Data processing method and device, computer equipment, storage medium and program product

Abstract

The invention relates to the technical field of agricultural science, and discloses a data processing method and device, computer equipment, a storage medium and a program product. The method comprises the steps that current object data are acquired, and the current object data comprise object data and meteorological data in a current generation interval; performing data conversion processing on the object data and the meteorological data in the current generation interval to obtain target processing data; inputting the target processing data into a pre-trained object prediction model, and outputting target object data of a next generation interval through the object prediction model; wherein the object prediction model is obtained by training object data samples and meteorological data samples in a plurality of historical generation intervals. By implementing the technical scheme of the invention, the target object data of the next generation interval can be predicted based on the object data and the meteorological data in the current generation interval, and the prediction efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of agricultural science and technology, and in particular to a data processing method, device, computer equipment, storage medium and program product. Background Art

[0002] Pests pose a serious threat to the ecology and economy, mainly in terms of damage to agriculture, disease transmission, and ecological imbalance. They not only directly lead to reduced crop yields, but also increase the economic burden on farmers.

[0003] To solve this problem, related technologies have constructed different prediction models to predict the occurrence patterns of pests. However, current prediction models cannot effectively capture the complex nonlinear relationships between pest occurrence data and often rely on large amounts of data that are difficult to obtain. At the same time, these models have poor anti-interference capabilities and are prone to falling into local optimality, resulting in insufficient prediction accuracy. Summary of the invention

[0004] In view of this, the present invention provides a data processing method, apparatus, computer device, storage medium and program product to solve the problems that the current prediction model cannot effectively capture complex nonlinear relationships, relies on data that is difficult to obtain and has insufficient anti-interference ability.

[0005] In a first aspect, the present invention provides a data processing method, comprising: obtaining current object data, the current object data including object data and meteorological data within a current generation interval; performing data conversion processing on the object data and meteorological data within the current generation interval to obtain target processing data; inputting the target processing data into a pre-trained object prediction model, and outputting target object data for the next generation interval through the object prediction model; wherein the object prediction model is trained based on object data samples and meteorological data samples within multiple historical generation intervals.

[0006] The data processing method provided by the embodiment of the present invention ensures that the model makes predictions based on the latest information by acquiring current object data and meteorological data, thereby improving the accuracy of the prediction. By converting the object data and meteorological data, the input format can be optimized, making the model easier to understand and process, thereby improving the prediction performance. Using a pre-trained object prediction model can save training time and utilize existing data knowledge to improve prediction efficiency and accuracy. The model is trained based on samples of multiple historical generations, making full use of the information of historical data and improving the generalization ability of the model.

[0007] In an optional implementation, data conversion processing is performed on the object data and meteorological data within the current generation interval to obtain target processing data, including: data cleaning of the object data and meteorological data to obtain a cleaned first data set; normalizing the first data set to obtain a second data set; and format conversion processing of the second data set to obtain target processing data.

[0008] The data processing method provided in the embodiment of the present invention can remove noise and outliers by cleaning the object data and meteorological data, ensure the quality of the data, and thus improve the accuracy of subsequent analysis and prediction. Normalizing the data can eliminate the influence of different dimensions and ranges on model training, making the model easier to converge, thereby improving the stability of the prediction. Format conversion can ensure that the data is presented in a manner suitable for model input, improve the flexibility of data processing, and facilitate subsequent analysis and application. High-quality and standardized data input can help the model better learn the laws in the data, thereby improving the accuracy and robustness of the prediction.

[0009] In an optional embodiment, the training step of the object prediction model includes: obtaining the object quantity corresponding to the object data sample and the meteorological data sample; creating a model structure of the object prediction model based on a preset network structure; training the model structure using the object data sample, the meteorological data sample and the object quantity, and outputting the predicted object quantity through the trained object prediction model; iterating the object prediction model based on the error information between the object quantity and the predicted object quantity.

[0010] The data processing method provided by the embodiment of the present invention ensures that the model training is based on actual data and improves the relevance and accuracy of the prediction by obtaining the object quantity corresponding to the object data sample and the meteorological data sample. By using the object data sample, the meteorological data sample and the object quantity for training, a variety of factors are comprehensively considered to improve the adaptability of the model to complex situations. By analyzing the error information between the object quantity and the predicted object quantity, the model can be continuously iterated and optimized to improve the prediction accuracy and stability. The iterative training process allows the model to gradually learn and correct errors, thereby enhancing the learning ability and accuracy of the model.

[0011] In an optional implementation, the model structure is trained using object data samples, meteorological data samples, and object quantities, including: dividing the object data samples, meteorological data samples, and object quantities according to generation intervals and integrating them to obtain a sample data set; and training the model structure using the sample data set.

[0012] The data processing method provided by the embodiment of the present invention can better capture the characteristics of the time series and improve the sensitivity of the model to timeliness by dividing and integrating the object data samples, meteorological data samples and object quantities according to the generation interval. Integrating data samples from different sources helps to enrich the input information of the model, thereby improving the model's ability to learn complex relationships. The structuring and standardization of the sample data set makes the training process more efficient and helps the model reduce the interference of noise and redundant information during learning. By dividing according to the generation interval, it can be ensured that the model training is closer to the actual situation, thereby improving the reliability and accuracy of the prediction. The integrated sample data set is convenient for unified processing, reduces the complexity of data preprocessing, and improves work efficiency.

[0013] In an optional embodiment, the object prediction model is iterated based on the error information between the object quantity and the predicted object quantity, including: determining the error information between the object quantity and the predicted object quantity based on a preset loss function; based on the error information, updating at least one weight parameter of the object prediction model to iterate the object prediction model.

[0014] The data processing method provided by the embodiment of the present invention utilizes the error information between the object quantity and the predicted object quantity to effectively identify the deficiencies of the model, thereby making targeted improvements. The preset loss function provides a clear standard to measure the performance of the model, making the training and iteration of the model more systematic and controllable. By updating at least one weight parameter, the learning strategy of the model can be flexibly adjusted to improve the adaptability of the model. This error-based information feedback mechanism enables the model to be gradually optimized in each iteration, enhancing the learning ability and accuracy of the model. Through iterative optimization, the complexity of the model can be properly controlled to reduce the risk of overfitting, thereby improving the performance of the model on new data.

[0015] In an optional embodiment, at least one weight parameter of the object prediction model is updated based on the error information, including: determining the gradient information of at least one weight parameter based on the error information; determining the exponentially weighted average and the exponentially weighted square average corresponding to the gradient information using a preset optimization algorithm; performing bias correction on the exponentially weighted average and the exponentially weighted square average; and updating at least one weight parameter based on the corrected exponentially weighted average and the exponentially weighted square average.

[0016] The data processing method provided by the embodiment of the present invention can more accurately guide the learning direction of the model and improve the convergence speed by determining the gradient information of the weight parameters based on the error information. By using the combination of the exponential weighted average and the square average, the learning rate can be dynamically adjusted to adapt to different training stages and reduce fluctuations in the learning process. Bias correction of the exponential weighted average and the square average can improve the accuracy of the update, especially in the early stages of training, which helps to avoid the deviation of the gradient estimate. By comprehensively considering the historical gradient information, the stability of the parameter update can be improved, unnecessary shocks can be reduced, and a smoother convergence process can be achieved. Therefore, this method can effectively utilize historical information, accelerate the convergence speed of model learning, and save computing resources and time.

[0017] In an optional embodiment, the object quantity is converted into a first level value, and the predicted object quantity is converted into a second level value; based on the difference between the first level value and the second level value, the maximum error value is determined; based on the ratio between the difference and the maximum error value, the accuracy of the object prediction model is determined.

[0018] The data processing method provided by the embodiment of the present invention converts the object quantity and the predicted object quantity into grade values ​​respectively, so that data of different data types or scales can be compared under the same standard, thereby improving the comparability of the data. Through clear difference values ​​and maximum error values, the accuracy of the model prediction can be intuitively evaluated to help understand the performance of the model. Expressing the accuracy as a ratio makes the results easier to understand and interpret, and facilitates comparison with other models or methods. According to the change of the maximum error value, the accuracy evaluation of the model can reflect the prediction performance in different situations and provide dynamic feedback.

[0019] In an optional implementation, a mapping relationship between preset standard object data and standard occurrence degree levels is obtained; based on the mapping relationship, a target occurrence degree level corresponding to the target object data is determined; and corresponding prevention and control information is generated based on the target occurrence degree level.

[0020] The data processing method provided by the embodiment of the present invention can ensure the consistency and repeatability of the evaluation process by establishing a mapping relationship between standard object data and standard occurrence degree levels, which is conducive to unified understanding and application. The corresponding target occurrence degree level is determined according to the target object data, so that the prevention and control information can be accurately targeted at the actual situation and enhance the effectiveness of decision-making. Using existing standard data for mapping can improve the reliability and accuracy of the model and ensure that the prevention and control measures are based on data and facts. Through clear grade classification, the generation of prevention and control information becomes more systematic, which is convenient for relevant personnel to understand and implement. Therefore, this method can quickly generate corresponding prevention and control information, which is convenient for timely measures to reduce potential risks.

[0021] In a second aspect, the present invention provides a data processing device, comprising: an acquisition module, used to acquire current object data, the current object data including object data and meteorological data within a current generation interval; a processing module, used to perform data conversion processing on the object data and meteorological data within the current generation interval to obtain target processed data; a prediction module, used to input the target processed data into a pre-trained object prediction model, and output the target object data of the next generation interval through the object prediction model; wherein the object prediction model is trained based on object data samples and meteorological data samples within multiple historical generation intervals.

[0022] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the data processing method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0023] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the data processing method of the first aspect or any corresponding embodiment thereof.

[0024] In a fifth aspect, the present invention provides a computer program product, comprising computer instructions for causing a computer to execute the data processing method of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0026] Figure 1 is a flow chart of a data processing method according to an embodiment of the present invention;

[0027] Figure 2 is a flow chart of another data processing method according to an embodiment of the present invention;

[0028] Figure 3 is a flow chart of another data processing method according to an embodiment of the present invention;

[0029] Figure 4 is a structural block diagram of a data processing device according to an embodiment of the present invention;

[0030] Figure 5It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of the present invention.

[0032] Pests pose a serious threat to the ecology and economy, mainly in terms of agricultural damage, disease transmission and ecological imbalance. They not only directly lead to crop yield reduction, affecting food supply and farmers' income, but may also pose a threat to human health by spreading pathogens. In addition, the spread of pests will also destroy the ecological balance, affect biodiversity, and further aggravate environmental problems.

[0033] To solve this problem, related technologies predict the occurrence patterns of pests by constructing different prediction models, such as statistical models, ecological models, and traditional neural network models. Statistical models are based on historical data and statistical principles, and aim to predict the probability and trend of pest occurrence. Its prediction data mainly comes from field surveys, meteorological stations, satellite remote sensing and other channels, involving historical data of pests, climate data, soil data, and crop data. The linear relationship between pests and environmental factors is analyzed by linear regression models to predict the probability and trend of pest occurrence. However, linear regression models can only simply establish linear relationships and are suitable for pest prediction with simple and obvious relationships. For example, ecological models analyze climate factors that affect pest survival and reproduction, soil factors that affect crop growth, and biological factors that have complex relationships with pests, while considering human factors such as agricultural operations and pesticide use. The model integrates multiple factors and parameters and can comprehensively predict the occurrence trend and degree of pests. However, the acquisition of these parameters and factors requires a lot of manpower and material resources. Data collection and processing are the most challenging parts of model construction, which directly affect the subsequent model effect and prediction accuracy. For example, traditional neural network models have the characteristics of self-learning and self-adaptation, can effectively handle nonlinear relationships, and overcome the limitations of traditional quantitative prediction methods. However, such models have poor anti-interference ability, and the local gradient descent method used may lead to falling into the local optimal solution, thus affecting the prediction accuracy of the actual occurrence of pests.

[0034] In view of this, the technical solution of the present invention constructs a long short-term memory network (Long Short-Term MemoryNetwork, LSTM), which can capture long-term dependencies in time series data, which is crucial for predicting the trend of pest occurrence, because the occurrence of pests is greatly affected by historical data. The prediction model can update the weight parameters in real time, timely predict the occurrence of the next generation of insects, and automatically convert them into corresponding occurrence levels and prevention and control measures. The training data set of the model combines historical data, real-time reports and meteorological data for joint training. Over time, changes in real-time data will prompt the model to be retrained to update the weight parameters of the prediction model. With the expansion of the data set, LSTM will retain the more influential features according to the relevance and importance of the data features, thereby improving the prediction accuracy. Through this real-time update, the prediction of the next generation of pests will be more accurate.

[0035] According to an embodiment of the present invention, a data processing method embodiment is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0036] In this embodiment, a data processing method is provided, which can be used in computer devices, such as desktop computers, notebook computers, etc. Figure 1 is a flow chart of a data processing method according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0037] Step S101, obtaining current object data, the current object data including object data and meteorological data within the current generation interval.

[0038] Current object data refers to the data related to model training and prediction collected within the current generation interval, which mainly includes object data and meteorological data.

[0039] The current generation interval refers to the time range for observing and recording a specific object (such as migratory pests) within a certain period of time (for example, a few days to a few weeks). This time period can be divided according to factors such as the pest life cycle, migration pattern, crop growth cycle, etc.

[0040] Object data refers to data related to migratory pests, including insect quantity data and damage rate data. Specifically, the number of adult pests is collected through field monitoring equipment. These data can be recorded daily through methods such as light trapping, reflecting the number of pests in the current generation interval. The damage rate of pests on crops calculated by the stripping method represents the ratio between the number of leaf bracts eaten by pests and the total sample size (such as 20 or 50 plants). These data can help assess the impact of pests on crops.

[0041] Meteorological data include temperature, relative humidity and rainfall in the region. Specifically, meteorological data comes from a meteorological service unit (such as weather editor), and daily weather data within the current generation interval is obtained.

[0042] Specifically, the number of trapped pests is recorded regularly (for example, every day) by field monitoring equipment. These data can be automatically reported by the monitoring equipment or manually collected and reported to the data reporting system. By implementing the stripping method, the selected crop samples are inspected, the number of leaf bracts eaten by pests (N) and the total number of samples (M) are recorded, and the damage rate (P=N / M*100%) is calculated. The daily meteorological data for a specific time period in the area is obtained from a meteorological service unit (such as Xiangji Weather). The above-mentioned insect quantity data, damage rate data and meteorological data are integrated through the data reporting system interface to form complete current object data for subsequent model training and prediction.

[0043] Step S102, performing data conversion processing on the object data and meteorological data in the current generation interval to obtain target processed data.

[0044] The target processing data refers to the data format obtained after data conversion processing and suitable for input into the object prediction model. Specifically, the object data (such as insect quantity, damage rate, etc.) and meteorological data (such as temperature, humidity, precipitation, etc.) in the current generation interval are converted and processed, and then integrated into a data set to form a unified data structure. Among them, the conversion to the input format required by the model, such as matrix or tensor, is not limited here.

[0045] Step S103, input the target processing data into the pre-trained object prediction model, and output the target object data of the next generation interval through the object prediction model; wherein the object prediction model is trained based on object data samples and meteorological data samples in multiple historical generation intervals.

[0046] The pre-trained object prediction model is a deep learning model that aims to predict the future state of an object based on historical data (including object data and meteorological data). The model learns the relationship and pattern between data through training with a large amount of historical data, and can identify specific rules to predict the future state of an object. For example, it can be a long short-term memory network (LSTM), and the model structure mainly includes an input layer, an LSTM layer, a fully connected layer, and an output layer. Through the connection of these layers, the LSTM model can coherently and autonomously learn the mutual characteristics of data from the input data, capture the complex nonlinear relationship and sequence pattern relationship between the input and output data sets, and can be applied to more complex and cumbersome pest prediction situations. The core part of the LSTM model is the LSTM unit, and each LSTM unit contains three gate structures: input gate, forget gate, output gate, and a memory cell. The input gate controls the addition of new information, the forget gate controls whether the old information is forgotten, and the output gate controls whether the information in the memory cell should be output.

[0047] The target object data of the next generation sub-interval refers to the prediction of the status (such as insect quantity, damage rate, etc.) of a specific object (such as migratory pests) within the prediction time range. This data is the result of the model output.

[0048] Object data samples refer to actual data about a specific object collected in the past generation interval. These samples contain different features about the target object and are the basis for model training, such as insect quantity, damage rate, crop type, pesticide application records, etc. Meteorological data samples refer to meteorological condition data corresponding to the object data samples, which are used to provide important environmental factors that affect the state of the object, such as temperature, humidity, precipitation, wind speed, etc. Specifically, the target processing data is organized into the format required by the model, ensuring that the dimensions and feature names of the data match the training data of the model. The organized data is input into the model using programming languages ​​(such as Python) and machine learning libraries (such as TensorFlow, Keras, PyTorch, etc.). The model will generate prediction results based on the input data, which are the target object data (such as predicted insect quantity, damage rate, etc.) for the next generation interval.

[0049] The data processing method provided by the embodiment of the present invention ensures that the model makes predictions based on the latest information by acquiring current object data and meteorological data, thereby improving the accuracy of the prediction. By converting the object data and meteorological data, the input format can be optimized, making the model easier to understand and process, thereby improving the prediction performance. Using a pre-trained object prediction model can save training time and utilize existing data knowledge to improve prediction efficiency and accuracy. The model is trained based on samples of multiple historical generations, making full use of the information of historical data and improving the generalization ability of the model.

[0050] In this embodiment, a data processing method is provided, which can be used in computer devices, such as desktop computers, notebook computers, etc. Figure 2 is a flow chart of a data processing method according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0051] Step S201, obtain current object data, which includes object data and meteorological data within the current generation interval. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0052] Step S202, performing data conversion processing on the object data and meteorological data in the current generation interval to obtain target processed data.

[0053] Specifically, the above step S202 includes:

[0054] Step S2021, performing data cleaning on the object data and the meteorological data to obtain a cleaned first data set.

[0055] The first data set refers to the data set of object data and meteorological data after data cleaning. Specifically, which data are outliers are defined according to business logic or statistical methods. For example, the standard deviation method is used to calculate the mean and standard deviation of the data, and data exceeding the mean ± 2 times the standard deviation is eliminated. Another example is the quartile method is used to calculate the quartile of the data, and data below Q-1.5×(Q3-Q1) and above Q3+1.5×(Q3-Q1) are eliminated. Outlier detection is performed on the object data and meteorological data respectively, and the above method is applied to eliminate the outliers. Check whether there are null values ​​in the data set. For numerical data, the mean, median, and mode are used to fill the missing values. For time series data, the previous and next values ​​are used to fill the missing values. If the number of missing values ​​is small, the records containing missing values ​​can be eliminated. Ensure that all data units are consistent, such as temperature in degrees Celsius and precipitation in millimeters. If there are data in different units, they need to be converted to ensure uniformity. Check whether there are duplicate records in the data set. If so, keep one record and eliminate other duplicate entries. Convert object data and meteorological data into appropriate data types (such as strings, integers, and floating-point numbers), and unify column names into standard formats such as lowercase and without spaces to facilitate subsequent operations.

[0056] Step S2022: normalize the first data set to obtain a second data set.

[0057] The second data set refers to a data set that has been normalized. The purpose of this process is to adjust the values ​​of different features in the first data set to the same dimension, eliminate the impact caused by different dimensions, and thus improve the training effect and accuracy of the model. For example, the values ​​of all features are converted to a relatively uniform range (such as between 0 and 1) so that the model is easier to converge during training. Specifically, the minimum-maximum normalization method can be used to calculate the minimum and maximum values ​​for each feature (such as the number of pests, yield, temperature, humidity, etc.). The minimum-maximum normalization formula is applied to each eigenvalue to convert it into a normalized value. Furthermore, the Z-score normalization method can also be used to calculate the mean and standard deviation for each feature, and each eigenvalue is converted using the Z-score normalization formula.

[0058] Step S2023, performing format conversion processing on the second data set to obtain target processed data.

[0059] Format conversion processing refers to the necessary conversion of the data structure or data type in the second data set in order to prepare for subsequent data analysis or model training. Specifically, ensure that the data type of each feature is consistent with its intended use. For example: numerical features (such as the number of pests, temperature, humidity, etc.) should be converted to floating point numbers or integers; categorical features (such as crop type, growth stage, etc.) should be converted to categorical or character data. Perform one-hot encoding on categorical feature variables so that they can be converted into numerical features suitable for model training. After the data processing is completed, ensure that the numerical feature data and the categorical feature variable data after one-hot encoding are sorted according to certain rules, and reset the index for subsequent processing.

[0060] Step S203, input the target processing data into the pre-trained object prediction model, and output the target object data of the next generation interval through the object prediction model; wherein the object prediction model is trained based on object data samples and meteorological data samples in multiple historical generation intervals. For details, please refer to Figure 1 Step S103 of the illustrated embodiment will not be described in detail here.

[0061] The data processing method provided in the embodiment of the present invention can remove noise and outliers by cleaning the object data and meteorological data, ensure the quality of the data, and thus improve the accuracy of subsequent analysis and prediction. Normalizing the data can eliminate the influence of different dimensions and ranges on model training, making the model easier to converge, thereby improving the stability of the prediction. Format conversion can ensure that the data is presented in a manner suitable for model input, improve the flexibility of data processing, and facilitate subsequent analysis and application. High-quality and standardized data input can help the model better learn the laws in the data, thereby improving the accuracy and robustness of the prediction.

[0062] In this embodiment, a data processing method is provided, which can be used in computer devices, such as desktop computers, notebook computers, etc. Figure 3 is a flow chart of a data processing method according to an embodiment of the present invention. Figure 3 As shown, the process includes the following steps:

[0063] Step S301, obtain current object data, which includes object data and meteorological data within the current generation interval. Figure 2 Step S201 of the illustrated embodiment will not be described in detail here.

[0064] Step S302: Perform data conversion processing on the object data and meteorological data in the current generation interval to obtain target processed data. Figure 2 Step S202 of the illustrated embodiment will not be described in detail here.

[0065] Step S303, input the target processing data into the pre-trained object prediction model, and output the target object data of the next generation interval through the object prediction model; wherein the object prediction model is trained based on object data samples and meteorological data samples in multiple historical generation intervals.

[0066] Specifically, the training steps of the object prediction model include:

[0067] Step a1, obtaining object quantities corresponding to object data samples and meteorological data samples.

[0068] The number of objects refers to the number of individuals of a specific object (such as migratory pests) actually captured or observed in a specific area and time. Specifically, the designated area can be monitored at specific time intervals (such as daily or weekly) to record the number of migratory pests, or the historical number of pests and corresponding meteorological data can be obtained from a professional database, which is not limited here.

[0069] Step a2: creating a model structure of an object prediction model based on a preset network structure.

[0070] The preset network structure refers to a deep learning model architecture that has been designed and is suitable for a specific task, such as a long short-term memory network (LSTM). The model structure of the object prediction model refers to the specific composition of the deep learning model designed for predicting the number of migratory pests. For example, after selecting LSTM as the network structure, the input layer receives object data samples and meteorological data samples, and 1-2 LSTM hidden layers extract time series features. The output layer is a neuron that predicts the number of pests at a certain moment in the future.

[0071] Step a3, using the object data samples, meteorological data samples and object quantity to train the model structure, and output the predicted object quantity through the trained object prediction model.

[0072] The predicted object quantity refers to the number of specific objects (such as migratory pests) predicted by the model, which is a quantitative prediction based on the input object data sample and meteorological data sample information. Specifically, the object data samples, meteorological data samples and object quantity are divided into a training set and a test set. For example, 90% of the data is used for training, and the remaining data is used to evaluate the performance of the model. After the object data samples, meteorological data samples and object quantity are integrated, a training set is formed, and the model is trained using the training set. The trained object prediction model outputs the predicted object quantity.

[0073] In some optional implementations, the above step a3 includes:

[0074] Step b1, dividing the object data samples, meteorological data samples and object quantities according to the generation intervals and integrating them to obtain a sample data set.

[0075] The sample data set is a data set containing multiple features related to the pest quantity. The features come from the object data samples, meteorological data samples and the object quantity. Specifically, after the object data samples, meteorological data samples and the object quantity are preprocessed and cleaned, a sample data set S is formed. For any data S i , containing the input data x i And the output data y i . Input data x i The output data y is composed of object data samples within a generation interval (such as the number of migratory pests) and meteorological data samples within the corresponding generation time interval (such as weather data, temperature, humidity, rainfall, etc.). i It is the object quantity within the next generation interval (the actual number of migratory pests determined in the next generation).

[0076] Step b2: Use the sample data set to train the model structure.

[0077] Specifically, the input data x i Passed to the selected model, the model will be based on these input data x i And the corresponding output data y i Learn the relationship between features and target variables to train the model structure.

[0078] In the above implementation, by dividing and integrating the object data samples, meteorological data samples and object quantities according to the generation interval, the characteristics of the time series can be better captured and the sensitivity of the model to timeliness can be improved. Integrating data samples from different sources helps to enrich the input information of the model, thereby improving the model's ability to learn complex relationships. The structuring and standardization of the sample data set makes the training process more efficient and helps the model reduce the interference of noise and redundant information during learning. By dividing according to the generation interval, it can be ensured that the model training is closer to the actual situation, thereby improving the reliability and accuracy of the prediction. The integrated sample data set is convenient for unified processing, which reduces the complexity of data preprocessing and improves work efficiency.

[0079] Step a4, iterating the object prediction model based on the error information between the object quantity and the predicted object quantity.

[0080] Error information refers to the difference between the actual observed amount of an object and the amount of the object predicted by the model. By analyzing this difference, the prediction model can be improved and optimized to increase its accuracy and reliability. Specifically, the error information is analyzed, the parameters of the object prediction model are adjusted based on the analysis results, and the optimized parameters are used to retrain to improve the prediction accuracy.

[0081] In some optional implementations, the above step a4 includes:

[0082] Step c1, using a preset loss function, determines the error information between the object quantity and the predicted object quantity.

[0083] The loss function is a tool used to measure the gap between the model's predicted value and the true value. For example, the mean squared error (MSE) can be used as the loss function, and the formula is:

[0084]

[0085] Among them, n is the number of samples, y i Represents the correct numerical result of the i-th sample, y i ' is the predicted value output by the prediction model. The smaller the value of the loss function, the better the prediction performance of the model.

[0086] By calculating the value of the loss function, we can get the prediction error of the model under the current weight parameters. This error information is the basis for subsequent weight updates. It can tell the model how accurate the current prediction is and how to adjust the parameters to improve the prediction.

[0087] Step c2: based on the error information, updating at least one weight parameter of the object prediction model to iterate the object prediction model.

[0088] In a neural network, the output of the model is determined by the input data and the weight parameters. In order to make the model's predictions more accurate, these weight parameters need to be adjusted based on the error information. The goal of weight updating is to minimize the loss function, thereby improving the accuracy of the prediction. Specifically, this process is an iterative process that is repeated for multiple cycles (epochs), and each cycle optimizes the model by calculating the loss function and updating the weight parameters. The ultimate goal is to gradually reduce the value of the loss function and gradually improve the prediction performance of the model.

[0089] In the above implementation, the error information between the object quantity and the predicted object quantity can be used to effectively identify the shortcomings of the model, so as to make targeted improvements. The preset loss function provides a clear standard to measure the performance of the model, making the training and iteration of the model more systematic and controllable. By updating at least one weight parameter, the learning strategy of the model can be flexibly adjusted to improve the adaptability of the model. This error-based information feedback mechanism enables the model to be gradually optimized in each iteration, enhancing the learning ability and accuracy of the model. Through iterative optimization, the complexity of the model can be properly controlled to reduce the risk of overfitting, thereby improving the performance of the model on new data.

[0090] In some optional implementations, the above step c2 includes:

[0091] Step d1, determining gradient information of at least one weight parameter based on the error information.

[0092] Gradient information represents the derivative of the loss function with respect to the weight parameter, which indicates the direction and rate of change of the loss function under the current parameter settings. Specifically, the chain rule can be used to calculate the gradient of the loss function with respect to the weight parameter, and the gradient of each weight can be obtained by combining the relationship between the model output and the weight.

[0093] Step d2, using a preset optimization algorithm, determines the exponentially weighted average value and the exponentially weighted square average value corresponding to the gradient information.

[0094] The preset optimization algorithm can be selected as the Adam optimization algorithm, which uses adaptive moment estimation to calculate the exponentially weighted average m_ of the gradient of each parameter. t , and the exponentially weighted squared average v_ of these gradients t , specifically, the calculation formula is as follows:

[0095] m_ t =β_1×m_(t-1)+(1-β_1)×g

[0096] v_ t =β_2×v_(t-1)+(1-β_2)×g 2

[0097] Among them, g is the current gradient, β_1, β_2 are hyperparameters.

[0098] Step d3, performing bias correction on the exponentially weighted average and the exponentially weighted square average.

[0099] Adjust the initial m_ t and v_ t To eliminate the deviation. Specifically, due to m_ t and v_ t In the initial stage, it will be affected by 0, and the correction formula is as follows:

[0100]

[0101] Step d4: updating at least one weight parameter based on the modified exponentially weighted average and the exponentially weighted square average.

[0102] Update the weight parameters to minimize the loss function. Specifically, the weight update formula is:

[0103] w_ t =w_ (t-1) –η×m_ t ' / (v_ t '+ε)(1 / 2)

[0104] Among them, w_ t is the current updated weight, w_ (t-1) is the weight at the previous moment, η is the learning rate, and ε is a small positive number used to avoid the denominator being zero.

[0105] In the above implementation, by determining the gradient information of the weight parameters based on the error information, the learning direction of the model can be guided more accurately and the convergence speed can be improved. By using the combination of the exponential weighted average and the square average, the learning rate can be dynamically adjusted to adapt to different training stages and reduce fluctuations in the learning process. Bias correction of the exponential weighted average and the square average can improve the accuracy of the update, especially in the early stages of training, which helps to avoid the deviation of the gradient estimate. By comprehensively considering the historical gradient information, the stability of the parameter update can be improved, unnecessary shocks can be reduced, and a smoother convergence process can be achieved. Therefore, this method can effectively utilize historical information, accelerate the convergence speed of model learning, and save computing resources and time.

[0106] Step a5, converting the object quantity into a first level value, and converting the predicted object quantity into a second level value.

[0107] First level value (F i ) is based on the actual observed pest population y i The converted level value can be defined according to a preset standard (such as insect quantity range). For example, 0-10 is level 1, 11-50 is level 2, 51-100 is level 3, and more than 100 is level 4.

[0108] Second level value (A i ) is the number of pests y predicted by the model i 'The converted grade values ​​use the same standard.

[0109] For example, if the actual number of insects is 30 and the predicted number of insects is 45, then the first level value F i is 2, the second level value is A i is 2.

[0110] Step a6, determining a maximum error value based on the difference between the first level value and the second level value.

[0111] Calculate the difference between the actual grade and the predicted grade to analyze the performance of the model. Specifically, difference = |F i -A i ∣, if the first level value (F i ) and the second level value (A i ) are the same, the difference is 0; if they are different, the difference is Fi and A i The absolute difference between .

[0112] Maximum error value (M i ) refers to the maximum difference calculated across all samples, which helps assess the worst-case performance of the model. For example, if the maximum difference in a series of predictions is 3, then the maximum error value M i is 3.

[0113] Step a7, determining the accuracy of the object prediction model based on the ratio between the difference value and the maximum error value.

[0114] The performance of the object prediction model is evaluated by calculating the accuracy (R). Specifically, the accuracy R can be calculated by the following formula:

[0115]

[0116] If the value of R is close to 1, it means that the prediction accuracy of the model is high and the difference between the actual result and the predicted result is small. If the value of R is close to 0, it means that the model prediction is inaccurate and the difference between the actual result and the predicted result is large.

[0117] In the above implementation, the object quantity and the predicted object quantity are converted into grade values ​​respectively, so that data of different data types or scales can be compared under the same standard, thereby improving the comparability of the data. Through clear difference values ​​and maximum error values, the accuracy of the model prediction can be intuitively evaluated to help understand the performance of the model. Expressing the accuracy as a ratio makes the results easier to understand and interpret, and facilitates comparison with other models or methods. According to the change of the maximum error value, the accuracy evaluation of the model can reflect the prediction performance in different situations and provide dynamic feedback.

[0118] Step S304: obtaining a mapping relationship between preset standard object data and standard occurrence degree levels.

[0119] Standard object data refers to specific values ​​of pest abundance or pest occurrence for specific pest species, collected based on historical monitoring, research and scientific data, which are used to define the characteristics and patterns of pest occurrence. Specifically, standard object data may include pest species, pest abundance data and historical records.

[0120] The standard occurrence level is a system for classifying the number of pests, which is used to assess the potential threat posed by pests to crops. These levels help agricultural workers quickly understand the severity of pests and take appropriate prevention and control measures. Specifically, each level can be divided according to the size of the pests. For example, level 1 means that the number of pests is very small and basically has no impact on crops; level 2 means that the number of pests is small and has a slight impact; level 3 means that the number of pests is moderate and has a certain impact on crops; level 4 means that the number of pests is large and has a serious impact on crops; level 5 means that the number of pests is extremely large and causes large-scale losses. Among them, the standard occurrence level can be formulated using existing standardized assessment methods, such as pest monitoring and assessment standards formulated by national or regional agricultural departments, or by referring to research data and suggestions from scientific research institutions, especially the occurrence patterns of pests under specific climatic conditions, to formulate standard occurrence levels, which are not limited here.

[0121] Specifically, the standard object data and the corresponding occurrence levels are sorted to form a systematic database. The insect quantity data of each pest should be marked with the corresponding occurrence level. The standard object data (insect quantity) is input into the system, and the mapping rules between insect quantity and occurrence level are defined according to the existing evaluation standards. Through the verification of historical data, check whether the mapping relationship is reasonable and make adjustments if necessary.

[0122] Taking rice leaf roller as an example, the corresponding relationship between the pest population and the occurrence level is as follows: when the pest population (unit: head / hundred bushes) is less than 30, it corresponds to level 1; when the pest population is between 30 and 100, it corresponds to level 2; when the pest population is between 101 and 200, it corresponds to level 3; when the pest population is between 201 and 300, it corresponds to level 4; and when the pest population is greater than 300, it corresponds to level 5.

[0123] Taking the corn pest fall armyworm as an example, the corresponding relationship between the insect population and the occurrence level is: when the insect population is between 0.1 and 5, it corresponds to level 1; when the insect population is between 5 and 10, it corresponds to level 2; when the insect population is between 10 and 30, it corresponds to level 3; when the insect population is between 30 and 80, it corresponds to level 4; and when the insect population is greater than 80, it corresponds to level 5.

[0124] Step S305: determining the target occurrence degree level corresponding to the target object data according to the mapping relationship.

[0125] The target occurrence level refers to the level of severity of the pest occurrence corresponding to it determined by analyzing the target object data based on a mapping relationship established in advance. Specifically, the collected target object data is compared with the standard object data using the mapping relationship established previously. The mapping relationship can be, for example, a table or rule that clarifies the correspondence between the number of insects and the level of occurrence. According to the insect quantity data of the target object, the corresponding level is searched in the mapping relationship. For example: if the number of insects in the target object data is 25, and the mapping relationship is searched to find that "the number of insects less than 30 corresponds to the occurrence level 1", then the target occurrence level is level 1. The determined target occurrence level is recorded for subsequent pest monitoring, early warning, and formulation of prevention and control strategies.

[0126] Step S306, generating corresponding prevention and control information according to the target occurrence level.

[0127] Prevention and control information refers to management strategies and measures formulated to deal with a specific level of pest occurrence. The generation of this information mainly depends on the previously determined target level of occurrence. Among them, prevention and control information can include prevention and control measures, early warning information, and monitoring plans. Specifically, a prevention and control information database is established in advance, which contains prevention and control measures and suggestions corresponding to different target levels of occurrence. This database can be established based on historical data, expert opinions, or scientific research results. For each level of occurrence, corresponding prevention and control measures are defined. For example, for level 1 (mild pests), manual capture or simple ecological control measures can be selected; for level 2 (mild pests), it is recommended to use biological pesticides and monitor pest changes; for level 3 (moderate pests), comprehensive prevention and control is required, and the use of chemical agents may be required; for level 4 (serious pests), it is strongly recommended to use chemical agents in an all-round way, conduct field monitoring, and adjust planting strategies when necessary. When the target level of occurrence is determined, the system queries the corresponding prevention and control measures from the prevention and control information database according to the level. Customize the queried prevention and control information according to the actual situation (such as weather, soil conditions, crop types, etc.). For example, if the pest occurs in the rainy season, it may be necessary to adjust the selection and application time of the agent. The queried and adjusted prevention and control measures will be compiled into a report, which may include the specific content of the prevention and control measures, the time and frequency of application, and suggestions for monitoring and evaluation, etc. The prevention and control information will be sent to relevant agricultural workers, farmers or managers to ensure that they can receive prevention and control measures and suggestions in a timely manner.

[0128] Furthermore, all data (insect quantity, occurrence level, and control measures) can be visualized and displayed on a pre-established Web platform. Users can view real-time data, historical trends, and control recommendations for different types of pests on the Web platform, and provide feedback or adjust subsequent management strategies.

[0129] The data processing method provided by the embodiment of the present invention can ensure the consistency and repeatability of the evaluation process by establishing a mapping relationship between standard object data and standard occurrence degree levels, which is conducive to unified understanding and application. The corresponding target occurrence degree level is determined according to the target object data, so that the prevention and control information can be accurately targeted at the actual situation and enhance the effectiveness of decision-making. Using existing standard data for mapping can improve the reliability and accuracy of the model and ensure that the prevention and control measures are based on data and facts. Through clear grade classification, the generation of prevention and control information becomes more systematic, which is convenient for relevant personnel to understand and implement. Therefore, this method can quickly generate corresponding prevention and control information, which is convenient for timely measures to reduce potential risks.

[0130] In the present embodiment, a data processing device is also provided, which is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0131] This embodiment provides a data processing device, such as Figure 4 As shown, including:

[0132] The first acquisition module 401 is used to acquire current object data, which includes object data and meteorological data within a current generation interval;

[0133] The processing module 402 is used to perform data conversion processing on the object data and meteorological data in the current generation interval to obtain target processing data;

[0134] The prediction module 403 is used to input the target processing data into the pre-trained object prediction model, and output the target object data of the next generation interval through the object prediction model; wherein the object prediction model is trained based on object data samples and meteorological data samples in multiple historical generation intervals.

[0135] In some optional implementations, the processing module 402 includes:

[0136] A cleaning submodule, used for cleaning the object data and the meteorological data to obtain a cleaned first data set;

[0137] A normalization submodule, used to perform normalization processing on the first data set to obtain a second data set;

[0138] The format conversion submodule is used to perform format conversion processing on the second data set to obtain target processed data.

[0139] In some optional implementations, the prediction module 403 includes:

[0140] An acquisition submodule, used to acquire object quantities corresponding to object data samples and meteorological data samples;

[0141] Creating a submodule for creating a model structure of an object prediction model based on a preset network structure;

[0142] A training submodule is used to train the model structure using object data samples, meteorological data samples and object quantity, and output the predicted object quantity through the trained object prediction model;

[0143] The iterative submodule is used to iterate the object prediction model based on the error information between the object quantity and the predicted object quantity.

[0144] In some optional embodiments, the training submodule includes:

[0145] An integration unit, used for integrating the object data samples, the meteorological data samples and the object quantity according to the generation intervals to obtain a sample data set;

[0146] The training unit is used to train the model structure using the sample data set.

[0147] In some optional implementations, the iterator module includes:

[0148] A first determining unit, used to determine error information between the object quantity and the predicted object quantity by using a preset loss function;

[0149] The first updating unit is used to update at least one weight parameter of the object prediction model based on the error information to iterate the object prediction model.

[0150] In some optional implementations, the first updating unit includes:

[0151] A second determining unit, configured to determine gradient information of at least one weight parameter based on the error information;

[0152] A third determining unit, used to determine an exponentially weighted average value and an exponentially weighted square average value corresponding to the gradient information by using a preset optimization algorithm;

[0153] A correction unit, used for performing bias correction on the exponentially weighted average and the exponentially weighted square average;

[0154] The second updating unit is used to update at least one weight parameter based on the modified exponentially weighted average value and the exponentially weighted square average value.

[0155] In some optional implementations, the prediction module 403 further includes:

[0156] a conversion submodule, for converting the object quantity into a first level value and converting the predicted object quantity into a second level value;

[0157] A first determination submodule, configured to determine a maximum error value based on a difference between the first level value and the second level value;

[0158] The second determination submodule is used to determine the accuracy of the object prediction model based on the ratio between the difference value and the maximum error value.

[0159] In some optional implementations, the data processing device further includes:

[0160] A second acquisition module is used to acquire a mapping relationship between preset standard object data and standard occurrence degree level;

[0161] A determination module, used to determine the target occurrence degree level corresponding to the target object data according to the mapping relationship;

[0162] The generation module is used to generate corresponding prevention and control information according to the target occurrence level.

[0163] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0164] The data processing device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0165] The data processing device provided by the embodiment of the present invention obtains current object data and meteorological data to ensure that the model makes predictions based on the latest information, thereby improving the accuracy of the predictions. By converting the object data and meteorological data, the input format can be optimized, making the model easier to understand and process, thereby improving the prediction performance. Using a pre-trained object prediction model can save training time and utilize existing data knowledge to improve prediction efficiency and accuracy. The model is trained based on samples of multiple historical generations, making full use of the information of historical data and improving the generalization ability of the model.

[0166] The embodiment of the present invention also provides a computer device having the above Figure 4 The data processing device shown.

[0167] See also Figure 5 , Figure 5 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 5 As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 5 A processor 10 is taken as an example.

[0168] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0169] The memory 20 stores instructions executable by at least one processor 10, so that the at least one processor 10 executes the method shown in the above embodiment.

[0170] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0171] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0172] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 5 The example of connecting through bus is taken in the following.

[0173] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0174] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0175] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0176] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A data processing method, characterized in that: The method comprises: Acquire current object data, wherein the current object data includes object data and meteorological data within a current generation interval; Performing data conversion processing on the object data and meteorological data within the current generation interval to obtain target processed data; Inputting the target processing data into a pre-trained object prediction model, and outputting target object data of the next generation sub-interval through the object prediction model; The object prediction model is obtained by training based on object data samples and meteorological data samples within multiple historical generation intervals.

2. The method according to claim 1, characterized in that: The data conversion process is performed on the object data and the meteorological data within the current generation interval to obtain the target processed data, including: Performing data cleaning on the object data and the meteorological data to obtain a cleaned first data set; Normalizing the first data set to obtain a second data set; The second data set is format converted to obtain the target processed data.

3. The method according to claim 1, characterized in that The training steps of the object prediction model include: Acquire the object quantities corresponding to the object data sample and the meteorological data sample; Creating a model structure of the object prediction model based on a preset network structure; The model structure is trained using the object data sample, the meteorological data sample and the object quantity, and the predicted object quantity is outputted through the trained object prediction model; The object prediction model is iterated based on error information between the object quantity and the predicted object quantity.

4. The method according to claim 3, characterized in that The training of the model structure by using the object data sample, the meteorological data sample and the object quantity includes: The object data samples, the meteorological data samples and the object quantity are divided according to generation intervals and then integrated to obtain a sample data set; The model structure is trained using the sample data set.

5. The method according to claim 3, characterized in that: The iterating the object prediction model based on the error information between the object quantity and the predicted object quantity includes: Determine the error information between the object quantity and the predicted object quantity by using a preset loss function; Based on the error information, at least one weight parameter of the object prediction model is updated to iterate the object prediction model.

6. The method according to claim 5, characterized in that The updating of at least one weight parameter of the object prediction model based on the error information includes: Based on the error information, determining gradient information of the at least one weight parameter; Determine the exponentially weighted average value and the exponentially weighted square average value corresponding to the gradient information by using a preset optimization algorithm; Performing bias correction on the exponentially weighted average and the exponentially weighted square average; The at least one weight parameter is updated based on the modified exponentially weighted average value and the exponentially weighted square average value.

7. The method according to any one of claims 3 to 6, characterized in that Also includes: converting the object quantity into a first level value and converting the predicted object quantity into a second level value; determining a maximum error value based on a difference between the first level value and the second level value; Based on the ratio between the difference value and the maximum error value, the accuracy of the object prediction model is determined.

8. The method according to claim 1, characterized in that: Also includes: Obtaining the mapping relationship between the preset standard object data and the standard occurrence degree level; Determining, according to the mapping relationship, a target occurrence degree level corresponding to the target object data; Generate corresponding prevention and control information according to the target occurrence level.

9. A data processing device, characterized in that: The device comprises: An acquisition module, used to acquire current object data, wherein the current object data includes object data and meteorological data within a current generation interval; A processing module, used for performing data conversion processing on the object data and meteorological data within the current generation interval to obtain target processed data; A prediction module, used for inputting the target processing data into a pre-trained object prediction model, and outputting target object data of the next generation sub-interval through the object prediction model; The object prediction model is obtained by training based on object data samples and meteorological data samples within multiple historical generation intervals.

10. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the data processing method according to any one of claims 1 to 8 by executing the computer instructions.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the data processing method according to any one of claims 1 to 8.

12. A computer program product, characterized in that The method comprises computer instructions for causing a computer to execute the data processing method according to any one of claims 1 to 8.

Images