A Method and System for Optimizing Rainy Season-Suitable Planting Configuration for Winter Wheat Based on Drought Year Type Identification

Abstract

This invention relates to the field of agricultural planting technology and discloses a method and system for optimizing the planting configuration of winter wheat based on drought year type identification. The method involves constructing a drought year type identification index system, collecting data on each drought year type identification index in real time within the target planting area, preprocessing the collected data, and inputting the preprocessed data into a drought year type identification model to identify the current drought year type in the target planting area. Based on the identified drought year type, combined with historical rainfall data and soil characteristics of the target planting area, the optimal planting configuration parameters for winter wheat are determined. Real-time monitoring of each parameter information is achieved using sensors, and the planting configuration parameters are adjusted based on the monitoring data to ensure optimal growth conditions for winter wheat under different drought year types. This invention can accurately identify different drought year types, providing a reliable basis for optimizing the planting configuration of winter wheat under suitable rainfall conditions.

Description

Technical Field

[0001] This invention relates to the field of agricultural planting technology, specifically to a method and system for optimizing the planting configuration of winter wheat based on drought year type identification. Background Technology

[0002] Winter wheat is an important food crop, and its yield and quality are significantly affected by climatic conditions, especially drought. Under different drought years, traditional winter wheat planting configurations often fail to make full use of rainfall resources, resulting in water waste and increased irrigation costs. At the same time, the yield and quality of winter wheat cannot be effectively guaranteed. Existing planting configurations lack accurate identification of drought years and targeted optimization strategies, and cannot reasonably adjust planting parameters such as sowing time, planting density, and irrigation plans according to the actual drought situation, making it difficult to achieve the goals of efficient water conservation and high yield and quality in winter wheat planting. Summary of the Invention

[0003] The purpose of this invention is to solve the above-mentioned problems by designing a method and system for optimizing the planting configuration of winter wheat based on drought year type identification.

[0004] The first aspect of this invention provides a method for optimizing the planting configuration of winter wheat based on drought year type identification, the method comprising the following steps:

[0005] A drought year type identification index system was constructed, and data of each drought year type identification index were collected in real time in the target planting area. The collected data were preprocessed to obtain preprocessed data.

[0006] The preprocessed data is input into the drought year type identification model, which is used to identify the current drought year type in the target planting area.

[0007] Based on the identified drought year type, combined with historical rainfall data and soil characteristics of the target planting area, the appropriate planting configuration parameters for winter wheat in rainy conditions are determined.

[0008] During the planting process, sensors are used to monitor various parameters in real time to obtain monitoring data. Based on the monitoring data, the planting configuration parameters are adjusted to ensure that winter wheat obtains the best growth conditions in different drought years.

[0009] Optionally, in a first implementation of the first aspect of the present invention, the drought year type identification index system includes at least precipitation, evaporation, soil moisture content, groundwater level, and meteorological drought index.

[0010] Optionally, in the second implementation of the first aspect of the present invention, the step of constructing a drought year type identification index system, collecting data of each drought year type identification index in real time within the target planting area, and preprocessing the collected data to obtain preprocessed data includes:

[0011] The drought year type identification index data collected in the target planting area are arranged according to time series and spatial location to form a three-dimensional spatiotemporal matrix;

[0012] A weight is initialized for each element in the three-dimensional spatiotemporal matrix. For each spatial location and index in the three-dimensional spatiotemporal matrix, the data sequence of the location and index at different time points is analyzed. By calculating the characteristics of the data in the time series, the attention weight in the time dimension is obtained.

[0013] For each time point and indicator, attention weights in the spatial dimension are calculated by analyzing the distance between spatial locations and the correlation of data.

[0014] By fusing the attention weights of the time dimension and the attention weights of the spatial dimension, a comprehensive spatiotemporal attention weight is obtained for each data element;

[0015] For locations with missing data, the surrounding data is weighted and summed based on the comprehensive spatiotemporal attention weights of the surrounding data elements to calculate the interpolation result of the missing data, thus completing the processing of the missing data.

[0016] The calculated interpolation results are used to replace the original missing data positions in the three-dimensional spatiotemporal matrix to obtain the spatiotemporal matrix after processing by the ST-Impute algorithm, thus obtaining the preprocessed data.

[0017] Optionally, in a third implementation of the first aspect of the present invention, the step of inputting the preprocessed data into a drought year type identification model, and identifying the current drought year type of the target planting area through the drought year type identification model, includes:

[0018] A CNN-BiLSTM-Attention spatiotemporal feature extraction module was built, domain adaptation was introduced, and self-supervised pre-training and semi-supervised learning were combined to train a drought year type recognition model.

[0019] The preprocessed data is input into the drought year type identification model. The model extracts the spatiotemporal features of the data through the CNN-BiLSTM-Attention spatiotemporal feature extraction module, and then performs domain-adaptive feature representation. Combined with the model parameters optimized by self-supervised and semi-supervised learning, the model outputs the identification result of the current drought year type of the target planting area.

[0020] Optionally, in the fourth implementation of the first aspect of the present invention, the construction of the CNN-BiLSTM-Attention spatiotemporal feature extraction module, the introduction of domain adaptation, and the combination of self-supervised pre-training and semi-supervised learning to train a drought year type recognition model include:

[0021] The training samples used for model training are organized according to the time and space dimensions to form a multidimensional data matrix. In the CNN layer, multiple convolutional kernels of different sizes and numbers are set. The data is extracted through convolution operations to capture local features of the data in the spatial dimension. Pooling operations are used to reduce the data dimensionality.

[0022] The feature vectors output by the CNN layer are input into the BiLSTM layer. The BiLSTM uses a gating mechanism to process time series data from both the forward and backward directions, learn the long-term dependencies of the data in the time dimension, and extract the dynamic change features between data at different time points.

[0023] Using the output of the BiLSTM layer as input, the attention mechanism layer automatically focuses on the spatiotemporal features related to drought year type identification by calculating the weights of different features.

[0024] The training samples used for model training are taken as the source domain and the target planting region is taken as the target domain. An adversarial training approach is adopted, and a domain discriminator is added to the model. During the training process, the CNN-BiLSTM-Attention spatiotemporal feature extraction module learns to extract feature representations that are insensitive to the domain, so that the data of the source domain and the target domain are aligned in the feature space.

[0025] Optionally, in the fifth implementation of the first aspect of the present invention, the step of constructing the CNN-BiLSTM-Attention spatiotemporal feature extraction module, introducing domain adaptation, and combining self-supervised pre-training and semi-supervised learning to train a drought year type recognition model further includes:

[0026] Pre-training is performed using a masked autoencoder approach. The input training samples are randomly masked, and the masked data is then fed into the CNN-BiLSTM-Attention spatiotemporal feature extraction module. The model reconstructs the original unmasked data, learns the latent feature representation of the data, and optimizes the model parameters. After pre-training, a small amount of labeled data and a large amount of unlabeled data are input into the model together. A supervised loss function is calculated using the labeled data, while an unsupervised loss function is calculated using the unlabeled data. The two loss functions are then weighted and summed to optimize the model parameters.

[0027] Optionally, in a sixth implementation of the first aspect of the present invention, determining the appropriate rainfall-suitable planting configuration parameters for winter wheat based on the identified drought year type and in conjunction with historical rainfall data and soil characteristics of the target planting area includes:

[0028] A certain number of gray wolf individuals are randomly generated in the feasible solution space to form an initial population. Each gray wolf individual represents a set of winter wheat planting configuration parameters.

[0029] Calculate the fitness value of each gray wolf individual and divide the gray wolf individuals into four levels: α, β, δ and ω according to the fitness value. The α wolf is the optimal solution, the β wolf and δ wolf assist the α wolf in making decisions, and the ω wolf is the poor solution.

[0030] α, β, and δ wolves guide other wolves in the pack to update their positions based on their own positions. By simulating the hunting behavior of wolf packs, they search for better solutions within the feasible solution space.

[0031] The updated population is sorted using non-dominated sorting, dividing the population into different non-dominated levels. Solutions at the same level are mutually non-dominated until the preset maximum number of iterations is reached, thus obtaining the planting configuration parameters for winter wheat.

[0032] A second aspect of the present invention provides a system for optimizing the planting configuration of winter wheat based on drought year type identification, the system comprising:

[0033] The preprocessing module is used to construct a drought year type identification index system, collect data of each drought year type identification index in real time in the target planting area, preprocess the collected data, and obtain preprocessed data.

[0034] The identification module is used to input the preprocessed data into the drought year type identification model, and to identify the current drought year type of the target planting area through the drought year type identification model;

[0035] The determination module is used to determine the appropriate planting configuration parameters for winter wheat based on the identified drought year type, combined with historical rainfall data and soil characteristics of the target planting area;

[0036] The adjustment module is used to monitor various parameters in real time through sensors during the planting process, obtain monitoring data, and adjust the planting configuration parameters based on the monitoring data so that winter wheat can obtain the best growth conditions in different drought years.

[0037] A third aspect of the present invention provides a device for optimizing the planting configuration of winter wheat in accordance with drought year type identification. The device includes a memory and at least one processor. The memory stores instructions. The at least one processor invokes the instructions in the memory to cause the device to perform the steps of the method for optimizing the planting configuration of winter wheat in accordance with drought year type identification as described in any of the preceding claims.

[0038] A fourth aspect of the present invention provides a computer-readable storage medium storing instructions that, when executed by a processor, implement the steps of the method for optimizing the planting configuration of winter wheat based on drought year type identification as described in any of the preceding claims.

[0039] The technical solution provided by this invention constructs a drought year type identification index system, collects data on various drought year type identification indicators in real time within the target planting area, preprocesses the collected data to obtain preprocessed data, inputs the preprocessed data into a drought year type identification model, and identifies the current drought year type in the target planting area through the drought year type identification model; based on the identified drought year type, combined with historical rainfall data and soil characteristics of the target planting area, determines the appropriate planting configuration parameters for winter wheat; during the planting process, sensors monitor various parameter information in real time to obtain monitoring data, and adjust the planting configuration parameters according to the monitoring data to achieve the desired results. This invention enables winter wheat to achieve optimal growth conditions under different drought years. It can accurately identify different drought years, providing a reliable basis for optimizing the planting configuration of winter wheat in rainy seasons. It can rationally adjust irrigation plans and planting parameters according to the drought year type, make full use of natural rainfall, reduce unnecessary irrigation water use, achieve high-efficiency water conservation in winter wheat planting, reduce water waste and irrigation costs. The optimized planting configuration can enable winter wheat to obtain suitable growth conditions under different drought years, improve the stress resistance of winter wheat, increase yield and improve quality. It can be flexibly adjusted according to the climate conditions, soil characteristics and planting habits of different regions, and has wide applicability. Attached Figure Description

[0040] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention.

[0041] Figure 1 A flowchart of the method for optimizing the planting configuration of winter wheat based on drought year type identification provided in an embodiment of the present invention;

[0042] Figure 2This is a schematic diagram of the structure of the winter wheat rain-suitable planting configuration optimization system based on drought year type identification provided in an embodiment of the present invention;

[0043] Figure 3 This is a schematic diagram of the structure of the winter wheat rain-suitable planting configuration optimization device based on drought year type identification provided in an embodiment of the present invention. Detailed Implementation

[0044] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” or “having,” and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, apparatus, product, or device that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or devices.

[0045] For ease of understanding, the specific process of the embodiments of the present invention is described below. Please refer to [link / reference]. Figure 1 The flowchart of the method for optimizing the planting configuration of winter wheat based on drought year type identification provided in this embodiment of the invention includes the following steps:

[0046] Step 101: Construct a drought year type identification index system, collect data of each drought year type identification index in real time in the target planting area, preprocess the collected data, and obtain preprocessed data.

[0047] Step 102: Input the preprocessed data into the drought year type identification model, and identify the current drought year type of the target planting area through the drought year type identification model;

[0048] Step 103: Based on the identified drought year type, combined with historical rainfall data and soil characteristics of the target planting area, determine the appropriate planting configuration parameters for winter wheat in terms of rainfall.

[0049] Step 104: During the planting process, the sensors monitor various parameters in real time to obtain monitoring data. Based on the monitoring data, the planting configuration parameters are adjusted so that winter wheat can obtain the best growth conditions in different drought years.

[0050] In this embodiment, drought-related influencing factors are clearly identified. Taking into account meteorological, soil, and hydrological factors, precipitation, evaporation, soil moisture content, groundwater level, and meteorological drought indices (such as the Standardized Precipitation Index (SPI) and precipitation anomaly percentage) are selected as basic indicators. These indicators reflect the regional drought situation from different dimensions; for example, precipitation directly reflects water replenishment, and soil moisture content reflects the availability of water for crops. Methods such as the Analytic Hierarchy Process (AHP) or Principal Component Analysis (PCA) are used to analyze the importance of each indicator in identifying drought year types, assigning corresponding weights to each indicator. A judgment matrix is ​​constructed through expert scoring (AHP), or principal components are extracted by dimensionality reduction of the data (PCA) to determine the contribution of each indicator in drought year type identification. Within the target planting area, meteorological stations, soil monitoring sensors, and groundwater monitoring wells are rationally deployed. Meteorological stations collect real-time meteorological data such as precipitation, evaporation, temperature, and humidity; soil monitoring sensors acquire data on soil moisture content, soil temperature, and soil electrical conductivity at different soil depths; and groundwater monitoring wells record changes in groundwater levels. Data is collected at set time intervals (e.g., hourly, daily) to form a raw dataset. The raw data is then checked to remove erroneous data caused by equipment failure or transmission errors, as well as abnormal data that is clearly inconsistent with reality (e.g., negative precipitation values). A spatiotemporal attention interpolation algorithm (ST-Impute) is applied to process missing data. Normalization or standardization methods are used to transform index data with different dimensions and value ranges into a unified interval (e.g., [0,1] or a standard normal distribution with a mean of 0 and a standard deviation of 1), eliminating the influence of dimensional differences on subsequent analysis and obtaining preprocessed data.

[0051] In this embodiment, the drought year type identification index data collected within the target planting area are arranged according to time series and spatial location to form a three-dimensional spatiotemporal matrix. A weight is initialized for each element in the three-dimensional spatiotemporal matrix. For each spatial location and index in the three-dimensional spatiotemporal matrix, the data series of that location and index at different time points are analyzed. By calculating the characteristics of the data in the time series, the attention weight in the time dimension is obtained. For each time point and index, the attention weight in the spatial dimension is calculated by analyzing the distance between spatial locations and the data correlation. The time dimension attention weight and the spatial dimension attention weight are fused to obtain the comprehensive spatiotemporal attention weight for each data element. For locations with missing data, the surrounding data is weighted and summed based on the comprehensive spatiotemporal attention weight of the surrounding data elements to calculate the interpolation result of the missing data, thus completing the missing data processing. The calculated interpolation result replaces the original missing data locations in the three-dimensional spatiotemporal matrix, resulting in a spatiotemporal matrix processed by the ST-Impute algorithm, thereby obtaining the preprocessed data.

[0052] In this embodiment, a CNN-BiLSTM-Attention spatiotemporal feature extraction module is built, domain adaptation is introduced, and self-supervised pre-training and semi-supervised learning are combined to train a drought year type identification model. The pre-processed data is input into the drought year type identification model. The model extracts the spatiotemporal features of the data through the CNN-BiLSTM-Attention spatiotemporal feature extraction module, and then the feature representation after domain adaptation is combined with the model parameters optimized by self-supervised and semi-supervised learning to output the identification result of the current drought year type of the target planting area.

[0053] In this embodiment, the training samples used for model training are organized according to time and spatial dimensions to form a multidimensional data matrix. In the CNN layer, multiple convolutional kernels of different sizes and numbers are set to extract features from the data through convolution operations, capturing local features in the spatial dimension. Pooling operations are used to reduce the data dimensionality. The feature vectors output from the CNN layer are input to the BiLSTM layer. The BiLSTM, through a gating mechanism, processes time-series data simultaneously from both forward and backward directions, learning the long-term dependencies of the data in the time dimension and extracting dynamic change features between data at different time points. The output of the BiLSTM layer is used as input in the attention mechanism layer, which automatically focuses on spatiotemporal features related to drought year type identification by calculating the weights of different features. The training samples used for model training are used as the source domain, and the target planting area is used as the target domain, employing adversarial... The training method involves adding a domain discriminator to the model. During training, the CNN-BiLSTM-Attention spatiotemporal feature extraction module learns to extract domain-insensitive feature representations, aligning the source and target domain data in the feature space. Pre-training is performed using a masked autoencoder approach, randomly masking the input training samples. The masked data is then fed into the CNN-BiLSTM-Attention spatiotemporal feature extraction module, where the model reconstructs the original unmasked data, learns the latent feature representations, and optimizes the model parameters. After pre-training, a small amount of labeled data and a large amount of unlabeled data are input into the model. A supervised loss function is calculated using the labeled data, while an unsupervised loss function is calculated using the unlabeled data. The two loss functions are then weighted and summed to optimize the model parameters.

[0054] In this embodiment, a certain number of gray wolf individuals are randomly generated within the feasible solution space to form an initial population. Each gray wolf individual represents a set of winter wheat planting configuration parameters. The fitness value of each gray wolf individual is calculated, and the gray wolf individuals are divided into four levels: α, β, δ, and ω according to their fitness values. The α wolf represents the optimal solution, the β and δ wolves assist the α wolf in making decisions, and the ω wolf represents a poor solution. The α, β, and δ wolves guide other wolves in the population to update their positions based on their own positions. By simulating the hunting behavior of the wolf pack, a better solution is searched within the feasible solution space. The updated population is then subjected to non-dominated sorting, dividing the population into different non-dominated levels. Solutions at the same level are mutually non-dominated until a preset maximum number of iterations is reached to obtain the winter wheat planting configuration parameters.

[0055] In this embodiment, drought years are classified into different levels, including mild drought years, moderate drought years, severe drought years, and extremely severe drought years. Sowing time: In mild drought years, sowing time can be appropriately delayed to take advantage of later rainfall; in severe and extremely severe drought years, sowing should be done earlier so that winter wheat can utilize limited soil moisture in the early growth stage. Planting density: As the severity of drought increases, planting density should be appropriately reduced to decrease competition for water and nutrients among plants. Irrigation plan: In mild drought years, the frequency and amount of irrigation should be reduced, relying mainly on natural rainfall; in severe and extremely severe drought years, a scientific irrigation plan should be developed, prioritizing irrigation during the critical water-demand period for winter wheat, and adopting water-saving irrigation technologies such as drip irrigation and sprinkler irrigation. Fertilization plan: Under drought conditions, the application of phosphorus and potassium fertilizers should be increased to improve the drought resistance of winter wheat; the application of nitrogen fertilizer should be reduced to avoid excessive water consumption due to excessive vegetative growth.

[0056] In this embodiment, a multi-objective gray wolf optimization algorithm (MOGWO) is developed, defining multiple optimization objectives such as maximizing yield, minimizing water consumption, and reducing planting costs, and constructing corresponding objective functions. A certain number of gray wolf individuals are randomly generated within the feasible solution space to form an initial population, with each individual representing a set of winter wheat planting configuration parameters (sowing time, planting density, irrigation plan, fertilization scheme, etc.). The fitness value of each individual is calculated, and wolf pack levels (α, β, δ, and ω) are determined according to fitness. α, β, and δ wolves guide other wolves to update their positions, and the search for better positions is achieved by simulating wolf pack hunting behavior. The optimal solution is obtained by performing non-dominated sorting to form the Pareto front, iterating repeatedly until the preset conditions are met, and obtaining the Pareto optimal solution set. A hybrid Bayesian network-genetic algorithm model is constructed, collecting historical rainfall data, soil characteristic data, and expert knowledge related to winter wheat planting in the target planting area to determine the causal relationships between variables, constructing the Bayesian network structure, and determining the network parameters through data learning. Within the range of variable values ​​determined by the Bayesian network, an initial population for the genetic algorithm is randomly generated, individuals are input into the Bayesian network, fitness values ​​are calculated based on the actual objective function, and the population is selected, crossovered, and... The mutation operation is repeated iteratively to obtain optimized planting configuration parameters. A dynamic parameter adjustment strategy is constructed using deep reinforcement learning (PPO algorithm). The actual planting environment of the target planting area is defined as the environment, and the algorithm model used to adjust the planting configuration parameters is called the agent. The environmental state includes information such as drought year type, real-time meteorological data, and soil moisture content. The agent's actions are adjustments to the planting configuration parameters. A reward function is designed to provide feedback on the quality of actions based on planting goals (such as increased yield, reduced water consumption, etc.). The agent's policy network and value network are initialized. The agent interacts in the environment, collects data and stores it in an experience replay buffer. Data is sampled from the buffer and the policy network and value network are updated using the PPO algorithm. The training is repeated until the policy converges. The planting configuration parameters are determined comprehensively. Combining the identified drought year type and referring to the Pareto optimal solution set obtained by the MOGWO algorithm, the parameters are optimized using a Bayesian network-genetic algorithm hybrid model. Based on the dynamic adjustment strategy constructed by deep reinforcement learning (PPO algorithm), and considering the historical rainfall data and soil characteristics of the target planting area, the appropriate rainfall-suitable planting configuration parameters for winter wheat, such as sowing time, planting density, irrigation plan, and fertilization scheme, are finally determined.

[0057] In this embodiment, soil moisture sensors, meteorological sensors (monitoring precipitation, evaporation, temperature, humidity, etc.), and plant growth sensors (monitoring plant height, leaf area, etc.) are rationally deployed within the planting area to collect real-time information on soil, meteorological, and plant growth parameters, obtaining raw monitoring data. The raw monitoring data is then cleaned to remove noise and outliers. Interpolation methods or knowledge graph-based reasoning methods are used to fill in missing data, ensuring data integrity and usability. A meteorological-soil coupling simulator is constructed, collecting long-term meteorological and soil data from the target planting area, and dividing it into training, validation, and test sets. A model architecture based on neural processes is built, including an encoder, attention mechanism module, and decoder. By minimizing the loss function between the predicted results and the actual data, the model is trained using optimization algorithms such as stochastic gradient descent, completing the construction of the meteorological-soil coupling simulator. A meta-learning (MAML) framework is applied to achieve rapid scene adaptation. Planting scenarios under different drought years, time periods, and planting area conditions are defined as different tasks. Model parameters are initialized, and meta-training is performed on multiple tasks. The model is fine-tuned using a small amount of task data, and the initial parameters are updated based on the loss. When encountering a new scenario, The initial parameters obtained from meta-training are used to fine-tune the model with a small amount of new scenario data, enabling it to quickly adapt to new scenarios. A spatiotemporal knowledge graph is constructed to achieve multidimensional decision-making reasoning, identifying entities (meteorological elements, soil properties, winter wheat growth stages, planting configuration parameters, etc.) and relationships (such as the impact of weather on soil moisture content) related to winter wheat planting. Information is extracted from historical data, expert knowledge, and literature, represented in triplet form, and stored in the knowledge graph database. The spatiotemporal knowledge graph is constructed and updated in real time. Real-time monitoring data is transformed into entities and relationships in the knowledge graph, and graph reasoning algorithms are used to analyze the interaction of different factors. The system deduces a suitable adjustment scheme for planting configuration parameters based on the feedback; it inputs the preprocessed monitoring data into a meteorological-soil coupled simulator, and uses the model parameters adapted through meta-learning to predict the changing trend of the meteorological-soil coupling state; it transforms the current monitoring data and simulator prediction results into entities and relationships in a knowledge graph, and deduces the optimal adjustment scheme for planting configuration parameters based on the spatiotemporal knowledge graph, and sends the adjustment instructions to the corresponding planting equipment to complete the adjustment of planting configuration parameters; at the same time, it continuously monitors the growth status of winter wheat and related environmental parameters after the adjustment, evaluates the adjustment effect, and uses it for subsequent model optimization and parameter adjustment strategy improvement.

[0058] Please see Figure 2 A schematic diagram of the structure of the winter wheat rain-suitable planting configuration optimization system based on drought year type identification provided in this embodiment of the invention. The system includes:

[0059] The preprocessing module is used to construct a drought year type identification index system, collect data of each drought year type identification index in real time in the target planting area, preprocess the collected data, and obtain preprocessed data.

[0060] The identification module is used to input the preprocessed data into the drought year type identification model, and to identify the current drought year type of the target planting area through the drought year type identification model;

[0061] The determination module is used to determine the appropriate planting configuration parameters for winter wheat based on the identified drought year type, combined with historical rainfall data and soil characteristics of the target planting area;

[0062] The adjustment module is used to monitor various parameters in real time through sensors during the planting process, obtain monitoring data, and adjust the planting configuration parameters based on the monitoring data so that winter wheat can obtain the best growth conditions in different drought years.

[0063] Figure 3 This is a schematic diagram of a winter wheat rain-suitable planting configuration optimization device 600 based on drought year type identification, provided by an embodiment of the present invention. This winter wheat rain-suitable planting configuration optimization device 600 can vary significantly due to different configurations or performance. It may include one or more central processing units (CPUs) 610 (e.g., one or more processors) and a memory 620, and one or more storage media 630 (e.g., one or more mass storage devices) storing application programs 633 or data 632. The memory 620 and storage media 630 can be temporary or persistent storage. The program stored in the storage media 630 may include one or more modules (not shown in the diagram), each module including a series of instruction operations on the winter wheat rain-suitable planting configuration optimization device 600 based on drought year type identification. Furthermore, the processor 610 may be configured to communicate with the storage media 630 and execute the series of instruction operations in the storage media 630 on the winter wheat rain-suitable planting configuration optimization device 600 based on drought year type identification to implement the method provided in the above embodiment.

[0064] The winter wheat rainfall-appropriate planting configuration optimization device 600 based on drought year type identification may also include one or more power supplies 640, one or more wired or wireless network interfaces 650, one or more input / output interfaces 660, and / or one or more operating devices 631, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will understand that... Figure 3The illustrated equipment structure for optimizing the planting configuration of winter wheat based on drought year type identification does not constitute a limitation on the computer equipment provided by this invention. It may include more or fewer components than illustrated, or combine certain components, or have different component arrangements.

[0065] The present invention also provides a computer-readable storage medium, which can be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when the instructions are executed on a computer, cause the computer to perform the various steps of the method for optimizing the planting configuration of winter wheat based on drought year type identification provided in the above embodiments.

[0066] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described equipment or apparatus / unit can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0067] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0068] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing the planting configuration of winter wheat based on drought year type identification, characterized in that, The method includes the following steps: A drought year type identification index system was constructed, and data of each drought year type identification index were collected in real time in the target planting area. The collected data were preprocessed to obtain preprocessed data. The preprocessed data is input into the drought year type identification model, which is used to identify the current drought year type in the target planting area. Based on the identified drought year type, combined with historical rainfall data and soil characteristics of the target planting area, the appropriate planting configuration parameters for winter wheat in rainy conditions are determined. During the planting process, sensors are used to monitor various parameters in real time to obtain monitoring data. Based on the monitoring data, the planting configuration parameters are adjusted so that winter wheat can obtain the best growth conditions in different drought years. The step of inputting the preprocessed data into the drought year type identification model, and identifying the current drought year type in the target planting area through the drought year type identification model, includes: A CNN-BiLSTM-Attention spatiotemporal feature extraction module was built, domain adaptation was introduced, and self-supervised pre-training and semi-supervised learning were combined to train a drought year type recognition model. The preprocessed data is input into the drought year type identification model. The model extracts the spatiotemporal features of the data through the CNN-BiLSTM-Attention spatiotemporal feature extraction module, and then the feature representation after domain adaptive processing is combined with the model parameters optimized by self-supervised and semi-supervised learning to output the identification result of the current drought year type of the target planting area. The process involves determining suitable rainfall-appropriate planting parameters for winter wheat based on the identified drought year type, combined with historical rainfall data and soil characteristics of the target planting area, including: A certain number of gray wolf individuals are randomly generated in the feasible solution space to form an initial population. Each gray wolf individual represents a set of winter wheat planting configuration parameters. Calculate the fitness value of each individual gray wolf, and divide the gray wolves into groups based on their fitness values. and Four levels, The wolf is the optimal solution. wolves and Wolf Support wolves make decisions. The wolf is a poor solution; Wolves guide other wolves in the pack to update their positions based on their own location, and by simulating the hunting behavior of wolf packs, they search for better solutions within the feasible solution space. The updated population is sorted using non-dominated sorting, dividing the population into different non-dominated levels. Solutions at the same level are mutually non-dominated until the preset maximum number of iterations is reached, thus obtaining the planting configuration parameters for winter wheat.

2. The method for optimizing winter wheat planting configuration based on drought year type identification as described in claim 1, characterized in that, The drought year type identification index system includes at least precipitation, evaporation, soil moisture content, groundwater level, and meteorological drought index.

3. The method for optimizing winter wheat planting configuration based on drought year type identification as described in claim 1, characterized in that, The aforementioned drought year type identification index system is constructed by collecting data on various drought year type identification indicators in real time within the target planting area. The collected data is then preprocessed to obtain preprocessed data, including: The drought year type identification index data collected in the target planting area are arranged according to time series and spatial location to form a three-dimensional spatiotemporal matrix; A weight is initialized for each element in the three-dimensional spatiotemporal matrix. For each spatial location and index in the three-dimensional spatiotemporal matrix, the data sequence of the location and index at different time points is analyzed. By calculating the characteristics of the data in the time series, the attention weight in the time dimension is obtained. For each time point and indicator, attention weights in the spatial dimension are calculated by analyzing the distance between spatial locations and the correlation of data. By fusing the attention weights of the time dimension and the attention weights of the spatial dimension, a comprehensive spatiotemporal attention weight is obtained for each data element; For locations with missing data, the surrounding data is weighted and summed based on the comprehensive spatiotemporal attention weights of the surrounding data elements to calculate the interpolation result of the missing data, thus completing the processing of the missing data. The calculated interpolation results are used to replace the original missing data positions in the three-dimensional spatiotemporal matrix to obtain the spatiotemporal matrix after processing by the ST-Impute algorithm, thus obtaining the preprocessed data.

4. The method for optimizing winter wheat planting configuration based on drought year type identification as described in claim 1, characterized in that, The aforementioned CNN-BiLSTM-Attention spatiotemporal feature extraction module incorporates domain adaptation and combines self-supervised pre-training and semi-supervised learning to train a drought year type recognition model, including: The training samples used for model training are organized according to the time and space dimensions to form a multidimensional data matrix. In the CNN layer, multiple convolutional kernels of different sizes and numbers are set. The data is extracted through convolution operations to capture local features of the data in the spatial dimension. Pooling operations are used to reduce the data dimensionality. The feature vectors output by the CNN layer are input into the BiLSTM layer. The BiLSTM uses a gating mechanism to process time series data from both the forward and backward directions, learn the long-term dependencies of the data in the time dimension, and extract the dynamic change features between data at different time points. Using the output of the BiLSTM layer as input, the attention mechanism layer automatically focuses on the spatiotemporal features related to drought year type identification by calculating the weights of different features. The training samples used for model training are taken as the source domain and the target planting region is taken as the target domain. An adversarial training approach is adopted, and a domain discriminator is added to the model. During the training process, the CNN-BiLSTM-Attention spatiotemporal feature extraction module learns to extract feature representations that are insensitive to the domain, so that the data of the source domain and the target domain are aligned in the feature space.

5. The method for optimizing winter wheat planting configuration based on drought year type identification as described in claim 1, characterized in that, The construction of the CNN-BiLSTM-Attention spatiotemporal feature extraction module, the introduction of domain adaptation, and the combination of self-supervised pre-training and semi-supervised learning to train a drought year type recognition model also include: Pre-training is performed using a masked autoencoder approach. The input training samples are randomly masked, and the masked data is then fed into the CNN-BiLSTM-Attention spatiotemporal feature extraction module. The model reconstructs the original unmasked data, learns the latent feature representation of the data, and optimizes the model parameters. After pre-training, a small amount of labeled data and a large amount of unlabeled data are input into the model together. A supervised loss function is calculated using the labeled data, while an unsupervised loss function is calculated using the unlabeled data. The two loss functions are then weighted and summed to optimize the model parameters.

6. A system for optimizing the planting configuration of winter wheat based on drought year type identification, characterized in that, The system includes: The preprocessing module is used to construct a drought year type identification index system, collect data of each drought year type identification index in real time in the target planting area, preprocess the collected data, and obtain preprocessed data. The identification module is used to input preprocessed data into the drought year type identification model, which identifies the current drought year type of the target planting area. The module involves building a CNN-BiLSTM-Attention spatiotemporal feature extraction module, introducing domain adaptation, and combining self-supervised pre-training and semi-supervised learning to train the drought year type identification model. The preprocessed data is then input into the drought year type identification model. The model extracts the spatiotemporal features of the data through the CNN-BiLSTM-Attention spatiotemporal feature extraction module, performs domain adaptation on the feature representation, and combines the optimized model parameters from self-supervised and semi-supervised learning to output the identification result of the current drought year type of the target planting area. The determination module is used to determine the optimal planting configuration parameters for winter wheat based on the identified drought year type, combined with historical rainfall data and soil characteristics of the target planting area. This involves randomly generating a certain number of gray wolf individuals within the feasible solution space to form an initial population, with each gray wolf representing a set of winter wheat planting configuration parameters; calculating the fitness value of each gray wolf individual; and classifying the gray wolf individuals according to their fitness values. and Four levels, The wolf is the optimal solution. wolves and Wolf Support wolves make decisions. The wolf is a poor solution; The wolf guides other wolves in the pack to update their positions based on its own position. By simulating the hunting behavior of the wolf pack, a better solution is searched in the feasible solution space. The updated population is then sorted into different non-dominated levels. Solutions in the same level are mutually non-dominated until the preset maximum number of iterations is reached, and the planting configuration parameters for winter wheat are obtained. The adjustment module is used to monitor various parameters in real time through sensors during the planting process, obtain monitoring data, and adjust the planting configuration parameters based on the monitoring data so that winter wheat can obtain the best growth conditions in different drought years.

7. A device for optimizing the planting configuration of winter wheat based on drought year type identification, characterized in that, The winter wheat rain-suitable planting configuration optimization device based on drought year type identification includes a memory and at least one processor. The memory stores instructions. The at least one processor calls the instructions in the memory to cause the winter wheat rain-suitable planting configuration optimization device based on drought year type identification to perform each step of the winter wheat rain-suitable planting configuration optimization method based on drought year type identification as described in any one of claims 1-5.

8. A computer-readable storage medium storing instructions thereon, characterized in that, When the instructions are executed by the processor, they implement each step of the method for optimizing the planting configuration of winter wheat based on drought year type identification as described in any one of claims 1-5.

Images