Agricultural disaster prediction requires the comprehensive use of GIS, satellite remote sensing, and other technologies to collect a wide range of multi-source information, including meteorological, soil, and crop data. After preprocessing, key information is extracted to construct a prediction model, which is then optimized and adjusted using expert knowledge to ultimately arrive at a predicted agricultural disaster outcome. This provides a scientific basis for agricultural disaster prevention and control decisions, assists in formulating strategies such as strengthening water conservancy infrastructure construction, promoting water-saving irrigation technologies, adjusting planting structures, and enhancing farmland management. It helps agricultural producers conduct timely and precise disaster management, effectively reducing disaster losses, ensuring national food security, promoting sustainable agricultural development, and playing a vital supporting role in stabilizing the national economy.

[0003] When predicting agricultural disasters, it is necessary to collect information from multiple sources and build a prediction model. For example, the soil nutrient dynamic prediction system based on satellite remote sensing and multi-source data fusion, patent publication number CN118606645A, includes a data collection unit, a data processing and fusion unit, a prediction model unit, and a user interaction unit. The data collection unit includes satellite remote sensing data and auxiliary data sources. The data processing and fusion unit includes preprocessing, feature engineering, and data fusion. The prediction model unit includes model development, real-time updating and learning. The user interaction unit includes result display and decision support. This system improves and optimizes existing soil nutrient analysis methods to meet the needs of modern precision agriculture.

[0004] The combination of multi-source data can improve the accuracy of agricultural disaster prediction, facilitate the timely formulation of relevant preventive measures, and reduce disaster losses. The agricultural ecological environment has a certain self-recovery function. For general agricultural disasters, they may automatically recover over time without human intervention or with no positive effect. The above method analyzes soil nutrients by integrating multi-source data, which facilitates timely land fertilization and other operations. However, it does not take into account the self-recovery effect of the agricultural ecological environment, resulting in poor prediction accuracy. Summary of the Invention

[0005] The purpose of this invention is to provide a method and system for dynamic prediction and emergency decision-making of agricultural disasters based on multi-source data, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for dynamic prediction and emergency decision-making of agricultural disasters based on multi-source data, comprising:

[0007] Multi-source data collection: Collect historical and real-time data, both of which include meteorological, soil, and crop data for the crop growth period and several periods prior to the growth period.

[0008] Level Preset: Crop data in historical and real-time data are grouped by time period and subjected to minimum-maximum normalization. The disaster level is then divided according to the data ratio to represent different degrees of disaster.

[0009] Establish a crop prediction model: In historical data, divide the crop growth period into time periods. Starting from the second time period, select any time period as the target time period. Use the crop data of the target time period as the output. Take the meteorological data, soil data, and crop data of the time period before the target time period, as well as the time period number of the target time period, as the input. Select different target time periods to combine multiple training samples and establish a crop prediction model.

[0010] Calculate the superpredictor factor: The superpredictor factor is calculated by using the superpredictor factor calculation method to calculate the superpredictor factor of historical data after the target time period relative to the historical data of the target time period. It represents the subsequent dynamic situation of crop data in the target time period and is used to reflect the self-recovery effect of the agricultural ecological environment.

[0011] Establishing a disaster prediction model: By combining crop prediction models and super-predictive factors through disaster prediction model establishment methods, a disaster prediction model can be obtained;

[0012] Decision-making: Input real-time data into the disaster prediction model, output the prediction results, and combine them with the disaster level to formulate a pre-set decision-making plan.

[0013] Preferably, the method for calculating the superpredictive factor includes:

[0014] Mark the time period in historical data that corresponds to the time period to be predicted as the specified time period, and calculate the rate of change of crop data between the specified time period and the next time period in historical data;

[0015] Using crop data change rate as output, and meteorological data, soil data, and crop data from the previous time period as input, a change rate prediction model is established, and the change rate prediction model is trained using multiple samples.

[0016] Substitute the meteorological, soil, and crop data from the time period preceding the time period to be predicted into the rate of change prediction model to obtain the predicted value of the rate of change of crop data in the time period following the time period to be predicted.

[0017] Based on the predicted rate of change in crop data, the estimated value of the change in crop data is calculated, and the estimated value is marked as the over-predictive factor.

[0018] Preferably, the method for calculating the superpredictive factor includes:

[0019] Mark the time period in historical data that corresponds to the time period to be predicted as the specified time period, calculate the rate of change of crop data between the specified time period and the next time period in historical data, and mark it as the first rate of change;

[0020] Calculate the crop data change rate between the two time periods before a specified time period in historical data and the previous time period, and mark it as the second change rate. Calculate the crop data change rate between the two time periods before a specified time period in real-time data and the previous time period, and mark it as the third change rate.

[0021] Using the first rate of change as the output, and meteorological data, soil data, and crop data from the previous time period as the input, a rate of change prediction model is established, and the rate of change prediction model is trained using multiple samples.

[0022] Substitute the meteorological, soil, and crop data from the time period preceding the time period to be predicted into the rate of change prediction model to obtain the predicted value of the rate of change of crop data in the time period following the time period to be predicted.

[0023] Calculate the central tendency measure of the second rate of change for multiple samples in historical data, and calculate the ratio of the third rate of change to the central tendency measure;

[0024] Based on the predicted rate of change in crop data, the estimated value of the change in crop data is calculated and multiplied by the ratio to obtain the over-predictive factor.

[0025] Preferably, the method for calculating the superpredictive factor includes:

[0026] Mark the time period in historical data that corresponds to the time period to be predicted as the specified time period, calculate the rate of change of crop data between the specified time period and the next time period in historical data, and mark it as the first rate of change;

[0027] Calculate the rate of change of crop data between every two time periods before a specified time period in historical data, and label it as the second rate of change. Calculate the rate of change of crop data between every two time periods before a predicted time period in real-time data, and label it as the third rate of change.

[0028] Using the first rate of change as the output, and meteorological data, soil data, and crop data from the previous time period as the input, a rate of change prediction model is established, and the rate of change prediction model is trained using multiple samples.

[0029] Substitute the meteorological, soil, and crop data from the period preceding the predicted period in the real-time data into the rate of change prediction model to obtain the predicted value of the rate of change of crop data between the predicted period and the subsequent period.

[0030] Calculate the measure of central tendency for multiple second rates of change with the same time period in historical data, calculate the ratio between the third rate of change with the measure of central tendency for the same time period, and then calculate the mean of multiple ratios.

[0031] Based on the predicted rate of change in crop data, the estimated value of the change in crop data is calculated, and then multiplied by the mean of multiple ratios to obtain the over-predictive factor.

[0032] Preferably, the disaster prediction model establishment method includes:

[0033] The optimization ratio is calculated using the superpredictive factor, specifically as follows:

[0034]

[0035] in Indicates the optimization ratio. Indicates the overpredictor factor. Represents the absolute value of the overpredictor factor;

[0036] By combining the crop prediction model with the optimization ratio, a disaster prediction model is obtained, specifically:

[0037]

[0038] in This represents the disaster prediction result. A value between 0 and 1 indicates the probability of a disaster. A value greater than 1 indicates that crops have reached their historical best condition, while a value less than 0 indicates that agricultural disasters have reached their historical worst condition and crops are in their historical worst condition. This represents the output of the crop prediction model. This indicates the optimization ratio.

[0039] Preferably, the establishment of the crop prediction model specifically includes:

[0040] Based on the crop growth period, each time period is marked as a target time period, starting from the second time period. A time coefficient is calculated for each target time period, and a crop prediction model is built based on the time coefficients and historical data. Specifically:

[0041]

[0042] in This represents the time coefficient for the i-th target time period. The time period to be predicted is the [number]th [year]. A target time period, This represents the predicted value of crop data. This represents the total number of target time periods, which is 1 less than the total number of time periods divided into crop production periods. This indicates the total number of data types in meteorological, soil, and crop data. This represents the feature value of the j-th data type among the meteorological, soil, and crop data from the preceding time period, with the prediction time period in mind. This represents the regression coefficient for the j-th data type within the i-th target time period.

[0043] Preferably, the preset level specifically includes:

[0044] Crop data include the normalized vegetation index;

[0045] Using time periods as the calculation unit, the normalized vegetation index in multiple samples is subjected to minimum-maximum normalization so that its value ranges between 0 and 1, and is marked as the integrity value.

[0046] The completeness values ​​of samples in each time period are divided into several disaster levels according to a certain distribution ratio, and corresponding human intervention decisions are made for each level.

[0047] Compared with the prior art, the beneficial effects of the present invention are:

[0048] By calculating the super-predictive factor for the predicted time period using historical data, and using the super-predictive factor to represent the dynamic situation of subsequent disasters during the predicted time period, the self-recovery effect of the agricultural ecological environment is quantified. Combined with the crop prediction model constructed from historical data, the dynamic situation of agricultural disasters during the predicted time period can be analyzed more accurately, facilitating the formulation of more accurate and detailed intervention decisions, and further improving the accuracy and effectiveness of emergency decision-making.

[0049] Meanwhile, when establishing crop prediction models, different time coefficients are set according to the crop's growth period. This allows for the prediction of crop status for each time period based on the different data of the crop itself at different time periods, which can further improve the accuracy of crop status prediction. Attached Figure Description

[0050] Figure 1 This is a flowchart illustrating the agricultural disaster dynamic prediction and emergency decision-making method of the present invention;

[0051] Figure 2 This is a flowchart illustrating the super-predictive factor calculation method in Embodiment 1 of the present invention;

[0052] Figure 3 This is a flowchart illustrating the super-predictive factor calculation method in Embodiment 2 of the present invention;

[0053] Figure 4 This is a flowchart illustrating the disaster prediction model establishment method in this invention;

[0054] Figure 5 This is a schematic diagram of the grade prediction process in this invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] In this application, for ease of understanding, the method steps used do not necessarily need to be executed in the order of steps in this embodiment during actual operation. In other embodiments, these steps may be performed simultaneously or in a different order.

[0057] Example 1:

[0058] The agricultural ecological environment has a certain self-recovery function. For general agricultural disasters, it may automatically recover over time without human intervention or without the positive effect of human intervention. Combining the self-recovery effect of the agricultural ecological environment when predicting agricultural disasters can improve the accuracy of disaster dynamic prediction and the effectiveness and accuracy of emergency decision-making.

[0059] like Figure 1 , Figure 2 , Figure 4 and Figure 5 As shown, the present invention provides a technical solution: a method for dynamic prediction and emergency decision-making of agricultural disasters based on multi-source data, comprising:

[0060] Multi-source data collection: Collect historical and real-time data, both of which include meteorological, soil, and crop data for the crop growth period and several periods prior to the growth period.

[0061] It should be noted that meteorological data can be obtained from weather stations or through on-site measurements, soil data can be obtained through sampling and testing, and crop data can be obtained through remote sensing.

[0062] Level Preset: Crop data in historical and real-time data are grouped by time period and subjected to minimum-maximum normalization. The disaster level is then divided according to the data ratio to represent different degrees of disaster.

[0063] Establish a crop prediction model: In historical data, divide the crop growth period into time periods. Starting from the second time period, select any time period as the target time period. Use the crop data of the target time period as the output. Take the meteorological data, soil data, and crop data of the time period before the target time period, as well as the time period number of the target time period, as the input. Select different target time periods to combine multiple training samples and establish a crop prediction model.

[0064] Calculate the superpredictor factor: The superpredictor factor is calculated by using the superpredictor factor calculation method to calculate the superpredictor factor of historical data after the target time period relative to the historical data of the target time period. It represents the subsequent dynamic situation of crop data in the target time period and is used to reflect the self-recovery effect of the agricultural ecological environment.

[0065] Establishing a disaster prediction model: By combining crop prediction models and super-predictive factors through disaster prediction model establishment methods, a disaster prediction model can be obtained;

[0066] Decision-making: Input real-time data into the disaster prediction model, output the prediction results, and combine them with the disaster level to formulate a pre-set decision-making plan.

[0067] like Figure 5 As shown, the preset levels specifically include:

[0068] Crop data include the normalized vegetation index;

[0069] Using time periods as the calculation unit, the normalized vegetation index in multiple samples is subjected to minimum-maximum normalization so that its value ranges between 0 and 1, and is marked as the integrity value.

[0070] The completeness values ​​of samples in each time period are divided into several disaster levels according to a certain distribution ratio, and corresponding human intervention decisions are made for each level.

[0071] It should be noted that, for ease of understanding, the following simulated data is used:

[0072] The Normalized Difference Vegetation Index (NDVI) was selected as a characteristic item of the crop parameter. The calculation method and acquisition method of the NDVI are existing technologies and will not be described in detail here. The range of the NDVI is between -1 and -1.

[0073] After the normalized vegetation index is min-max normalized, its value range is controlled between 0 and 1. The closer it is to 0, the worse the physiological state of the crop is, and the closer it is to 1, the better the physiological state of the crop is.

[0074] Taking corn (whose growing season is from May to September, and the time period is divided by month; in actual operation, the time period can be divided into more detailed periods according to the density of the collected data, such as by week. The more detailed the time period division, the more accurate the prediction results. The specific division method is not limited) as an example, assuming that the crop data after minimum-maximum normalization are shown in Tables 1 and 2 below (now we need to predict the disaster level in July in the real-time data):

[0075] Table 1: Historical Crop Data (Previous Years' Data)

[0076] May June July August September Sample 1 0.8 0.8 0.7 0.7 0.9 Sample 2 0.7 0.6 0.6 0.6 0.8 Sample 3 0.8 0.7 0.7 0.6 0.8 … … … … … …

[0077] Table 2: Real-time crop data (current year data)

[0078] May June July August September Data from that year 0.6 0.7 To be predicted

[0079] Then, taking each time period as the calculation unit, and taking July as an example, it is assumed that there is 1 sample of crop data between 0 and 0.3 (inclusive) in the historical data, which is classified as a severe disaster level; there are 3 samples between 0.3 and 0.5 (inclusive), which are classified as a moderate disaster level; there are 6 samples between 0.5 and 0.7 (inclusive), which are classified as a slight disaster level; and there are 3 samples between 0.7 and 1 (inclusive), which are classified as no disaster level.

[0080] Furthermore, technical personnel are responsible for formulating emergency decisions for each disaster level (such as increasing irrigation intensity, adjusting irrigation time, fertilizing and spraying pesticides, etc.). (The specific division ratio, the number of levels, and the formulation of emergency decisions can be determined by technical personnel based on experience and crop planting guidance materials, without any specific restrictions.)

[0081] Establishing crop prediction models specifically includes:

[0082] Based on the time periods of the crop growth cycle, each time period is marked as a target time period starting from the second time period. The time coefficient of each target time period is calculated, and a crop prediction model is established based on the time coefficient and historical data.

[0083] First, the data for each time period is grouped. Then, a multiple linear regression model is built for each time period separately. The data types of meteorological, soil, and crop data from the time period preceding the target time period are combined into a column vector, and the time coefficients of each time period are combined into a row vector. Then, the corresponding regression coefficients are assigned to the corresponding time periods through vector-matrix multiplication. Finally, coefficients that do not belong to the target time period are set to zero using the time coefficients. This yields the crop prediction model, specifically:

[0084]

[0085] in This represents the time coefficient for the i-th target time period. The time period to be predicted is the [number]th [year]. A target time period, This represents the predicted value of crop data. This represents the total number of target time periods, which is 1 less than the total number of time periods divided into crop production periods. This indicates the total number of data types in meteorological, soil, and crop data. This represents the feature value of the j-th data type among the meteorological, soil, and crop data from the preceding time period, with the prediction time period in mind. This represents the regression coefficient for the j-th data type within the i-th target time period.

[0086] The above formula is composed of multiple multiple linear regression models. During training, the least squares method can be used, with the mean squared error as the loss function, to obtain... The parameter matrix.

[0087] It should be noted that, for ease of understanding, simulated data are used as shown in Tables 3 and 4 below (meteorological and soil data are represented using only a single data type for ease of calculation and method demonstration; in actual operations, multiple data types can be used to improve forecast accuracy):

[0088] Table 3: Historical meteorological data (temperature, in °C, the same below)

[0089] May June July August September Sample 1 22 26 30 32 26 Sample 2 20 29 34 34 28 Sample 3 22 28 31 32 26 … … … … … …

[0090] Table 4: Historical Soil Data (Moisture Content, % as used below)

[0091] May June July August September Sample 1 24 22 20 21 23 Sample 2 23 18 16 17 18 Sample 3 24 21 21 20 21 … … … … … …

[0092] The crop data uses the simulation data from Table 1 above. The model is then trained (the specific training process is existing technology and will not be detailed here). The parameters are shown in Table 5 below (where i=1 represents the first target time period, i.e., June):

[0093] Table 5: Parameter table

[0094] 1 2 3 4 1 (Temperature) 0.02 0.01 0.01 0.02 2 (Moisture content) -0.01 -0.01 -0.01 -0.01 3 (NDVI) 0.89 0.86 0.71 0.86

[0095] Assume the real-time data is as shown in Table 6 below:

[0096] Table 6: Real-time data (meteorological data and soil data)

[0097] May June July August September Temperature 19 21 Moisture content 28 36

[0098] We now need to forecast crop data for July. =2 (June is the first target period, and July is the second target period, therefore...) =2) Substituting into the crop prediction model, we get: ≈0.4, therefore we can conclude that the crop data for July is approximately 0.4.

[0099] like Figure 2 As shown, the methods for calculating the superpredictor factor include:

[0100] Mark the time period in historical data that corresponds to the time period to be predicted as the specified time period, and calculate the rate of change of crop data between the specified time period and the next time period in historical data;

[0101] Using crop data change rate as output, and meteorological data, soil data, and crop data from the previous time period as input, a change rate prediction model is established, and the change rate prediction model is trained using multiple samples.

[0102] Substitute the meteorological, soil, and crop data from the time period preceding the time period to be predicted into the rate of change prediction model to obtain the predicted value of the rate of change of crop data in the time period following the time period to be predicted.

[0103] Based on the predicted rate of change in crop data, the estimated value of the change in crop data is calculated, and the estimated value is marked as the over-predictive factor.

[0104] It should be noted that, for ease of understanding, the following simulated data is used:

[0105] Taking July as an example, select July as the specified time period and calculate the rate of change of crop data between July and June:

[0106] Sample 1: (0.7 - 0.7) ÷ 0.7 = 0;

[0107] Sample 2: (0.6 - 0.6) ÷ 0.6 = 0;

[0108] Sample 3: (0.6-0.7)÷0.7≈-0.14;

[0109] ...

[0110] Using the crop data change rate as output, and taking meteorological, soil, and crop data from the previous time period as input, a change rate prediction model is established, specifically as follows:

[0111]

[0112] in This is the predicted rate of change for crop data. , , These represent meteorological data (temperature), soil data (humidity), and crop data (NDVI) for the period preceding the specified time period, respectively. , , These represent the regression coefficients of meteorological data (temperature), soil data (humidity), and crop data (NDVI) for the period preceding the specified time period.

[0113] Training with simulated data yields... ≈-0.01, ≈0.01, The predicted value of the crop data change rate for the next time period after the predicted time period is approximately 0.27, obtained through the trained model.

[0114] Based on the predicted rate of change in crop data, the estimated value of the change in crop data is calculated:

[0115] 0.27 × 0.4 = 0.108;

[0116] Therefore, the overpredictive factor is 0.108.

[0117] like Figure 4 As shown, the methods for establishing disaster prediction models include:

[0118] The optimization ratio is calculated using the superpredictive factor, specifically as follows:

[0119]

[0120] in Indicates the optimization ratio. Indicates the overpredictor factor. Represents the absolute value of the overpredictor factor;

[0121] By combining the crop prediction model with the optimization ratio, a disaster prediction model is obtained, specifically:

[0122]

[0123] in This represents the disaster prediction result. A value between 0 and 1 indicates the probability of a disaster. A value greater than 1 indicates that crops have reached their historical best condition, while a value less than 0 indicates that agricultural disasters have reached their historical worst condition and crops are in their historical worst condition. This represents the output of the crop prediction model. Indicates the optimization ratio

[0124] It should be noted that, for ease of understanding, the following simulated data is used:

[0125] Taking July as an example:

[0126] Using the super-predictive factor and crop prediction model obtained from the previous calculations, further calculations can be performed to obtain the following results:

[0127] Optimization ratio ≈0.121;

[0128] Based on the aforementioned calculated crop data forecast value of 0.4 for July, the disaster prediction result is calculated. =0.4 + 0.121 = 0.521;

[0129] Based on the disaster level classification for July, 0.521 falls between 0.5 and 0.7, which is considered a minor disaster. Technicians can then make emergency decisions based on the level of minor disaster, thereby improving the accuracy and effectiveness of emergency decisions.

[0130] Example 2:

[0131] The superpredictor factor represents the dynamic changes in crop data over a period of time after the prediction period, and is used to reflect the self-recovery effect of the agricultural ecological environment. In Example 1, the superpredictor factor is calculated by considering the changes in multi-source data. However, there are many influencing factors in the agricultural ecological environment, and situations outside of multi-source data may lead to large calculation errors in the superpredictor factor. Therefore, this example provides another superpredictor factor calculation method to improve the accuracy of the superpredictor factor.

[0132] like Figure 3 As shown, the methods for calculating the superpredictor factor include:

[0133] Mark the time period in historical data that corresponds to the time period to be predicted as the specified time period, calculate the rate of change of crop data between the specified time period and the next time period in historical data, and mark it as the first rate of change;

[0134] Calculate the crop data change rate between the two time periods before a specified time period in historical data and the previous time period, and mark it as the second change rate. Calculate the crop data change rate between the two time periods before a specified time period in real-time data and the previous time period, and mark it as the third change rate.

[0135] Using the first rate of change as the output, and meteorological data, soil data, and crop data from the previous time period as the input, a rate of change prediction model is established, and the rate of change prediction model is trained using multiple samples.

[0136] Substitute the meteorological, soil, and crop data from the time period preceding the time period to be predicted into the rate of change prediction model to obtain the predicted value of the rate of change of crop data in the time period following the time period to be predicted.

[0137] Calculate the central tendency measure of the second rate of change for multiple samples in historical data, and calculate the ratio of the third rate of change to the central tendency measure;

[0138] Based on the predicted rate of change in crop data, the estimated value of the change in crop data is calculated and multiplied by the ratio to obtain the over-predictive factor.

[0139] It should be noted that, for ease of understanding, the following simulated data is used:

[0140] Taking July as an example, select July as the specified time period and calculate the rate of change (first rate of change) of crop data between July and June:

[0141] Sample 1: (0.7 - 0.7) ÷ 0.7 = 0;

[0142] Sample 2: (0.6 - 0.6) ÷ 0.6 = 0;

[0143] Sample 3: (0.6-0.7)÷0.7≈-0.14;

[0144] ...

[0145] Using the crop data change rate as output, and taking meteorological, soil, and crop data from the previous time period as input, a change rate prediction model is established, specifically as follows:

[0146]

[0147] in This is the predicted rate of change for crop data. , , These represent meteorological data (temperature), soil data (humidity), and crop data (NDVI) for the period preceding the specified time period, respectively. , , These represent the regression coefficients of meteorological data (temperature), soil data (humidity), and crop data (NDVI) for the period preceding the specified time period.

[0148] Training with simulated data yields... ≈-0.01, ≈0.01, The predicted value of the crop data change rate for the next time period after the predicted time period is approximately 0.27, obtained through the trained model.

[0149] Based on the predicted rate of change in crop data, the estimated value of the change in crop data is calculated as: 0.27 × 0.4 = 0.108.

[0150] Calculate the second rate of change:

[0151] Sample 1: (0.8 - 0.8) ÷ 0.8 = 0;

[0152] Sample 2: (0.6-0.7)÷0.7≈0.143;

[0153] Sample 3: (0.7-0.8)÷0.8=0.125;

[0154] ...

[0155] Calculate the third rate of change: (0.7-0.6) ÷ 0.6 ≈ 0.167.

[0156] The average of the second rate of change (the average is used here as a measure of central tendency, but the median or mode can also be used, and there is no specific restriction) is 0.09, and the ratio of the third rate of change to this average is 1.856.

[0157] The overpredictor factor is the product of the estimated value of the crop data change and the ratio, and the calculated overpredictor factor is approximately 0.2.

[0158] This calculation method compares the magnitude of changes in historical data with the magnitude of changes in real-time data, comprehensively quantifies the impact of factors other than multi-source data on the agricultural ecological environment, and ensures that the calculation of super-predictive factors can cover the impact of more unknown factors, thereby improving the accuracy of super-predictive factors.

[0159] Example 3

[0160] In Example 2, the variation range of historical data is compared with the variation range of real-time data. However, this only considers the comparison of the situation in the period before the prediction period, which has a poor overall representative effect on the agricultural ecological environment. Based on this, this example provides another method for calculating the super-predictive factor based on Example 2, which further improves the accuracy of the super-predictive factor.

[0161] Methods for calculating the superpredictor factor include:

[0162] Mark the time period in historical data that corresponds to the time period to be predicted as the specified time period, calculate the rate of change of crop data between the specified time period and the next time period in historical data, and mark it as the first rate of change;

[0163] Calculate the rate of change of crop data between every two time periods before a specified time period in historical data, and label it as the second rate of change. Calculate the rate of change of crop data between every two time periods before a predicted time period in real-time data, and label it as the third rate of change.

[0164] Using the first rate of change as the output, and meteorological data, soil data, and crop data from the previous time period as the input, a rate of change prediction model is established, and the rate of change prediction model is trained using multiple samples.

[0165] Substitute the meteorological, soil, and crop data from the period preceding the predicted period in the real-time data into the rate of change prediction model to obtain the predicted value of the rate of change of crop data between the predicted period and the subsequent period.

[0166] Calculate the measure of central tendency for multiple second rates of change with the same time period in historical data, calculate the ratio between the third rate of change with the measure of central tendency for the same time period, and then calculate the mean of multiple ratios.

[0167] Based on the predicted rate of change in crop data, the estimated value of the change in crop data is calculated, and then multiplied by the mean of multiple ratios to obtain the over-predictive factor.

[0168] It should be noted that, for ease of understanding, the following simulated data is used:

[0169] Taking July as an example, the estimated value of crop data change in Example 2 is 0.108 (the specific calculation process is the same as described above, and will not be repeated here).

[0170] Calculate the second rate of change:

[0171] Sample 1: (0.8-0.8)÷0.8=0; (0.7-0.8)÷0.8=-0.125; (0.7-0.7)÷0.7=0; (0.9-0.7)÷0.7≈0.286;

[0172] Sample 2: (0.6-0.7)÷0.7≈0.143; (0.6-0.6)÷0.6=0; (0.6-0.6)÷0.6=0; (0.8-0.6)÷0.6≈0.333;

[0173] Sample 3: (0.7-0.8)÷0.8=0.125; (0.7-0.7)÷0.7=0; (0.6-0.7)÷0.7=-0.143; (0.8-0.6)÷0.6≈0.333;

[0174] ...

[0175] Calculate the third rate of change: (0.7-0.6) ÷ 0.6 = 0.167.

[0176] The average of the second rate of change (the average is used here as a measure of central tendency; the median or mode can also be used, and there is no specific restriction) is 0.08. The ratio of the third rate of change to this average is approximately 2.09.

[0177] The overpredictor factor is the product of the estimated value of the crop data change and the ratio, and the calculated overpredictor factor is approximately 0.226.

[0178] By substituting the crop data change rates of multiple time periods for each sample into the calculation of the super-predictive factor, and fully considering the differences between each sample, this embodiment uses the entire sample as the data reference when comprehensively quantifying the impact of factors other than multi-source data on the agricultural ecological environment (Example 2 uses one time period as the reference) compared to Example 2. This expands the calculation base, enabling the super-predictive factor to more accurately quantify the self-recovery effect of the agricultural ecological environment, thereby improving the accuracy of subsequent disaster prediction.

[0179] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended embodiments and their equivalents.

Claims

1. A method for dynamic prediction and emergency decision-making of agricultural disasters based on multi-source data, characterized by: include: Multi-source data collection: Collect historical and real-time data, both of which include meteorological, soil, and crop data for the crop growth period and several periods prior to the growth period. Level Preset: Crop data in historical and real-time data are grouped by time period and subjected to minimum-maximum normalization. The disaster level is then divided according to the data ratio to represent different degrees of disaster. Establish a crop prediction model: In historical data, divide the crop growth period into time periods. Starting from the second time period, select any time period as the target time period. Use the crop data of the target time period as the output. Take the meteorological data, soil data, and crop data of the time period before the target time period, as well as the time period number of the target time period, as the input. Select different target time periods to combine multiple training samples and establish a crop prediction model. Calculate the superpredictor factor: The superpredictor factor is calculated by the superpredictor factor calculation method to determine the superpredictor factor of historical data after the target time period relative to the historical data of the target time period. It represents the subsequent dynamic situation of crop data in the target time period and is used to reflect the self-recovery effect of the agricultural ecological environment. Establishing a disaster prediction model: By combining crop prediction models and super-predictive factors through disaster prediction model establishment methods, a disaster prediction model can be obtained; Decision-making: Input real-time data into the disaster prediction model, output the prediction results, and combine them with the disaster level to formulate a pre-set decision-making plan; The method for establishing the disaster prediction model includes: The optimization ratio is calculated using the superpredictive factor, specifically as follows: in Indicates the optimization ratio. Indicates the overpredictor factor. Represents the absolute value of the overpredictor factor; By combining the crop prediction model with the optimization ratio, a disaster prediction model is obtained, specifically: in This represents the disaster prediction result. A value between 0 and 1 indicates the probability of a disaster. A value greater than 1 indicates that crops have reached their historical best condition, while a value less than 0 indicates that agricultural disasters have reached their historical worst condition and crops are in their historical worst condition. This represents the output of the crop prediction model. This indicates the optimization ratio.

2. The method for dynamic prediction and emergency decision-making of agricultural disasters based on multi-source data according to claim 1, characterized in that: The method for calculating the superpredictive factor includes: Mark the time period in historical data that corresponds to the time period to be predicted as the specified time period, and calculate the rate of change of crop data between the specified time period and the next time period in historical data; Using crop data change rate as output, and meteorological data, soil data, and crop data from the previous time period as input, a change rate prediction model is established, and the change rate prediction model is trained using multiple samples. Substitute the meteorological, soil, and crop data from the time period preceding the time period to be predicted into the rate of change prediction model to obtain the predicted value of the rate of change of crop data in the time period following the time period to be predicted. Based on the predicted rate of change in crop data, the estimated value of the change in crop data is calculated, and the estimated value is marked as the over-predictive factor.

3. The method for dynamic prediction and emergency decision-making of agricultural disasters based on multi-source data according to claim 1, characterized in that: The method for calculating the superpredictive factor includes: Mark the time period in historical data that corresponds to the time period to be predicted as the specified time period, calculate the rate of change of crop data between the specified time period and the next time period in historical data, and mark it as the first rate of change; Calculate the crop data change rate between the two time periods before a specified time period in historical data and the previous time period, and mark it as the second change rate. Calculate the crop data change rate between the two time periods before a specified time period in real-time data and the previous time period, and mark it as the third change rate. Using the first rate of change as the output, and meteorological data, soil data, and crop data from the previous time period as the input, a rate of change prediction model is established, and the rate of change prediction model is trained using multiple samples. Substitute the meteorological, soil, and crop data from the time period preceding the time period to be predicted into the rate of change prediction model to obtain the predicted value of the rate of change of crop data in the time period following the time period to be predicted. Calculate the central tendency measure of the second rate of change for multiple samples in historical data, and calculate the ratio of the third rate of change to the central tendency measure; Based on the predicted rate of change in crop data, the estimated value of the change in crop data is calculated and multiplied by the ratio to obtain the over-predictive factor.

4. The method for dynamic prediction and emergency decision-making of agricultural disasters based on multi-source data according to claim 1, characterized in that: The method for calculating the superpredictive factor includes: Mark the time period in historical data that corresponds to the time period to be predicted as the specified time period, calculate the rate of change of crop data between the specified time period and the next time period in historical data, and mark it as the first rate of change; Calculate the rate of change of crop data between every two time periods before a specified time period in historical data, and label it as the second rate of change. Calculate the rate of change of crop data between every two time periods before a predicted time period in real-time data, and label it as the third rate of change. Using the first rate of change as the output, and meteorological data, soil data, and crop data from the previous time period as the input, a rate of change prediction model is established, and the rate of change prediction model is trained using multiple samples. Substitute the meteorological, soil, and crop data from the period preceding the predicted period in the real-time data into the rate of change prediction model to obtain the predicted value of the rate of change of crop data between the predicted period and the subsequent period. Calculate the measure of central tendency for multiple second rates of change with the same time period in historical data, calculate the ratio between the third rate of change with the measure of central tendency for the same time period, and then calculate the mean of multiple ratios. Based on the predicted rate of change in crop data, the estimated value of the change in crop data is calculated, and then multiplied by the mean of multiple ratios to obtain the over-predictive factor.

5. The method for dynamic prediction and emergency decision-making of agricultural disasters based on multi-source data according to claim 1, characterized in that: The establishment of the crop prediction model specifically includes: Based on the crop growth period, each time period is marked as a target time period, starting from the second time period. A time coefficient is calculated for each target time period, and a crop prediction model is built based on the time coefficients and historical data. Specifically: in This represents the time coefficient for the i-th target time period. The time period to be predicted is the [number]th [year]. A target time period, This represents the predicted value of crop data. This represents the total number of target time periods, which is 1 less than the total number of time periods divided into crop production periods. This indicates the total number of data types in meteorological, soil, and crop data. This represents the feature value of the j-th data type among the meteorological, soil, and crop data from the preceding time period, with the prediction time period in mind. This represents the regression coefficient for the j-th data type within the i-th target time period.

6. The method for dynamic prediction and emergency decision-making of agricultural disasters based on multi-source data according to claim 1, characterized in that: The preset level specifically includes: Crop data include the normalized vegetation index; Using time periods as the calculation unit, the normalized vegetation index in multiple samples is subjected to minimum-maximum normalization so that its value ranges between 0 and 1, and is marked as the integrity value. The completeness values ​​of samples in each time period are divided into several disaster levels according to a certain distribution ratio, and corresponding human intervention decisions are made for each level.

7. An agricultural disaster dynamic prediction and emergency decision-making system based on multi-source data, characterized in that: include: Data collection module: used to collect historical and real-time data. Both historical and real-time data include meteorological, soil, and crop data for the crop growth period and several periods before the growth period. Preprocessing module: This module is used to group crop data from historical and real-time data according to time periods and perform min-max normalization, and to classify disaster levels according to data proportions to represent different degrees of disaster. Central Processing Module: Used to build crop prediction models: In historical data, divided according to the crop growth period, starting from the second period, any period is selected as the target period. Crop data of the target period is used as output, and meteorological, soil, and crop data of the period preceding the target period, as well as the period number of the target period, are used as input. Different target periods are selected to combine multiple training samples and build crop prediction models; Calculate superpredictor factors: The superpredictor factor of historical data after the target period is calculated relative to the historical data of the target period, representing the subsequent dynamics of crop data in the target period, and is used to reflect the self-recovery effect of the agricultural ecological environment; Build disaster prediction models: The crop prediction model and the superpredictor factors are combined using disaster prediction model building methods to obtain disaster prediction models; Output module: Inputs real-time data into the disaster prediction model, outputs the prediction results, and combines them with the disaster level to formulate a preset decision-making plan; The method for establishing the disaster prediction model includes: The optimization ratio is calculated using the superpredictive factor, specifically as follows: in Indicates the optimization ratio. Indicates the overpredictor factor. Represents the absolute value of the overpredictor factor; By combining the crop prediction model with the optimization ratio, a disaster prediction model is obtained, specifically: in This represents the disaster prediction result. A value between 0 and 1 indicates the probability of a disaster. A value greater than 1 indicates that crops have reached their historical best condition, while a value less than 0 indicates that agricultural disasters have reached their historical worst condition and crops are in their historical worst condition. This represents the output of the crop prediction model. This indicates the optimization ratio.

Images