Crop growth state simulation method and system based on meteorological grid data

Abstract

According to the crop growth state simulation method and system based on the meteorological grid data provided by the embodiment of the invention, the growth state of crops can be accurately predicted by deeply analyzing and simulating the meteorological environment element variables, and then an accurate crop yield prediction result is obtained. The prediction is not only helpful for planning planting in advance, but also can provide a scientific basis for agricultural operators, and helps to optimize crop planting structures and resource allocation. The scheme emphasizes the importance of mining cross characteristics shared among multiple state simulation original data, can deeply understand how various factors affect crop growth, finds out optimal growth conditions, and adjusts meteorological environment parameters accordingly. The meteorological environment parameters are adjusted according to the crop yield prediction result, so that the crop yield can be improved, the resource utilization efficiency can be improved, and the production cost is reduced.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a method and system for simulating crop growth status based on meteorological grid data. Background Art

[0002] With the development of science and technology, data-driven agricultural forecasting models have received increasing attention. Such models collect and analyze large amounts of data to try to identify various factors that affect crop growth and yield, in order to provide accurate forecasts for farmers and agricultural operators.

[0003] However, accurately predicting the growth status and yield of crops is still a challenging task. This is mainly because the growth status and yield of crops are affected by a variety of meteorological environmental factors, including temperature, humidity, wind speed, rainfall, etc. The relationship between these factors is complex, and in-depth analysis and simulation are often required to obtain accurate prediction results.

[0004] Traditional forecasting methods usually fail to consider all meteorological and environmental factors that affect crop growth and can only make forecasts based on a subset of these factors, so the accuracy of their forecast results is limited. In addition, these methods also lack effective tools to process and analyze large amounts of complex meteorological data, which further limits their forecasting capabilities. Summary of the invention

[0005] In order to improve the above problems, the present application provides a method and system for simulating the growth status of crops based on meteorological grid data.

[0006] In a first aspect, a method for simulating crop growth status based on meteorological grid data is provided, which is applied to a data simulation system, and the method comprises:

[0007] Obtain meteorological environment element variables corresponding to the meteorological grid monitoring information; wherein the meteorological environment element variables are used to simulate the growth status of crops under a plurality of crop growth status labels, and obtain the crop growth status simulation results corresponding to the plurality of crop growth status labels in the target farming area;

[0008] When simulating the growth state of crops by using the growth state simulation algorithms corresponding to the plurality of crop growth state labels, determining the state simulation original data of the growth state simulation algorithms corresponding to the plurality of crop growth state labels; wherein the state simulation original data is a description vector to be input into the growth state simulation algorithm for state simulation;

[0009] Extracting cross-features shared between state feature mapping relationship networks corresponding to a plurality of state simulation original data; wherein the state feature mapping relationship network is obtained by performing data feature mining on the state simulation original data;

[0010] The plurality of state simulation original data are updated with the cross feature to obtain a plurality of state simulation update data; wherein the plurality of state simulation original data and the plurality of state simulation update data are corresponding to each other;

[0011] Based on the several state simulation update data, the crop growth state simulation results corresponding to the target farming area under the several crop growth state labels are generated; wherein, the several crop growth state simulation results are used to combine to obtain the crop yield prediction result of the target farming area, and the crop yield prediction result is used as a guide for adjusting the artificial meteorological environment parameters.

[0012] In some exemplary technical solutions, determining the state simulation raw data of the growth state simulation algorithm corresponding to the plurality of crop growth state tags respectively includes:

[0013] Acquiring crop growth monitoring information, wherein the crop growth monitoring information is information obtained by monitoring crops in the target farming area, and the crop growth monitoring information is used to generate a simulation result reflecting the crop growth status of the target farming area;

[0014] Taking the crop growth monitoring information as a simulation reference, the state simulation original data of the growth state simulation algorithm corresponding to the several crop growth state labels are determined, and the simulation reference is used to determine the crop growth trend characteristics when simulating the meteorological environment element variables.

[0015] In some exemplary technical solutions, the obtaining of crop growth monitoring information includes:

[0016] At least one data set obtained by monitoring the target farming area is obtained as the crop growth monitoring information, wherein the data set is a result of monitoring the target farming area based on a set crop growth status label.

[0017] In some exemplary technical solutions, extracting cross-features shared between state feature mapping relationship networks corresponding to a plurality of state simulation raw data respectively includes:

[0018] Performing convolution operations on the plurality of state simulation original data respectively to obtain the state feature mapping relationship networks respectively corresponding to the plurality of state simulation original data;

[0019] Local downsampling is performed on a plurality of state feature mapping relationship networks to obtain a state involvement vector, wherein the state involvement vector is used to reflect the cross-features shared between the plurality of state feature mapping relationship networks.

[0020] In some exemplary technical solutions, the convolution operation is performed on the plurality of state simulation raw data respectively to obtain the state feature mapping relationship network corresponding to the plurality of state simulation raw data respectively, including:

[0021] Performing convolution operations on the plurality of state simulation raw data respectively to obtain state simulation convolution vectors corresponding to the plurality of crop growth state labels respectively;

[0022] Acquire sensor monitoring element vectors corresponding to the plurality of crop growth status labels, respectively, wherein the sensor monitoring element vectors are linear vectors obtained based on sensor variables corresponding to the crop growth status labels, and the sensor monitoring element vectors are used to reflect periodic data of the corresponding crop growth status labels, wherein the plurality of sensor monitoring element vectors correspond to the plurality of state simulation convolution vectors;

[0023] According to the correspondence, the state simulation convolution vector and the sensor monitoring element vector under the same crop growth state label are vector integrated to obtain the state feature mapping relationship network corresponding to the several state simulation original data respectively.

[0024] In some exemplary technical solutions, the step of obtaining the sensor monitoring element vectors corresponding to the plurality of crop growth status labels respectively includes:

[0025] Acquire sensor variables corresponding to the plurality of crop growth status labels, respectively, wherein the sensor variables are used to reflect sensor distribution characteristics for generating corresponding crop growth status simulation results, and the sensor variables include sensor area labels and sensor category labels, wherein the sensor area labels are used to reflect the distribution of sensors in the sensor monitoring feature space relative to the target farming area, and the sensor category labels are used to characterize the monitoring category of the sensors in the sensor monitoring feature space relative to the target farming area;

[0026] Preprocessing a plurality of sensor variables respectively to obtain preprocessing results respectively corresponding to the plurality of sensor variables, wherein the preprocessing results are linear vectors obtained by preprocessing the sensor variables in a growth monitoring task link;

[0027] The knowledge extraction algorithm is set to extract knowledge elements from the preprocessing result to obtain the sensor monitoring element vectors corresponding to the plurality of crop growth status labels.

[0028] In some exemplary technical solutions, locally downsampling a plurality of state feature mapping relationship networks to obtain a state involvement vector includes:

[0029] Determine a set of meteorological environment impact features respectively represented by the plurality of state feature mapping relationship networks, wherein the set of meteorological environment impact features is used to reflect a set of a plurality of meteorological environment impact features in a growth monitoring task link when the state feature mapping relationship network is obtained;

[0030] Focusing a number of meteorological environment impact features at the same feature position in a number of meteorological environment impact feature sets to obtain a number of local focusing weights;

[0031] The plurality of local focusing weights are downsampled to obtain the state involvement vector.

[0032] In some exemplary technical solutions, the updating of the plurality of state simulation original data using the cross feature to obtain a plurality of state simulation update data includes:

[0033] Acquire a state involvement vector representing the cross feature;

[0034] According to the crop growth state labels corresponding to the plurality of state simulation original data and the state involvement vector, meteorological environment associated growth description vectors corresponding to the plurality of crop growth state labels are obtained, wherein the meteorological environment associated growth description vectors are used to reflect the contribution of the state involvement vector to the state simulation original data at the time series description level, wherein the plurality of meteorological environment associated growth description vectors and the plurality of state simulation original data have correspondence;

[0035] According to the correspondence, the plurality of state simulation update data are obtained by associating the growth description vector with the meteorological environment under the same crop growth state label and the state simulation original data.

[0036] In some exemplary technical solutions, the meteorological environment-associated growth description vectors corresponding to the plurality of crop growth state labels respectively corresponding to the plurality of state simulation raw data and the state involvement vectors are obtained, including:

[0037] Determine sensor monitoring knowledge graphs corresponding to the plurality of crop growth status labels, respectively, wherein the sensor monitoring knowledge graph is a knowledge graph generated based on the sensor used when determining the corresponding crop growth status label;

[0038] Using the sensor monitoring knowledge graph corresponding to the crop growth status label monitoring category as a template, performing knowledge migration on the state involvement vector to obtain knowledge migration vectors representing the respective corresponding knowledge to the plurality of crop growth status labels;

[0039] According to the plurality of crop growth status labels and the corresponding knowledge transfer vectors, the meteorological environment associated growth description vectors respectively corresponding to the plurality of crop growth status labels in the growth monitoring task link are obtained.

[0040] In some exemplary technical solutions, the meteorological environment associated growth description vectors corresponding to the plurality of crop growth status labels in the growth monitoring task link are obtained based on the plurality of crop growth status labels and the corresponding knowledge transfer vectors, including:

[0041] Obtaining the thermal values ​​of the crop growth states represented by the plurality of crop growth state labels respectively;

[0042] Under the same crop growth status label, the meteorological environment associated growth description vectors corresponding to the plurality of crop growth status labels in the growth monitoring task link are obtained according to the crop growth status thermal value and the knowledge transfer vector.

[0043] In some exemplary technical solutions, under the same crop growth status label, according to the crop growth status thermal value and the knowledge transfer vector, the meteorological environment associated growth description vectors corresponding to the plurality of crop growth status labels in the growth monitoring task link are obtained, including:

[0044] Determine a meteorological environment impact feature set represented by a state feature mapping relationship network corresponding to a plurality of crop growth state labels, wherein the meteorological environment impact feature set includes a plurality of meteorological environment impact features, and each meteorological environment impact feature corresponds to a crop growth state thermal value;

[0045] Taking the crop growth state thermal value corresponding to the meteorological environment impact feature as the meteorological environment impact feature value, deriving several meteorological environment impact features in the meteorological environment impact feature set to obtain a meteorological environment impact feature chain set having the same meteorological environment impact feature value;

[0046] According to the meteorological environment impact feature chain set and the knowledge transfer vector, the meteorological environment associated growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link are obtained.

[0047] In some exemplary technical solutions, the meteorological environment-related growth description vectors corresponding to the plurality of crop growth status labels in the growth monitoring task link are obtained based on the meteorological environment impact feature chain set and the knowledge transfer vector, including:

[0048] Extracting knowledge elements from the meteorological environment impact feature chain set using a deep learning branch to obtain meteorological environment impact feature knowledge;

[0049] Under the same crop growth status label, the meteorological environment impact characteristic knowledge and the knowledge transfer vector are aggregated to obtain the meteorological environment associated growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link.

[0050] In some exemplary technical solutions, based on the correspondence, the meteorological environment-associated growth description vector and the state simulation original data under the same crop growth state label are vector integrated to obtain the plurality of state simulation update data:

[0051] Mapping the meteorological environment-related growth description vectors corresponding to the plurality of crop growth status labels to the simulated feature relationship network to obtain growth environment-related mapping vectors corresponding to the plurality of crop growth status labels;

[0052] Under the same crop growth state label, the growth environment association mapping vector and the state simulation original data are vector-integrated to obtain the state simulation update data corresponding to a plurality of crop growth state labels.

[0053] In some exemplary technical solutions, each crop growth state label corresponds to x growth state simulation algorithms, where x is a positive integer;

[0054] The step of generating the crop growth state simulation results corresponding to the crop growth state labels in the target farming area according to the plurality of state simulation update data includes:

[0055] Performing crop growth state simulation on the state simulation update data of the yth growth state simulation algorithm corresponding to the plurality of crop growth state labels, to obtain the state simulation original data of the y+1th growth state simulation algorithm corresponding to the plurality of crop growth state labels, where y is a positive integer not greater than x;

[0056] In response to the xth growth state simulation algorithm, obtaining a state simulation reasoning vector output by the xth growth state simulation algorithm corresponding to the plurality of crop growth state labels respectively;

[0057] Based on a plurality of state simulation inference vectors, the crop growth state simulation results corresponding to the target farming area under the plurality of crop growth state labels are generated.

[0058] In some exemplary technical solutions, the generating of the crop growth state simulation results corresponding to the target farming area under the plurality of crop growth state labels respectively based on the plurality of state simulation inference vectors includes:

[0059] Performing a cyclic crop growth state simulation according to the state simulation reasoning vector under the plurality of crop growth state labels until a cyclic condition is met, thereby obtaining state simulation decision vectors corresponding to the plurality of crop growth state labels, wherein the state simulation decision vectors are used to reflect a linear vector obtained after simulating the growth state of the crop with respect to the meteorological environment element variables;

[0060] The plurality of state simulation decision vectors are respectively transmitted through simulation output branches to generate the crop growth state simulation results corresponding to the plurality of crop growth state labels in the target farming area.

[0061] In a second aspect, a data simulation system is provided, comprising a processor and a memory communicating with each other, wherein the processor is used to retrieve a computer program from the memory and implement the method described in the first aspect by running the computer program.

[0062] According to a third aspect, a computer-readable storage medium is provided, on which a computer program is stored, and the computer program implements the method described in the first aspect when running.

[0063] The crop growth state simulation method and system based on meteorological grid data provided in the embodiment of the present application can accurately predict the growth state of crops and obtain accurate crop yield prediction results by deeply analyzing and simulating meteorological environmental factor variables. This prediction not only helps to plan planting in advance, but also provides a scientific basis for agricultural operators to help optimize crop planting structure and resource allocation. This scheme emphasizes the importance of mining the cross-features shared between multiple state simulation raw data, which can more deeply understand how various factors affect crop growth, find the optimal growth conditions, and adjust meteorological environmental parameters accordingly. Adjusting meteorological environmental parameters according to crop yield prediction results can not only increase crop yields, but also improve resource utilization efficiency and reduce production costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0065] Figure 1 A flowchart of a method for simulating crop growth status based on meteorological grid data provided in an embodiment of the present application. DETAILED DESCRIPTION

[0066] In order to better understand the above technical scheme, the technical scheme of the present application is described in detail below through the accompanying drawings and specific embodiments. It should be understood that the embodiments of the present application and the specific features in the embodiments are detailed descriptions of the technical scheme of the present application, rather than limitations on the technical scheme of the present application. In the absence of conflict, the embodiments of the present application and the technical features in the embodiments can be combined with each other.

[0067] Figure 1 A method for simulating crop growth status based on meteorological grid data is shown, which is applied to a data simulation system. The method includes the following steps 110 to 150.

[0068] Step 110, obtaining meteorological environmental factor variables corresponding to the meteorological grid monitoring information, wherein the meteorological environmental factor variables are used to simulate the crop growth status under a plurality of crop growth status labels, and obtaining the crop growth status simulation results corresponding to the plurality of crop growth status labels in the target farming area.

[0069] Step 120: When simulating the growth state of the crops by the growth state simulation algorithms corresponding to the several crop growth state labels respectively, determining the state simulation original data of the growth state simulation algorithms corresponding to the several crop growth state labels respectively, wherein the state simulation original data is a description vector to be entered into the growth state simulation algorithm for state simulation.

[0070] Step 130: extracting cross-features shared between state feature mapping relationship networks corresponding to a plurality of state simulation original data, wherein the state feature mapping relationship network is obtained by performing data feature mining on the state simulation original data.

[0071] Step 140: Update the plurality of state simulation original data with the cross feature to obtain a plurality of state simulation updated data, wherein the plurality of state simulation original data and the plurality of state simulation updated data correspond to each other.

[0072] Step 150: Generate crop growth state simulation results corresponding to the target farming area under the plurality of crop growth state labels according to the plurality of state simulation update data; the plurality of crop growth state simulation results are used to combine and obtain crop yield prediction results for the target farming area; the crop yield prediction results are used as a guide for adjusting artificial meteorological environment parameters.

[0073] In step 110, the meteorological grid monitoring information is a large amount of detailed meteorological data collected by special monitoring equipment within a certain geographical range. Each grid represents a specific geographical location, and each location has its own unique meteorological environment element data.

[0074] Meteorological environmental factor variables refer to various elements and parameters that describe the meteorological environment, such as temperature, humidity, rainfall, wind speed, etc. They are used as model input variables to simulate the growth status of crops.

[0075] Crop growth status labels are classification labels for crop growth stages, such as germination period, growth period, flowering period, maturity period, etc. Each label corresponds to a specific growth stage.

[0076] The target crop area refers to the crop planting area predicted and simulated by the model. In this area, the meteorological and environmental factor variables are used to simulate the crop growth status.

[0077] The crop growth status simulation results are obtained by simulating the meteorological environment element variables and the crop growth status labels. It shows the expected status of crops at various growth stages under a given meteorological environment. These results can provide a reference for crop planting management.

[0078] Furthermore, meteorological environmental factor variables (or meteorological environmental factor characteristics) are data describing climate and weather conditions. They include the following main aspects:

[0079] Temperature: This is one of the most basic meteorological factors and has a direct impact on the growth of crops. For example, different crops have their own suitable growth temperature range;

[0080] Humidity: includes relative humidity and absolute humidity, which reflects the amount of moisture in the air and has an important impact on the growth environment of crops and the occurrence of pests and diseases;

[0081] Rainfall: This is a measure of the total amount of precipitation in a region, which has a direct impact on the irrigation needs of crops and soil moisture conditions;

[0082] Wind speed and direction: High wind speed may cause physical damage to crops, while wind direction may affect agricultural activities such as pesticide spraying;

[0083] Daylight duration and solar radiation: For crops that are closely related to photosynthesis, daylight duration and the intensity of solar radiation are important factors affecting their growth;

[0084] Atmospheric pressure: Although it has little impact on crop growth, it may also have some impact under certain conditions.

[0085] These meteorological and environmental factor variables are important inputs for constructing crop growth simulation models. They can reflect the specific conditions of the crop growth environment and are used to predict the growth status and yield of crops.

[0086] It can be seen that when meteorological environmental factor variables are represented as feature vectors, each dimension represents a specific meteorological environmental factor. In specific applications, these variables may need to be preprocessed, such as standardization or normalization, in order to be input into the model. The following is an example feature vector: Feature vector = [temperature, relative humidity, absolute humidity, rainfall, wind speed, wind direction, sunshine duration, solar radiation, atmospheric pressure].

[0087] For example, suppose the weather data at a certain time and place is: temperature 25°C, relative humidity 60%, absolute humidity 15g / m 3 , rainfall 0mm, wind speed 5m / s, wind direction east, sunshine duration 10 hours, solar radiation 500W / m 2 , atmospheric pressure 1013hPa, then the characteristic vector of the meteorological environment can be expressed as: [25, 60, 15, 0, 5, 90, 10, 500, 1013]. The wind direction is expressed in degrees, and the east wind is 90 degrees. In practical applications, more meteorological environment factors can be considered, and other forms of data preprocessing can be performed according to the needs of the model.

[0088] In other examples, crop growth status label: This is a classification mark for the various stages of crops in their growth cycle. For example, the growth process of a wheat plant can be divided into the following main stages: germination, seedling, jointing, heading, filling and maturity. Each stage corresponds to a label, such as "germination", "seedling", etc. The simulation result of crop growth status is the expected growth status of crops calculated by the model using meteorological environmental factor variables and crop growth status labels. For example, if there is a model that inputs today's meteorological environmental factor variables, the model may predict that wheat will enter the "jointing stage" tomorrow. The above simulation results can help understand the possible growth conditions of wheat in the future and make corresponding agricultural management decisions accordingly.

[0089] In step 120, the growth state simulation algorithm is a calculation method that receives meteorological environmental factor variables and crop growth state labels as input and outputs predicted crop growth states. The algorithm may be based on various machine learning or artificial intelligence technologies, such as decision trees, random forests, neural networks, etc. The state simulation raw data refers to the data input into the growth state simulation algorithm for simulation, including meteorological environmental factor variables and crop growth state labels. These data are usually processed to meet the needs of the algorithm, such as normalization, encoding, etc. The description vector is a representation of the input data, which integrates multiple related data (such as temperature, humidity, rainfall, etc.) into a vector. For example, if the temperature, humidity and wind speed are taken as three features, the description vector of a sample point may be [20, 0.8, 5], representing a temperature of 20°C, a humidity of 80%, and a wind speed of 5m / s, respectively.

[0090] For example, a decision tree-based growth state simulation algorithm is being used. This algorithm uses meteorological environmental factor variables (such as temperature, humidity, rainfall, etc.) and current crop growth state labels to predict the next crop growth state.

[0091] For example, one might have the following illustrative decision tree:

[0092] If the current state is "germination stage": if the temperature is >15℃ and the rainfall is >10mm, then the next stage is predicted to be "seedling stage"; otherwise, the next stage is still predicted to be "germination stage";

[0093] If the current state is "seedling stage": if the temperature is >20℃ and the rainfall is <30mm, then the next stage is predicted to be "jointing stage"; otherwise, the next stage is still predicted to be "seedling stage"

[0094] …

[0095] In practical applications, growth state simulation algorithms usually use more input features and may be based on more complex machine learning models, such as random forests, neural networks, etc. At the same time, they also take into account more crop growth state labels and the specific conditions of each growth stage.

[0096] In step 130, the state feature mapping relationship network is a model or structure that represents the correlation between data features. For example, in the simulation of crop growth status, meteorological environmental factors such as temperature, humidity, and rainfall may be closely related to the growth status of crops. The state feature mapping relationship network is used to reveal the relationship between these features and the growth status. Shared cross-features: In multiple different state simulation raw data sets, there may be some common features, which are shared cross-features. For example, for wheat and corn, both crops require water and sunlight, so "rainfall" and "sunshine duration" may become shared cross-features. Data feature mining is a process of extracting useful information or knowledge from raw data. In this process, various statistical methods, machine learning algorithms and other tools may be used to discover hidden patterns, correlations, trends, etc. in the data. For example, in the state simulation raw data, it may be found through feature mining that when the temperature and humidity reach a certain value, the growth state of the crop will change.

[0097] In terms of the state feature mapping network, for example, a farmland is being studied where wheat is planted. In order to simulate the growth state of wheat, some meteorological environmental factor data may be collected, such as temperature, humidity, rainfall, etc. At the same time, it is also necessary to observe the growth state of wheat, such as its growth rate, color, pests and diseases, etc.

[0098] In this process, a state feature mapping relationship network can be established, which describes the relationship between various meteorological environmental elements (i.e., state features) and the growth state of wheat. For example, it may be found that when the temperature is 20-25°C and the humidity is 60-70%, the growth state of wheat is the best. Conversely, if the temperature is too high or too low, or the humidity is insufficient, the growth state of wheat may deteriorate. Such a relationship can be expressed through graphics, tables, algorithms, etc. to form a mapping relationship network.

[0099] In terms of shared cross-characteristics, for example, corn is also planted on the same farmland. Like wheat, the growth state of corn is also affected by factors such as temperature, humidity, and rainfall.

[0100] In this case, it can be found that although wheat and corn are different crops, they are both affected by the same meteorological and environmental factors. These meteorological and environmental factors that affect the two crops can be regarded as shared cross-characteristics.

[0101] For example, it may be found that for both wheat and corn, their growth status is optimal when the temperature is 20-25°C and the humidity is 60-70%. This means that the two features of "temperature" and "humidity" are shared cross-features for simulating the growth status of wheat and corn.

[0102] In step 140, after the cross features are extracted, these features are used to update the original state simulation data. The update process may involve some adjustments or transformations to the original data to reflect the impact of the cross features. For example, if a shared cross feature (such as temperature) is found to have a significant impact on the growth state of crops, the weight of this feature may be increased in the simulation data. The data obtained after the update is "state simulation update data". The correspondence here means that there is a mapping relationship or connection between the original state simulation data and the updated state simulation data. That is, each state simulation original data can be converted into a state simulation update data in some way (such as adding, deleting or modifying features, etc.). For example, if the weight of the "temperature" feature in the original data is increased by 10%, then there is a corresponding relationship between the original data and the updated data: updated data = original data + 10% * temperature feature.

[0103] In step 150, the crop yield prediction result is the expected crop yield obtained by simulating the crop growth state and combining it with the relationship between the meteorological environment elements. For example, if it is found that a certain crop grows best under certain temperature and humidity conditions, then under these conditions, it can be predicted that the crop yield will reach the highest. This expected crop yield is the crop yield prediction result.

[0104] Furthermore, the crop yield forecast results are used as a guide for the adjustment of artificial meteorological environmental parameters, which means that the meteorological environmental parameters of the farmland can be adjusted according to the crop yield forecast results to optimize the growth conditions of crops and increase the yield of crops. For example, if it is predicted that the yield of a certain crop will increase at a higher temperature, the temperature of the farmland can be adjusted by artificial means (such as building greenhouses, etc.) to make it close to the predicted optimal temperature. Similarly, if the forecast results show that rainfall has a greater impact on the yield of crops, measures such as irrigation can also be taken to adjust the humidity of the farmland. Such adjustments are made based on the crop yield forecast results, with the aim of achieving higher crop yields.

[0105] The above technical solution is introduced below through a complete example.

[0106] First, the environmental factor data of a piece of farmland, such as temperature, humidity and rainfall, are obtained through the meteorological grid monitoring system. Then, using these data, the growth state simulation is carried out under different wheat growth states (such as seedling stage, jointing stage, heading stage, etc.), and the simulation results of the farmland under different growth states are obtained.

[0107] During the simulation process, the state simulation raw data of the simulation algorithm corresponding to each growth state label is determined. These raw data are description vectors and will be entered into the growth state simulation algorithm for state simulation.

[0108] Next, we extract the cross-features shared between the state feature mapping networks from the raw data of these state simulations. This may involve some data analysis and machine learning techniques to find out which features play a key role in all growth state simulations. For example, we may find that temperature and humidity are very important factors in all growth states.

[0109] These shared cross-features are then used to update the state simulation raw data to generate state simulation update data. For example, the weights of temperature and humidity features may be increased because they have a significant impact on all growth states.

[0110] Finally, by simulating and updating the data based on these states, simulation results of the target farmland under different growth states can be generated. By combining these simulation results, the wheat yield prediction results of the farmland can be obtained. Then, the meteorological environment parameters of the farmland can be adjusted according to this prediction result to achieve the purpose of optimizing growth conditions and increasing yields. For example, if the prediction results show that higher temperature and humidity will lead to better growth effects and higher yields, the temperature and humidity of the farmland can be artificially adjusted by building greenhouses and other means.

[0111] In summary, this technical solution can effectively predict the growth status of crops through in-depth analysis and simulation of meteorological environmental factor variables, thereby helping to obtain more accurate crop yield forecast results. This accurate prediction not only helps farmers make crop planting plans in advance, but also provides a scientific basis for agricultural managers to optimize crop planting structure and agricultural resource allocation.

[0112] In addition, the technical solution also emphasizes the importance of shared cross-features. By mining the shared cross-features between multiple state simulation raw data, we can have a deeper understanding of how various factors affect crop growth, which will help find the optimal crop growth conditions and adjust the meteorological environment parameters of the farmland accordingly.

[0113] Finally, adjusting meteorological environmental parameters according to crop yield forecast results can not only increase crop yields, but also improve resource utilization efficiency, reduce agricultural production costs, and further increase farmers' income. In general, this technical solution has important practical significance and broad application prospects for agricultural production.

[0114] In some possible embodiments, the step 120 of determining the state simulation raw data of the growth state simulation algorithm corresponding to the plurality of crop growth state tags respectively includes steps 121 and 122 .

[0115] Step 121: Acquire crop growth monitoring information, where the crop growth monitoring information is information obtained by monitoring crops in the target farming area, and the crop growth monitoring information is used to generate a simulation result reflecting the crop growth status of the target farming area.

[0116] Step 122: using the crop growth monitoring information as a simulation reference, determining the state simulation original data of the growth state simulation algorithm corresponding to the plurality of crop growth state labels, wherein the simulation reference is used to determine the crop growth trend characteristics when simulating the meteorological environment element variables.

[0117] In the embodiment of this technical solution, step 120 includes step 121 and step 122. Step 121 involves obtaining crop growth monitoring information. This information is obtained by monitoring crops in the target farming area, and includes various data on the growth status of crops, such as plant height, leaf area index, etc. These crop growth monitoring information will be used to generate crop growth status simulation results reflecting the target farming area. Step 122 is to use the crop growth monitoring information as a simulation reference to determine the state simulation raw data of the growth state simulation algorithm corresponding to each crop growth state label. "Simulation reference" means using the actually collected crop growth monitoring information as a reference to simulate the growth trend characteristics of crops under different meteorological environmental element variables. For example, if you know the meteorological conditions such as temperature and humidity on a certain day, you can predict the growth status of crops based on these conditions.

[0118] In this way, we first obtain crop growth information through field monitoring, and then determine the state simulation raw data corresponding to each crop growth state label based on this. In this way, we can not only more accurately reflect the crop growth status in the target farming area, but also predict the possible growth trend characteristics of crops under different meteorological environments through simulation. This is undoubtedly of great significance for optimizing crop planting, increasing crop yields and agricultural resource utilization efficiency.

[0119] In some preferred embodiments, obtaining crop growth monitoring information in step 121 includes: obtaining at least one data set obtained by monitoring the target farming area as the crop growth monitoring information, wherein the data set is the result of monitoring the target farming area based on a set crop growth status label.

[0120] In some preferred embodiments, the step 121 of obtaining crop growth monitoring information is more specific. This step includes obtaining at least one data set obtained by monitoring the target farming area as crop growth monitoring information. This "data set" is the result obtained by monitoring based on the set crop growth status tag.

[0121] For example, a series of crop growth status labels may be set, such as "germination period", "tillering period", "heading period", etc. Then, during field monitoring, relevant data will be collected based on these labels. For example, in the "germination period", information such as seed germination rate and growth rate may be recorded; in the "tillering period", the average number of tillers per plant may be observed and recorded. All this information is integrated together to form a data set, which is the so-called crop growth monitoring information.

[0122] In this way, very detailed and accurate crop growth information can be obtained, which is very helpful for simulating crop growth status and predicting crop yields. At the same time, it also enables us to understand and grasp the specific impact of various meteorological and environmental factors on crop growth, further guide crop planting and management, optimize resource allocation, and improve agricultural production efficiency.

[0123] In some alternative embodiments, step 130 of extracting cross features shared between state feature mapping relationship networks corresponding to a plurality of state simulation raw data includes steps 131 and 132.

[0124] Step 131: Perform convolution operations on the plurality of state simulation raw data respectively to obtain the state feature mapping relationship networks corresponding to the plurality of state simulation raw data respectively.

[0125] Step 132: Locally downsample the plurality of state feature mapping relationship networks to obtain a state involvement vector, wherein the state involvement vector is used to reflect the cross-features shared between the plurality of state feature mapping relationship networks.

[0126] In this alternative embodiment, step 130 includes step 131 and step 132. Step 131 involves performing a convolution operation on a number of state simulation raw data. "Convolution operation" is a common signal processing technique, which is used here to extract characteristic information of the growth state of crops. Through this operation, a state feature mapping relationship network corresponding to each state simulation raw data can be obtained. These relationship networks reflect how meteorological and environmental factors affect the growth state of crops. Step 132 involves local downsampling of a number of state feature mapping relationship networks. "Local downsampling" is a technique for reducing data dimensionality, used to extract key features and reduce computational complexity. This step will generate a state involvement vector, which can reflect the cross-features shared between a number of state feature mapping relationship networks.

[0127] Through the above two steps, key cross-features can be extracted from a large amount of state simulation raw data, and based on this, more accurate crop growth state prediction can be made. This method can effectively utilize the information of various meteorological environmental factors and avoid excessive computational complexity. It is a practical and efficient crop growth state prediction technology.

[0128] Under some optional design ideas, in step 131, convolution operations are performed on the several state simulation original data respectively to obtain the state feature mapping relationship network corresponding to the several state simulation original data respectively, including steps 1311 to 1313.

[0129] Step 1311: Perform convolution operations on the plurality of state simulation raw data respectively to obtain state simulation convolution vectors corresponding to the plurality of crop growth state labels respectively.

[0130] Step 1312: obtain sensor monitoring element vectors corresponding to the several crop growth status labels respectively, wherein the sensor monitoring element vectors are linear vectors obtained based on sensor variables corresponding to the crop growth status labels, and the sensor monitoring element vectors are used to reflect the periodic data of the corresponding crop growth status labels, wherein there is a correspondence between the several sensor monitoring element vectors and the several state simulation convolution vectors.

[0131] Step 1313: Based on the correspondence, perform vector integration on the state simulation convolution vector and the sensor monitoring element vector under the same crop growth state label to obtain the state feature mapping relationship network corresponding to the plurality of state simulation original data.

[0132] Under this optional design idea, step 131 includes steps 1311 to 1313. Step 1311 is to perform convolution operations on several state simulation raw data respectively to obtain state simulation convolution vectors corresponding to each crop growth state label. "Convolution vector" is feature information extracted from the raw data through convolution operation, which reflects the impact of various meteorological environmental factors on the growth state of crops. Step 1312 involves obtaining sensor monitoring element vectors corresponding to each crop growth state label. "Sensor monitoring element vector" is a linear vector obtained based on the sensor variables corresponding to the crop growth state label, which reflects the periodic data of the corresponding crop growth state label. Sensor variables may include temperature, humidity, light intensity, etc., which are environmental parameters collected in real time by monitoring equipment. Step 1313 is to integrate the state simulation convolution vector and the sensor monitoring element vector under the same crop growth state label according to the above correspondence, so as to obtain the state feature mapping relationship network corresponding to each state simulation raw data.

[0133] In this way, through convolution operations and vector integration, important features reflecting the growth status of crops are extracted from the original state simulation data, while taking into account the impact of environmental parameters on crop growth. This method can more accurately predict the growth status and yield of crops, optimize agricultural resource allocation, and improve agricultural production efficiency.

[0134] In some possible embodiments, obtaining the sensor monitoring element vectors corresponding to the several crop growth status labels respectively in step 1312 includes: obtaining sensor variables corresponding to the several crop growth status labels respectively, the sensor variables are used to reflect the sensor distribution characteristics for generating the corresponding crop growth status simulation results, the sensor variables include sensor area labels and sensor category labels, the sensor area labels are used to reflect the distribution of the sensors in the sensor monitoring feature space relative to the target farming area, and the sensor category labels characterize the monitoring categories of the sensors in the sensor monitoring feature space relative to the target farming area; preprocessing the several sensor variables respectively to obtain preprocessing results corresponding to the several sensor variables respectively, the preprocessing results are linear vectors obtained by preprocessing the sensor variables in the growth monitoring task link; extracting knowledge elements from the preprocessing results by setting a knowledge extraction algorithm to obtain the sensor monitoring element vectors corresponding to the several crop growth status labels respectively.

[0135] In this possible embodiment, step 1312 involves obtaining sensor monitoring element vectors corresponding to several crop growth status labels. This process includes the following sub-steps: First, obtain the sensor variables corresponding to each crop growth status label. "Sensor variables" reflect the sensor distribution characteristics of generating the corresponding crop growth status simulation results. Sensor variables include "sensor area labels" and "sensor category labels". The sensor area label reflects the distribution of the sensor relative to the target farming area in the sensor monitoring feature space, for example, whether the sensor is located in the center, edge or corner of the field; and the sensor category label characterizes the monitoring category of the sensor relative to the target farming area in the sensor monitoring feature space, for example, the sensor is used to monitor temperature, humidity, light intensity or other meteorological environmental factors. Then, preprocess each sensor variable to obtain a preprocessing result corresponding to each sensor variable. Preprocessing usually includes operations such as data cleaning and normalization, which are intended to reduce noise, eliminate outliers and convert data into a form suitable for further processing. The preprocessing result is a linear vector obtained by preprocessing the sensor variable in the growth monitoring task link. Finally, the knowledge extraction algorithm is used to extract knowledge elements from the preprocessing results to obtain sensor monitoring element vectors corresponding to the growth status labels of each crop. This step is to extract useful information from the preprocessing results to facilitate subsequent analysis and simulation.

[0136] Therefore, this technical solution can more accurately reflect the growth status of crops by acquiring and processing sensor variables, providing an important reference for optimizing crop planting, increasing crop yields and agricultural resource utilization efficiency.

[0137] Under some possible design ideas, the step 132 performs local downsampling on a plurality of state feature mapping relationship networks to obtain a state involvement vector, including steps 1321 to 1323.

[0138] Step 1321, determine the meteorological environment impact feature sets respectively represented by the several state feature mapping relationship networks, and the meteorological environment impact feature set is used to reflect the set of several meteorological environment impact features in the growth monitoring task link when the state feature mapping relationship network is obtained.

[0139] Step 1322: Focusing a plurality of meteorological environment impact features at the same feature position in a plurality of meteorological environment impact feature sets to obtain a plurality of local focusing weights.

[0140] Step 1323: downsample the plurality of local focusing weights to obtain the state involvement vector.

[0141] Under this possible design idea, step 132 includes steps 1321 to 1323. Step 1321 first needs to determine the meteorological environment impact feature sets represented by several state feature mapping relationship networks. The meteorological environment impact feature set is used to reflect the set of several meteorological environment impact features in the growth monitoring task link when the state feature mapping relationship network is obtained. These features may include meteorological environmental factors such as temperature, humidity, and light intensity that affect the growth of crops. Step 1322 is to focus on several meteorological environment impact features in the same feature position in several meteorological environment impact feature sets to obtain several local focusing weights. Focusing refers to focusing attention or weight on certain important features and reducing consideration of other less important features. The local focusing weight is the weight of each meteorological environment impact feature obtained by this method. Step 1323 is to downsample the several local focusing weights to obtain the state involvement vector. Downsampling is a technology to reduce the amount of data. It reduces data complexity and speeds up calculations by randomly selecting part of the data in the data set or compressing the data in some way. The state involvement vector is a vector representing the growth state of crops obtained after downsampling.

[0142] It can be seen that by locally focusing and downsampling the state feature mapping relationship network, the most representative and discriminative meteorological environment impact characteristics can be extracted, thereby more accurately reflecting and predicting the growth status of crops, which helps to improve agricultural production efficiency and agricultural product quality.

[0143] In some other embodiments, the step 140 described in which the plurality of state simulation original data are updated with the cross feature to obtain a plurality of state simulation updated data includes steps 141 to 143.

[0144] Step 141: Obtain a state involvement vector representing the cross feature.

[0145] Step 142: Based on the crop growth state labels corresponding to the several state simulation original data and the state involvement vector, obtain meteorological environment associated growth description vectors corresponding to the several crop growth state labels, wherein the meteorological environment associated growth description vector is used to reflect the contribution of the state involvement vector to the state simulation original data at the time series description level, wherein there is a correspondence between the several meteorological environment associated growth description vectors and the several state simulation original data.

[0146] Step 143: According to the correspondence, the plurality of state simulation update data are obtained by associating the growth description vector with the meteorological environment under the same crop growth state label and the state simulation original data.

[0147] In these embodiments, step 140 describes updating a number of state simulation raw data with cross features to obtain a number of state simulation update data. This includes the following sub-steps: step 141 is to obtain a state involvement vector that characterizes the cross features. The cross features here refer to the correlation or interactivity between the meteorological environment impact features obtained by different types of sensors. The state involvement vector is a vector representing the growth state of crops obtained after downsampling in the previous step. Step 142 is to obtain a meteorological environment-related growth description vector corresponding to a number of crop growth state labels based on the crop growth state labels and state involvement vectors corresponding to the state simulation raw data. "Meteorological environment-related growth description vector" reflects the contribution of the state involvement vector to the state simulation raw data at the temporal description level, that is, it reveals how different meteorological environmental factors change over time and affect the growth state of crops. At the same time, there is a correspondence between each meteorological environment-related growth description vector and the state simulation raw data, which means that each description vector corresponds to a raw data. Step 143 is to obtain a number of state simulation update data based on this correspondence by associating the growth description vector and the state simulation original data under the same crop growth state label with the meteorological environment. That is, each update data is generated by the corresponding description vector and the original data.

[0148] This design, by introducing cross-features and state-involved vectors, can more accurately reflect and predict the growth status of crops, providing a more scientific and precise basis for decision-making in agricultural production. In addition, this solution can also update simulation data in real time, allowing the model to self-optimize over time, further improving the accuracy and reliability of predictions.

[0149] Furthermore, step 142 as described above obtains meteorological environment associated growth description vectors corresponding to the plurality of crop growth state labels respectively based on the crop growth state labels respectively corresponding to the plurality of state simulation original data and the state involvement vectors, including steps 1421 to 1423.

[0150] Step 1421: determine the sensor monitoring knowledge graphs corresponding to the plurality of crop growth status labels respectively, wherein the sensor monitoring knowledge graph is a knowledge graph generated based on the sensors used when determining the corresponding crop growth status labels.

[0151] Step 1422: Using the sensor monitoring knowledge graph corresponding to the crop growth status label monitoring category as a template, perform knowledge migration on the state involvement vector to obtain knowledge migration vectors corresponding to the plurality of crop growth status labels.

[0152] Step 1423: Based on the plurality of crop growth status labels and the corresponding knowledge transfer vectors, obtain the meteorological environment associated growth description vectors corresponding to the plurality of crop growth status labels in the growth monitoring task link.

[0153] In this technical solution, the content of step 142 is to obtain the meteorological environment-related growth description vectors corresponding to the several crop growth state labels based on the crop growth state labels and state involvement vectors corresponding to the several state simulation raw data. It includes the following sub-steps: Step 1421 is to determine the sensor monitoring knowledge graph corresponding to the several crop growth state labels. The "sensor monitoring knowledge graph" is a knowledge graph generated based on the sensor used when determining the corresponding crop growth state label. The knowledge graph is a structured knowledge representation method that displays the relationship between entities in the form of a graph to facilitate machine understanding and processing. Step 1422 is to use the sensor monitoring knowledge graph corresponding to the crop growth state label monitoring category as a template to perform knowledge transfer on the state involvement vector to obtain the knowledge transfer vector representing the corresponding to the several crop growth state labels. "Knowledge transfer" refers to applying the knowledge that has been learned to new scenarios or tasks, and the "knowledge transfer vector" is a vector obtained through knowledge transfer, which represents the projection or mapping of the state involvement vector on the knowledge graph. Step 1423 is to obtain the meteorological environment-related growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link according to the several crop growth status labels and the corresponding knowledge transfer vectors. That is, each description vector is jointly generated by the corresponding crop growth status label and the knowledge transfer vector.

[0154] This design, through the introduction of sensor monitoring knowledge graph and knowledge transfer technology, can effectively apply existing knowledge to new scenarios or tasks, improving the generalization ability of the model. In addition, this solution can also more accurately reflect and predict the growth status of crops, providing a more scientific and accurate decision-making basis for agricultural production.

[0155] In some preferred embodiments, step 1423 obtains the meteorological environment associated growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link based on the several crop growth status labels and the corresponding knowledge transfer vectors, including: obtaining the thermal values ​​of the crop growth status respectively represented by the several crop growth status labels; under the same crop growth status label, obtaining the meteorological environment associated growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link based on the crop growth status thermal values ​​and the knowledge transfer vectors.

[0156] In this preferred embodiment, step 1423 is further refined into the following operations: First, it is necessary to obtain the thermal values ​​of the crop growth status represented by several crop growth status labels. "Thermal value of crop growth status" is a numerical indicator to measure the intensity or activity of crop growth status. For example, if a status label represents "good growth", then its corresponding thermal value may be relatively high; conversely, if a status label represents "slow growth", then its corresponding thermal value may be relatively low. Then, under the same crop growth status label, based on the crop growth status thermal value and the knowledge transfer vector, the meteorological environment-related growth description vectors corresponding to several crop growth status labels in the growth monitoring task link are obtained. This step mainly combines the thermal value of the crop growth status and the knowledge transfer vector to generate a more comprehensive and accurate description vector.

[0157] In this way, by introducing the thermal value of crop growth status, the growth status of crops can be measured and reflected more accurately, improving the prediction accuracy of the model. In addition, by combining the thermal value and the knowledge transfer vector to generate a description vector, multi-source information can be effectively integrated, further improving the expression and generalization capabilities of the model. Therefore, this technical solution helps to improve agricultural production efficiency, improve the quality of agricultural products, and provide a more scientific and accurate decision-making basis for agricultural production.

[0158] Furthermore, under the same crop growth status label, according to the crop growth status thermal value and the knowledge transfer vector, the meteorological environment associated growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link are obtained, including: determining a meteorological environment impact feature set represented by a state feature mapping relationship network corresponding to the several crop growth status labels, the meteorological environment impact feature set includes several meteorological environment impact features, and each meteorological environment impact feature corresponds to a crop growth status thermal value; taking the crop growth status thermal value corresponding to the meteorological environment impact feature as the meteorological environment impact feature value, deriving several meteorological environment impact features in the meteorological environment impact feature set to obtain a meteorological environment impact feature chain set with the same meteorological environment impact feature value; according to the meteorological environment impact feature chain set and the knowledge transfer vector, the meteorological environment associated growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link are obtained.

[0159] In this technical solution, based on the thermal value of crop growth state and the knowledge transfer vector, the process of obtaining the meteorological environment-related growth description vectors corresponding to several crop growth state labels in the growth monitoring task link is further refined into the following operations: First, determine the meteorological environment impact feature set represented by the state feature mapping relationship network corresponding to several crop growth state labels. These feature sets include several meteorological environment impact features, each of which corresponds to a thermal value of crop growth state. Then, with the thermal value of crop growth state corresponding to the meteorological environment impact feature as the feature value, several features in the meteorological environment impact feature set are derived to obtain a meteorological environment impact feature chain set with the same feature value. "Feature chain set" refers to a set formed by features with the same feature value, which can better reflect the correlation between different features. Finally, based on the meteorological environment impact feature chain set and the knowledge transfer vector, the meteorological environment-related growth description vectors corresponding to several crop growth state labels in the growth monitoring task link are obtained. This step is mainly to combine the feature chain set and the knowledge transfer vector to generate a more comprehensive and accurate description vector.

[0160] In this way, by introducing the state feature mapping relationship network, the thermal value of crop growth state and the knowledge transfer vector, as well as the concept of feature chain, the growth state of crops can be measured and reflected more accurately, improving the prediction accuracy of the model. In addition, by fusing multiple information to generate description vectors, multi-source information can be effectively integrated, further improving the expression and generalization capabilities of the model. Therefore, this technical solution helps to improve agricultural production efficiency, improve the quality of agricultural products, and provide a more scientific and accurate decision-making basis for agricultural production.

[0161] Furthermore, based on the meteorological environment impact feature chain set and the knowledge transfer vector, the meteorological environment associated growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link are obtained, including: extracting knowledge elements from the meteorological environment impact feature chain set with a deep learning branch to obtain meteorological environment impact feature knowledge; under the same crop growth status label, aggregating the meteorological environment impact feature knowledge and the knowledge transfer vector to obtain the meteorological environment associated growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link.

[0162] In this technical solution, the process of obtaining meteorological environment-related growth description vectors corresponding to several crop growth status labels in the growth monitoring task link according to the meteorological environment impact feature chain set and the knowledge transfer vector is further refined into the following operations: First, the meteorological environment impact feature chain set is extracted with a deep learning branch to obtain meteorological environment impact feature knowledge. "Deep learning branch" refers to a branch or subsystem using a deep learning method, which can automatically learn and extract the inherent laws and characteristics of data. "Knowledge element extraction" is the important information or knowledge extracted from the meteorological environment impact feature chain set by the deep learning branch, which is called "meteorological environment impact feature knowledge". Then, under the same crop growth status label, the meteorological environment impact feature knowledge and the knowledge transfer vector are aggregated to obtain the meteorological environment-related growth description vectors corresponding to several crop growth status labels in the growth monitoring task link. "Aggregation" is the process of merging multiple elements or data into a whole. Here, it mainly merges the meteorological environment impact feature knowledge and the knowledge transfer vector to generate a more comprehensive and accurate description vector.

[0163] In this way, by extracting knowledge elements through deep learning branches and aggregating them with knowledge transfer vectors, the growth status of crops can be more accurately reflected and predicted. In addition, the solution utilizes the self-learning and extraction capabilities of deep learning to extract the factors that have the greatest impact on crop growth from a large amount of complex meteorological and environmental data, thereby more accurately predicting the growth status of crops and providing a more scientific and accurate basis for decision-making in agricultural production.

[0164] In some other embodiments, based on the correspondence, the meteorological environment associated growth description vector and the state simulation original data under the same crop growth status label are vector integrated to obtain the several state simulation update data: the meteorological environment associated growth description vectors corresponding to several crop growth status labels are mapped to the simulation feature relationship network to obtain the growth environment associated mapping vectors corresponding to several crop growth status labels; under the same crop growth status label, the growth environment associated mapping vector and the state simulation original data are vector integrated to obtain the state simulation update data corresponding to several crop growth status labels.

[0165] In these embodiments, according to the correspondence, the meteorological environment-associated growth description vector and the state simulation original data under the same crop growth state label are vector-integrated to obtain a number of state simulation update data, including the following steps: First, the meteorological environment-associated growth description vectors corresponding to the several crop growth state labels are mapped to the simulation feature relationship network to obtain the growth environment-associated mapping vectors corresponding to the several crop growth state labels. The "simulation feature relationship network" is a network structure that can represent the relationship and interaction between various features and is used to simulate the complex situation of the real world. The "growth environment-associated mapping vector" is a vector obtained through the mapping operation, which reflects the impact of the growth environment on the growth state of the crops.

[0166] Then, under the same crop growth state label, the growth environment associated mapping vector and the state simulation original data are vector integrated to obtain the state simulation update data corresponding to several crop growth state labels. "Vector integration" is a common machine learning technology that integrates the information of multiple vectors (such as growth environment associated mapping vectors and state simulation original data) to generate a new, more comprehensive vector (such as state simulation update data).

[0167] This design, by introducing the simulated feature relationship network and vector integration technology, can better integrate various information, thereby more accurately reflecting and predicting the growth status of crops. In addition, this solution can also update the simulation data in real time, so that the model can self-optimize over time, further improving the accuracy and reliability of the prediction. Therefore, this technical solution helps to improve agricultural production efficiency, improve the quality of agricultural products, and provide a more scientific and accurate decision-making basis for agricultural production.

[0168] In some examples, each crop growth state label corresponds to x growth state simulation algorithms, where x is a positive integer. Then, the step 150 described in which the crop growth state simulation results corresponding to the target farming area under the crop growth state labels are generated according to the plurality of state simulation update data includes steps 151 to 153.

[0169] Step 151, perform crop growth state simulation on the state simulation update data of the yth growth state simulation algorithm corresponding to the several crop growth state labels, to obtain the state simulation original data of the y+1th growth state simulation algorithm corresponding to the several crop growth state labels, where y is a positive integer not greater than x.

[0170] Step 152: In response to the xth growth state simulation algorithm, a state simulation reasoning vector output by the xth growth state simulation algorithm corresponding to the plurality of crop growth state labels is obtained.

[0171] Step 153: Generate the crop growth state simulation results corresponding to the target farming area under the plurality of crop growth state labels based on the plurality of state simulation reasoning vectors.

[0172] In this technical solution, each crop growth state label corresponds to x growth state simulation algorithms, where x is a positive integer. The process of generating crop growth state simulation results corresponding to a target crop area under a plurality of crop growth state labels according to a plurality of state simulation update data includes the following steps: Step 151 is to simulate the crop growth state for the state simulation update data of the yth growth state simulation algorithm corresponding to a plurality of crop growth state labels, and obtain the state simulation original data of the y+1th growth state simulation algorithm corresponding to a plurality of crop growth state labels, where y is a positive integer not greater than x; Step 152 is to obtain the state simulation reasoning vector output by the xth growth state simulation algorithm corresponding to a plurality of crop growth state labels after passing through the xth growth state simulation algorithm. "State simulation reasoning vector" refers to the vector obtained after the state is inferred by the simulation algorithm, which reflects the understanding and prediction of the crop growth state by the simulation algorithm; Step 153 is to generate the crop growth state simulation results corresponding to a target crop area under a plurality of crop growth state labels based on a plurality of state simulation reasoning vectors. That is to say, each simulation result is generated by the corresponding reasoning vector.

[0173] In summary, the above technical solution can simulate and predict the growth status of crops from different angles and levels by introducing multiple growth status simulation algorithms, thereby improving the accuracy and reliability of the prediction. In addition, this solution can also update the simulation data in real time and generate simulation results through inference vectors, so that the model can self-optimize over time, further improving the accuracy and reliability of the prediction. Therefore, this technical solution helps to improve agricultural production efficiency, improve the quality of agricultural products, and provide a more scientific and accurate decision-making basis for agricultural production.

[0174] In some optional embodiments, step 153 generates the crop growth state simulation results corresponding to the target farming area under the several crop growth state labels based on the several state simulation reasoning vectors, including steps 1531 to 1533.

[0175] Step 1531, perform a cyclic crop growth state simulation according to the state simulation reasoning vector under the several crop growth state labels until the loop condition is met, and obtain state simulation decision vectors corresponding to the several crop growth state labels respectively, and the state simulation decision vector is used to reflect the linear vector obtained after the meteorological environment element variable is simulated for the growth state of the crop.

[0176] Step 1532: The plurality of state simulation decision vectors are respectively transmitted through simulation output branches to generate the crop growth state simulation results corresponding to the plurality of crop growth state labels in the target farming area.

[0177] In these optional embodiments, the process of generating crop growth state simulation results corresponding to a target farming area under a plurality of crop growth state labels based on a plurality of state simulation reasoning vectors includes the following steps: Step 1531 is to perform a cyclic crop growth state simulation according to the state simulation reasoning vector under a plurality of crop growth state labels until the cyclic condition is met, and obtain a state simulation decision vector corresponding to a plurality of crop growth state labels. The state simulation decision vector is used to reflect the linear vector obtained after the meteorological environment element variable is simulated for the growth state of crops. "Linear vector" refers to a vector that can be obtained by multiplication and vector addition in a vector space. Step 1532 is to generate crop growth state simulation results corresponding to a target farming area under a plurality of crop growth state labels through a simulation output branch respectively through a plurality of state simulation decision vectors. "Simulation output branch" is a system or subsystem that can receive state simulation decision vectors and generate simulation results based on these vectors.

[0178] In this way, by introducing loop simulation and simulation output branches, the growth status of crops can be more accurately reflected and predicted, improving the accuracy and reliability of predictions. In addition, this solution can also update simulation data in real time and generate simulation results through state simulation decision vectors, so that the model can self-optimize over time, further improving the accuracy and reliability of predictions. Therefore, this technical solution helps to improve agricultural production efficiency, improve the quality of agricultural products, and provide a more scientific and accurate decision-making basis for agricultural production.

[0179] On the basis of the above, a data simulation system is provided, comprising a processor and a memory communicating with each other, wherein the processor is used to retrieve a computer program from the memory and implement the above method by running the computer program.

[0180] Based on the above, a computer-readable storage medium is provided, on which a computer program is stored, and the computer program implements the above method when running.

[0181] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for simulating crop growth status based on meteorological grid data, characterized in that: Applied to a data simulation system, the method comprises: Obtain meteorological environment element variables corresponding to the meteorological grid monitoring information; wherein the meteorological environment element variables are used to simulate the growth status of crops under a plurality of crop growth status labels, and obtain the crop growth status simulation results corresponding to the plurality of crop growth status labels in the target farming area; When simulating the growth state of crops by using the growth state simulation algorithms corresponding to the plurality of crop growth state labels, determining the state simulation original data of the growth state simulation algorithms corresponding to the plurality of crop growth state labels; wherein the state simulation original data is a description vector to be input into the growth state simulation algorithm for state simulation; Extracting cross-features shared between state feature mapping relationship networks corresponding to a plurality of state simulation original data; wherein the state feature mapping relationship network is obtained by performing data feature mining on the state simulation original data; The plurality of state simulation original data are updated with the cross feature to obtain a plurality of state simulation update data; wherein the plurality of state simulation original data and the plurality of state simulation update data are corresponding to each other; Based on the several state simulation update data, the crop growth state simulation results corresponding to the target farming area under the several crop growth state labels are generated; wherein, the several crop growth state simulation results are used to combine to obtain the crop yield prediction result of the target farming area, and the crop yield prediction result is used as a guide for adjusting the artificial meteorological environment parameters.

2. The method according to claim 1, characterized in that: The step of determining the state simulation raw data of the growth state simulation algorithm corresponding to the plurality of crop growth state tags respectively comprises: Acquiring crop growth monitoring information, wherein the crop growth monitoring information is information obtained by monitoring crops in the target farming area, and the crop growth monitoring information is used to generate a simulation result reflecting the crop growth status of the target farming area; Taking the crop growth monitoring information as a simulation reference, determining the state simulation original data of the growth state simulation algorithm corresponding to the plurality of crop growth state tags respectively, wherein the simulation reference is used to determine the crop growth trend characteristics when simulating the meteorological environment element variables; The obtaining of crop growth monitoring information includes: obtaining at least one data set obtained from monitoring the target farming area as the crop growth monitoring information, wherein the data set is a result obtained from monitoring the target farming area based on a set crop growth status label.

3. The method according to claim 1, characterized in that The step of extracting the cross features shared between the state feature mapping relationship networks corresponding to the state simulation raw data respectively includes: Performing convolution operations on the plurality of state simulation original data respectively to obtain the state feature mapping relationship networks respectively corresponding to the plurality of state simulation original data; Local downsampling is performed on a plurality of state feature mapping relationship networks to obtain a state involvement vector, wherein the state involvement vector is used to reflect the cross-features shared between the plurality of state feature mapping relationship networks.

4. The method according to claim 3, characterized in that The convolution operation is performed on the plurality of state simulation original data respectively to obtain the state feature mapping relationship network corresponding to the plurality of state simulation original data respectively, including: Performing convolution operations on the plurality of state simulation raw data respectively to obtain state simulation convolution vectors corresponding to the plurality of crop growth state labels respectively; Acquire sensor monitoring element vectors corresponding to the plurality of crop growth status labels, respectively, wherein the sensor monitoring element vectors are linear vectors obtained based on sensor variables corresponding to the crop growth status labels, and the sensor monitoring element vectors are used to reflect periodic data of the corresponding crop growth status labels, wherein the plurality of sensor monitoring element vectors correspond to the plurality of state simulation convolution vectors; According to the correspondence, the state simulation convolution vector and the sensor monitoring element vector under the same crop growth state label are vector integrated to obtain the state feature mapping relationship network corresponding to the plurality of state simulation raw data respectively; Among them, the obtaining of sensor monitoring element vectors corresponding to the several crop growth status labels respectively includes: obtaining sensor variables corresponding to the several crop growth status labels respectively, the sensor variables are used to reflect the sensor distribution characteristics of generating corresponding crop growth status simulation results, the sensor variables include sensor area labels and sensor category labels, the sensor area labels are used to reflect the distribution of sensors in the sensor monitoring feature space relative to the target farming area, and the sensor category labels characterize the monitoring category of the sensors in the sensor monitoring feature space relative to the target farming area; preprocessing the several sensor variables respectively to obtain preprocessing results corresponding to the several sensor variables respectively, the preprocessing results are linear vectors obtained by preprocessing the sensor variables in the growth monitoring task link; extracting knowledge elements from the preprocessing results by setting a knowledge extraction algorithm to obtain the sensor monitoring element vectors corresponding to the several crop growth status labels respectively.

5. The method according to claim 3, characterized in that: The locally downsampling a plurality of state feature mapping relationship networks to obtain a state involvement vector includes: Determine a set of meteorological environment impact features respectively represented by the plurality of state feature mapping relationship networks, wherein the set of meteorological environment impact features is used to reflect a set of a plurality of meteorological environment impact features in a growth monitoring task link when the state feature mapping relationship network is obtained; Focusing a number of meteorological environment impact features at the same feature position in a number of meteorological environment impact feature sets to obtain a number of local focusing weights; The plurality of local focusing weights are downsampled to obtain the state involvement vector.

6. The method according to claim 1, characterized in that The updating of the plurality of state simulation original data by using the cross feature to obtain a plurality of state simulation update data includes: Acquire a state involvement vector representing the cross feature; According to the crop growth state labels corresponding to the plurality of state simulation original data and the state involvement vector, meteorological environment associated growth description vectors corresponding to the plurality of crop growth state labels are obtained, wherein the meteorological environment associated growth description vectors are used to reflect the contribution of the state involvement vector to the state simulation original data at the time series description level, wherein the plurality of meteorological environment associated growth description vectors and the plurality of state simulation original data have correspondence; According to the correspondence, the plurality of state simulation update data are obtained by associating the growth description vector with the meteorological environment under the same crop growth state label and the state simulation original data.

7. The method according to claim 6, characterized in that The method of obtaining meteorological environment-related growth description vectors corresponding to the plurality of crop growth state labels and the state involvement vectors respectively corresponding to the plurality of crop growth state labels according to the plurality of state simulation original data comprises: determining sensor monitoring knowledge graphs corresponding to the plurality of crop growth state labels respectively, wherein the sensor monitoring knowledge graph is a knowledge graph generated based on the sensor used when determining the corresponding crop growth state labels; using the sensor monitoring knowledge graph corresponding to the crop growth state label monitoring category as a template, performing knowledge migration on the state involvement vector to obtain knowledge migration vectors representing the plurality of crop growth state labels respectively corresponding to the plurality of crop growth state labels; obtaining the meteorological environment-related growth description vectors corresponding to the plurality of crop growth state labels respectively in the growth monitoring task link according to the plurality of crop growth state labels and the corresponding knowledge migration vectors; Among them, the obtaining of the meteorological environment-related growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link according to the several crop growth status labels and the corresponding knowledge transfer vectors includes: obtaining the crop growth status thermal values ​​respectively represented by the several crop growth status labels; under the same crop growth status label, obtaining the meteorological environment-related growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link according to the crop growth status thermal values ​​and the knowledge transfer vectors; Among them, under the same crop growth state label, according to the crop growth state thermal value and the knowledge transfer vector, the meteorological environment associated growth description vectors corresponding to the several crop growth state labels in the growth monitoring task link are obtained, including: determining a meteorological environment impact feature set represented by a state feature mapping relationship network corresponding to the several crop growth state labels, the meteorological environment impact feature set includes several meteorological environment impact features, and each meteorological environment impact feature corresponds to a crop growth state thermal value; using the crop growth state thermal value corresponding to the meteorological environment impact feature as the meteorological environment impact feature value, deriving several meteorological environment impact features in the meteorological environment impact feature set to obtain a meteorological environment impact feature chain set with the same meteorological environment impact feature value; according to the meteorological environment impact feature chain set and the knowledge transfer vector, the meteorological environment associated growth description vectors corresponding to the several crop growth state labels in the growth monitoring task link are obtained; Among them, according to the meteorological environment impact feature chain set and the knowledge transfer vector, the meteorological environment associated growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link are obtained, including: using a deep learning branch to extract knowledge elements from the meteorological environment impact feature chain set to obtain meteorological environment impact feature knowledge; under the same crop growth status label, the meteorological environment impact feature knowledge and the knowledge transfer vector are aggregated to obtain the meteorological environment associated growth description vectors corresponding to the several crop growth status labels in the growth monitoring task link.

8. The method according to claim 6, characterized in that According to the correspondence, the meteorological environment-related growth description vector and the state simulation original data under the same crop growth state label are vector-integrated to obtain the plurality of state simulation update data: Mapping the meteorological environment-related growth description vectors corresponding to the plurality of crop growth status labels to the simulated feature relationship network to obtain growth environment-related mapping vectors corresponding to the plurality of crop growth status labels; Under the same crop growth state label, the growth environment association mapping vector and the state simulation original data are vector-integrated to obtain the state simulation update data corresponding to a plurality of crop growth state labels.

9. The method according to claim 1, characterized in that: Each crop growth state label corresponds to x growth state simulation algorithms, where x is a positive integer; The step of generating the crop growth state simulation results corresponding to the crop growth state labels in the target farming area according to the plurality of state simulation update data includes: Performing crop growth state simulation on the state simulation update data of the yth growth state simulation algorithm corresponding to the plurality of crop growth state labels, to obtain the state simulation original data of the y+1th growth state simulation algorithm corresponding to the plurality of crop growth state labels, where y is a positive integer not greater than x; In response to the xth growth state simulation algorithm, obtaining a state simulation reasoning vector output by the xth growth state simulation algorithm corresponding to the plurality of crop growth state labels respectively; Generate the crop growth state simulation results corresponding to the target farming area under the crop growth state labels respectively based on the plurality of state simulation reasoning vectors; Among them, the method of generating the crop growth state simulation results corresponding to the target farming area under the several crop growth state labels based on several state simulation reasoning vectors includes: performing cyclic crop growth state simulation according to the state simulation reasoning vector under the several crop growth state labels until the cyclic condition is met, and obtaining state simulation decision vectors corresponding to the several crop growth state labels, the state simulation decision vectors being used to reflect the linear vector obtained after the growth state simulation of the meteorological environment element variables is performed on the crops; and generating the crop growth state simulation results corresponding to the target farming area under the several crop growth state labels respectively through simulation output branches.

10. A data simulation system, characterized in that: The data simulation system includes a processor and a memory that communicate with each other, and the processor is used to call a computer program from the memory and implement the method according to any one of claims 1 to 9 by running the computer program.

Images