Peanut yield simulation model construction method based on reinforcement learning

Abstract

The invention relates to the technical field of crop yield models, in particular to a peanut yield simulation model construction method based on reinforcement learning, and the method comprises the steps: collecting meteorological index parameters and field management parameters; calculating a meteorological index characterization value based on the meteorological index parameters; analyzing a day-by-day heat-time characterization value; determining a target sample development stage based on the day-by-day thermal time characterization value; determining a target sample leaf area index characterization value; analyzing the characterization value of the light interception rate; analyzing the characterization value of the light energy utilization rate temperature correction coefficient; analyzing the characterization value of the soil moisture stress coefficient; analyzing the characterization value of the total biomass; analyzing a biomass day-by-day transfer characterization value; analyzing a day-by-day seed growth characterization value; and when the target sample is in a mature stage, calculating the final yield of the target sample based on the accumulated seed dry weight characterization value and the day-by-day seed growth characterization value. According to the method, the accuracy of the peanut yield simulation model is improved, and a precise decision is provided for optimizing field management, so that the actual yield of peanuts is improved.

Description

Technical Field

[0001] This application relates to the field of crop yield modeling technology, and in particular to a method for constructing a peanut yield simulation model based on reinforcement learning. Background Technology

[0002] Peanuts are an important oilseed crop. In 2023, my country's peanut planting area reached 4,797,000 hectares, with a total output of 19.231 million tons. China ranked second globally in planting area and first in total output. Henan Province ranked first in China in both peanut planting area and output, accounting for 27.2% and 33.2% of the national total, respectively. In recent years, the Henan Provincial Government has increased the construction of large-scale demonstration zones for green, high-yield, and high-efficiency peanut production, continuing to expand the planting area of ​​high-quality peanuts, particularly those high in oil and oleic acid, to increase yield per unit area, improve quality, and enhance overall economic benefits.

[0003] Peanuts are a warm-season, drought-tolerant crop that thrives in moist conditions but is susceptible to waterlogging. The degree of matching between temperature, water, light, and soil climate characteristics during the growing season plays a crucial role in peanut growth, development, and yield, especially the influence of climatic factors during critical periods. Current technologies have limited research on the relationship between peanut growth and meteorological conditions for peanut varieties unique to Henan Province. Existing technologies often employ single-development-stage models, and their simulation effects are yet to be evaluated, with a lack of yield models. Therefore, it is necessary to combine this with the actual production conditions in Henan Province, systematically studying the photosynthetic, water, and dry matter accumulation processes of peanuts to determine parameters for each growth stage. This will allow for the construction of a localized growth model, which can then be applied to a cultivation expert system to support high-yield peanut production and decision-making.

[0004] Chinese Patent Publication No. CN118211730A discloses a method for predicting crop yield based on meteorological feature matching and crop growth models. The method includes the following steps: collecting and preprocessing local crop management information and historical meteorological data; using meteorological data matching to simulate and predict unknown meteorological data for the current year; limiting yield prediction results based on a yield abundance / shortness index; assimilating the crop growth model using real-time remote sensing data; constructing a random forest model to obtain regional yield prediction results; and weighted calculation of the final yield prediction result. This invention constructs a method for simulating unknown meteorological data by combining meteorological data matching and a yield abundance / shortness index, and combines this with a crop growth model to predict regional yield. It establishes a yield prediction system integrating multiple methods, significantly reducing the cost and time of yield prediction. It can make accurate predictions of different crop yields early on, determine the overall supply and demand situation, and accurately estimate the total yield of the year approximately 45 days before harvest.

[0005] Chinese Patent Publication No. CN111898922A discloses a multi-scale crop yield assessment method and system. The assessment method includes: acquiring first reanalysis data, ERA5 reanalysis data, and crop parameters; outputting simulated leaf area index and simulated yield using the WOFOST model; calculating candidate meteorological indicators using the first reanalysis data and ERA5 reanalysis data, and determining seasonal weather variables in conjunction with the simulated yield; determining a yield assessment model based on the simulated yield and the seasonal weather variables; acquiring remote sensing data from the winter wheat greening stage to maturity stage; the remote sensing data includes 12 leaf area index images per year; correcting the remote sensing leaf area index based on the simulated leaf area index to determine the corrected remote sensing leaf area index; and inputting the corrected remote sensing leaf area index and the seasonal weather variables into the yield assessment model based on the SCYM multi-scale crop framework to determine the multi-scale crop yield. This invention can improve the accuracy of yield assessment.

[0006] Therefore, it is evident that the existing technology has the following problems: The lack of quantitative research on the relationship between peanut growth and development and meteorological conditions in existing technologies has led to the problems of single models for peanut growth and development period and low accuracy of models simulating peanut growth and development parameters, resulting in a decrease in actual peanut yield. Summary of the Invention

[0007] To address this, the present invention provides a method for constructing a peanut yield simulation model based on reinforcement learning, which overcomes the problem that the lack of quantitative research on the relationship between peanut growth and development and meteorological conditions leads to a single model for peanut growth and development period and meteorological conditions, and low accuracy of the model for simulating peanut growth and development parameters, resulting in a decline in actual peanut yield.

[0008] To achieve the above objectives, this invention provides a method for constructing a peanut yield simulation model based on reinforcement learning, comprising: Collect meteorological index parameters of the target agricultural meteorological experimental station within a historical period and field management parameters of the target sample within a historical period; Calculate the daily thermal time characterization value based on the aforementioned meteorological index parameters; The developmental stage of the target sample is determined based on the daily thermal time characterization values; The leaf area index characterization value of the target sample is determined based on the developmental stage and daily thermal time characterization value of the target sample. Calculate the daily solar radiation based on the aforementioned meteorological parameters; The light interception rate is calculated based on the extinction coefficient and the leaf area index of the target sample. Calculate the temperature correction coefficient for light energy utilization based on the aforementioned meteorological index parameters; The soil moisture stress coefficient characterization value was calculated based on the meteorological index parameters and field management parameters. Calculate the light energy utilization rate moisture correction coefficient based on the soil moisture stress coefficient characterization value; The total biomass characterization value is calculated based on the daily solar radiation, light interception rate characterization value, light energy utilization rate temperature correction coefficient, and light energy utilization rate water correction coefficient. Calculate the daily biomass transfer characterization value based on the aforementioned meteorological index parameters; The daily seed growth characterization value is calculated based on the total biomass characterization value, the daily transfer characterization value, and the food conversion coefficient. When the target sample is in the mature stage, the final yield of the target sample is calculated based on the cumulative seed dry weight characterization value and the daily seed growth characterization value. The meteorological parameters include temperature, sunshine, precipitation, wind speed, and relative humidity. The field management parameters include planting density, sowing date, and irrigation amount.

[0009] Furthermore, the process of calculating the daily thermal time characterization value of the meteorological index parameters includes: Extract temperature parameters from the target agricultural meteorological experimental station over a historical period; If the temperature parameter is less than the predetermined lower temperature threshold or greater than the predetermined upper temperature threshold, then the daily thermal time characterization value is determined to be the predetermined minimum value. If the temperature parameter is greater than or equal to the predetermined lower temperature threshold and less than or equal to the predetermined upper temperature threshold, then the daily thermal time characterization value is determined to be the product of the difference between the predetermined suitable temperature threshold and the predetermined lower temperature threshold and the predetermined temperature function.

[0010] Furthermore, the process of calculating the daily transfer rate characterization value from the daily thermal time characterization value includes: The ratio of the daily thermal time characterization value to the predetermined total thermal time characterization threshold is determined as the daily transfer volume characterization value.

[0011] Furthermore, the process of determining the developmental stage of the target sample using the daily thermal time characterization values ​​includes: If the daily thermal time characterization value is less than or equal to the first thermal time characterization threshold, then the target sample is determined to be from sowing to emergence stage; If the daily thermal time characterization value is greater than the predetermined first thermal time characterization threshold and less than or equal to the predetermined second thermal time characterization threshold, then the target sample is determined to be in the stage from seedling emergence to the start of seed growth. If the daily thermal time characterization value is greater than the predetermined second thermal time characterization threshold and less than or equal to the predetermined third thermal time characterization threshold, then the target sample is determined to be the stage from the start of seed growth to the cessation of leaf growth. If the daily thermal time characterization value is greater than the predetermined third thermal time characterization threshold and less than or equal to the predetermined fourth thermal time characterization threshold, then the target sample is determined to be the leaf cessation to maturity stage. If the daily thermal time characterization value is greater than the predetermined total thermal time characterization threshold, the target sample is determined to be in the mature stage.

[0012] Furthermore, the process of determining the leaf area index characterization value of the target sample at the developmental stage includes: If the target sample is from sowing to emergence, then the leaf area index is determined to be the predetermined minimum value. If the target sample is from seedling emergence to the start of seed growth, then the leaf area index is determined to be the sum of the daily increase in leaf area index and the leaf area index value of the previous day. If the target sample is from the beginning of seed growth to the leaf cessation stage, then the leaf area index is determined to be the product of the amount of total dry matter allocated to the leaf and the specific leaf area. If the target sample is the leaf growth cessation stage to maturity, then the leaf area index characterization value is determined to be a function of the daily thermal time characterization value and the mature leaf area index characterization value.

[0013] Furthermore, the process of determining daily solar radiation using the meteorological index parameters includes: Extract sunshine parameters from the target agricultural meteorological experimental station over a historical period; The daily solar radiation is calculated from the sunshine duration based on the evapotranspiration model.

[0014] Furthermore, the process of calculating the light interception rate characterization value from the extinction coefficient and the target sample leaf area index characterization value includes: Extract the characterization values ​​of the light emission coefficient and the leaf area index of the target sample; The product of the extinction coefficient and the leaf area index of the target sample is calculated as the data characterization value; The difference between the predetermined maximum value and the negative data representation value raised to the power of the natural constant is the light interception rate representation value.

[0015] Furthermore, the process of determining the temperature correction coefficient for light energy utilization efficiency using the meteorological index parameters includes: Extract historical temperature parameters from the target agricultural meteorological experimental station over a given period; If the temperature parameter is less than or equal to the predetermined lower temperature threshold or greater than or equal to the predetermined upper temperature threshold, then the temperature correction factor for light energy utilization is determined to be the predetermined minimum value. If the temperature parameter is greater than the predetermined lower temperature threshold and less than the predetermined suitable lower temperature threshold, then the light energy utilization temperature correction coefficient is determined as the ratio of the difference between the temperature parameter and the predetermined lower temperature threshold to the difference between the predetermined suitable lower temperature threshold and the predetermined lower temperature threshold. If the temperature parameter is greater than the predetermined upper limit threshold of suitable temperature but less than the predetermined upper limit threshold of temperature, then the temperature correction factor for light energy utilization is determined to be the ratio of the difference between the predetermined upper limit threshold of suitable temperature and the temperature parameter to the difference between the predetermined upper limit threshold of temperature and the predetermined upper limit threshold of suitable temperature. If the temperature parameter is greater than or equal to the predetermined lower limit of the suitable temperature and less than or equal to the predetermined upper limit of the suitable temperature, then the temperature correction factor for light energy utilization is determined to be the predetermined maximum value. Furthermore, the process of calculating the soil moisture stress coefficient characterization value based on the field management parameters, and calculating the light energy use efficiency moisture correction coefficient based on the soil moisture stress coefficient characterization value, includes: The soil moisture stress coefficient is equal to the light energy utilization rate moisture correction coefficient.

[0016] Furthermore, the process of calculating the daily total biomass characterization value from the meteorological index characterization value includes: Extract the daily solar radiation, light interception rate, temperature correction coefficient for light energy utilization, and moisture correction coefficient for light energy utilization. The total biomass characterization value is determined by multiplying the daily solar radiation, light interception rate, light energy utilization rate temperature correction coefficient, and light energy utilization rate moisture correction coefficient characterization values ​​with a predetermined biomass threshold and summing them with the daily total biomass characterization value per square meter.

[0017] Furthermore, the process of calculating the daily biomass transfer characterization value from the daily thermal time characterization value and the total biomass characterization value includes: The ratio of the daily thermal time characterization value to the predetermined total thermal time characterization threshold is determined as the daily biomass transfer characterization value.

[0018] Furthermore, the process of calculating the daily seed growth characterization value from the total biomass characterization value, the daily biomass transfer characterization value, and the grain conversion coefficient, and calculating the final yield of the target sample from the cumulative seed dry weight characterization value and the daily seed growth characterization value, includes: Extract total biomass characterization values, daily biomass transfer characterization values, and food conversion coefficient; The summation of the total biomass characterization value, the daily biomass transfer characterization value, and the food conversion coefficient is determined as the daily seed growth characterization value. The cumulative seed dry weight characterization value and the daily seed growth characterization value are summed to determine the final yield of the target sample.

[0019] Compared with existing technologies, the beneficial effects of this invention are that it provides a method for constructing a peanut yield simulation model based on reinforcement learning. This invention integrates historical meteorological indicators and field management parameters from a target agricultural meteorological experimental station to construct a mechanism-driven crop yield simulation system. This system first calculates the daily cumulative thermal time based on meteorological data to accurately determine the crop development stage; it then dynamically calculates the leaf area index (LAI) based on the development stage, and further derives the canopy light interception rate along with solar radiation, extinction coefficient, and LAI, forming a quantitative description of photosynthesis. By integrating solar radiation, LAI, and light energy utilization efficiency temperature correction coefficients, and introducing a light energy utilization efficiency water correction mechanism based on soil moisture stress coefficient, the system effectively characterizes the impact of water conditions on biomass formation. The system accurately simulates the accumulation process of total biomass. Finally, based on organ allocation logic and the grain conversion coefficient, the system dynamically simulates the seed dry matter accumulation process and achieves accurate yield prediction through cumulative calculation at maturity. The closed-loop simulation process significantly improves the accuracy and explanatory power of yield prediction, providing a scientific basis for crop management decisions, resource optimization, and sustainable agricultural development.

[0020] In particular, this invention constructs a data acquisition module, a developmental stage module, a leaf area index module, a total biomass module, a water balance module, and a dry matter distribution and yield module. These modules are interconnected through a tight data and feedback mechanism, forming a dynamic crop growth simulation system. The data acquisition module provides the basic data for the entire system, driving the developmental stage module to determine the key stages of crop growth. This stage information then controls the algorithm switching between the leaf area index module and the dry matter distribution module. The leaf area index, as a core variable, directly affects the accumulation of total biomass and is input into the water balance module along with meteorological and soil data. The water stress coefficient calculated by the water balance module then serves as a key correction factor, feeding back to the leaf area index and total biomass modules, forming a closed loop that accurately responds to environmental changes. Through integrated simulation of multiple processes, this invention achieves precise and mechanistic prediction of peanut yield formation. It not only provides farmers with scientific yield forecasts to assist in decision-making but also optimizes field management measures such as irrigation by analyzing the relationship between water stress and growth and development, thereby effectively improving water resource utilization efficiency and promoting the transformation from traditional agriculture to data-driven precision agriculture. This system has significant value in both production guidance and scientific research.

[0021] In particular, this invention constructs a simulation system with a clear mechanism and precise feedback by systematically calculating peanut growth and development parameters. The calculated meteorological index values ​​provide the energy basis for driving biophysical processes in the model; the calculation of daily thermal time values ​​and the determination of developmental stages enable dynamic quantitative tracking of crop phenological processes, allowing the model to switch algorithms according to growth and development stages; based on this, the calculated leaf area index and light interception rate values ​​accurately quantify the canopy structure and its photosynthetic capacity, providing core input for biomass simulation; and the soil moisture stress coefficient is calculated through field management parameters, further yielding… The light energy utilization rate, temperature, and moisture correction coefficients cleverly couple the moisture status of the soil, plants, and atmosphere continuum into the biomass accumulation process, significantly improving the model's adaptability and accuracy in variable environments. Finally, by representing the total biomass value and allocating dry matter according to developmental stages until the final yield is obtained, the physiological and ecological chain of photosynthetic product formation, allocation, and economic yield formation is fully reproduced. The chain calculation and organic integration of parameters together realize the mechanistic, dynamic, and precise simulation of crop growth and yield formation processes, providing strong decision support for yield forecasting, disaster assessment, and management optimization.

[0022] In particular, this invention achieves a precise and quantitative description of peanut yield by constructing a mechanism-driven, process-oriented dynamic simulation system. The model dynamically determines the developmental stage by calculating daily thermal hours, ensuring that the simulation process is synchronized with the actual phenological rhythm of peanuts. By coupling leaf area index and light interception rate, canopy photosynthesis is accurately quantified, laying the foundation for biomass simulation. Temperature correction coefficients and water stress coefficients are introduced to correct light energy utilization, effectively capturing the stress effects of temperature and water conditions on peanut growth and significantly improving the simulation accuracy under variable weather conditions. Through dry matter allocation logic, the model rationally distributes accumulated biomass to pods according to developmental stages, achieving reliable prediction of peanut yield. It not only provides accurate yield forecasts but also analyzes key factors affecting yield, providing strong scientific basis and decision support for water and fertilizer management, variety selection, and climate risk assessment in peanut production. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating the steps of the peanut yield simulation model construction method based on reinforcement learning in an embodiment of the present invention. Figure 2 This is a simulation framework diagram of the peanut yield simulation model construction method based on reinforcement learning according to an embodiment of the present invention; Figure 3 This is a module relationship diagram of the peanut yield simulation model construction method based on reinforcement learning in an embodiment of the present invention; Figure 4This is a logic diagram for determining the developmental stage of a target sample based on daily thermal time characterization values, as described in an embodiment of the present invention. Detailed Implementation

[0024] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0025] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0026] Please see Figure 1 The diagram shows the steps of a peanut yield simulation model construction method based on reinforcement learning according to an embodiment of the present invention. The present invention provides a peanut yield simulation model construction method based on reinforcement learning, comprising: Step S1: Collect meteorological index parameters of the target agricultural meteorological experimental station and field management parameters of the target sample during the historical period; calculate the daily thermal time characterization value based on the meteorological index parameters; Step S2: Determine the developmental stage of the target sample based on the daily thermal time characterization value; determine the leaf area index characterization value of the target sample based on the developmental stage of the target sample; Step S3: Calculate the light interception rate characterization value based on the extinction coefficient and the leaf area index characterization value of the target sample; calculate the light energy utilization rate temperature correction coefficient based on the meteorological index characterization value; calculate the soil moisture stress coefficient characterization value based on the field management parameters and meteorological parameters; calculate the light energy utilization rate water correction coefficient based on the soil moisture stress coefficient characterization value; calculate the total biomass characterization value based on the meteorological index characterization value and the light energy utilization rate. Step S4: Calculate the daily biomass conversion characterization value based on the daily thermal time characterization value; calculate the daily seed growth characterization value based on the total biomass characterization value, the daily biomass conversion characterization value, and the grain conversion coefficient; when the target sample is in the mature stage, calculate the final yield of the target sample based on the cumulative seed dry weight characterization value and the daily seed growth characterization value.

[0027] In this embodiment, staggered sowing experiments were conducted from 2020 to 2023 at the Zhengzhou Agricultural Meteorological Experiment Station (113°39′E, 34°43′N, 110.0m) and the Shangqiu Agricultural Meteorological Observatory (115°32′E, 34°26′N, 105.0m). The high-oleic peanut variety "Yuhua 65" was used. The experiments were conducted from 2020 to 2023, with the 2020 and 2021 experiments conducted in Zhengzhou, and the 2022 and 2023 experiments conducted in Shangqiu. Three sowing periods were set for each year. The sowing method was direct seeding on flat plots, with a row spacing of 40cm, a plant spacing of 20cm, a sowing depth of 5cm, and 2 seeds per plant. Each sowing period was replicated three times, with each replicate having a plot area of ​​30m². 2 The plots were randomly arranged, and the previous crop was winter wheat.

[0028] Meteorological data were obtained from surface meteorological observation stations in Zhengzhou and Shangqiu. The distance between the field test sites and the meteorological observation fields was less than 50m. Both stations belong to the warm temperate continental monsoon climate zone. During the peanut growing season, the average annual precipitation in Zhengzhou and Shangqiu is 560.0mm and 603.1mm, respectively, the average temperature is 21.9℃ and 21.5℃, and the sunshine duration is 1295.9h and 1247.5h, respectively.

[0029] In this embodiment, a mechanism-driven crop yield simulation system was constructed by integrating historical meteorological indicators and field management parameters from the target agricultural meteorological experimental station. This system first calculates key characteristic values ​​such as the temperature correction coefficient for light energy use efficiency, daily solar radiation, and extinction coefficient based on meteorological data. Then, it accurately determines the crop development stage through daily thermal time accumulation. Combining the dynamic calculation of leaf area index (LAI) values ​​with the developmental stage, and further deriving the canopy light interception rate with the extinction coefficient, a quantitative description of photosynthesis is formed. By integrating solar radiation, LAI, and the temperature correction coefficient for light energy use efficiency, the system accurately simulates the accumulation process of total biomass. Simultaneously, a temperature correction mechanism for light energy use efficiency based on soil moisture stress coefficient is introduced to effectively characterize the impact of water conditions on biomass formation. Finally, based on organ allocation logic and the grain conversion coefficient, the system dynamically simulates the seed dry matter accumulation process and achieves accurate yield prediction through cumulative calculation at maturity. The closed-loop simulation process significantly improves the accuracy and explanatory power of yield prediction, providing a scientific basis for crop management decisions, resource optimization, and sustainable agricultural development.

[0030] Please see Figure 2 The diagram shown illustrates the simulation framework of a peanut yield simulation model construction method based on reinforcement learning, as described in this invention. The invention provides a peanut yield simulation model construction method based on reinforcement learning, comprising: The data acquisition module is used to collect meteorological index parameters of the target agricultural meteorological experimental station and field management parameters of the target sample during the historical period. The development stage module, which is connected to the data acquisition module, is used to calculate the daily heat time based on meteorological index parameters and field management parameters and to accumulate the daily heat time. The current development stage of peanut is determined based on the comparison between the accumulated heat time and the preset heat time threshold for each development stage. The leaf area index module, which is connected to the development stage module, is used to determine the current development stage, call the leaf area algorithm corresponding to the development stage, and calculate the daily leaf area index based on the daily heat time, the sowing density parameter, and biomass. The water balance module, which is connected to the data acquisition module and the total biomass module, is used to calculate the daily soil water storage based on the precipitation in the daily meteorological data, the soil moisture content in the soil data and the daily leaf area index, through the soil water balance equation, and to determine the water stress coefficient based on the comparison results of the daily soil water storage with the preset field capacity and wilting humidity. The total biomass module, which is connected to the leaf area index module, is used to calculate the daily biomass per square meter based on the solar radiation in the daily meteorological data, the calculated daily leaf area index, the light energy utilization temperature correction coefficient, and the water stress coefficient. The dry matter allocation and yield module, which is connected to the developmental stage module and the total biomass module, is used to allocate the daily biomass per square meter to the leaves, stems, roots and seeds according to a preset organ allocation coefficient based on the determined current developmental stage, and to accumulate and calculate the seed dry matter weight as the simulated peanut yield value.

[0031] Please see Figure 3 The diagram shown illustrates the module relationships of a peanut yield simulation model construction method based on reinforcement learning, as described in this invention. The invention provides a peanut yield simulation model construction method based on reinforcement learning, comprising: The data acquisition module provides basic data for the entire system, driving the developmental stage module to determine the key stages of crop growth. This stage information then controls the algorithm switching between the leaf area index module and the dry matter distribution module. As a core variable, the leaf area index directly affects the accumulation of total biomass. It is also input into the water balance module along with meteorological and soil data. The water stress coefficient calculated by the latter serves as a key correction factor and is fed back to the leaf area index and total biomass modules, forming a closed loop that accurately responds to environmental changes.

[0032] In this embodiment, through integrated simulation of multiple processes, precise and mechanistic prediction of peanut yield formation process is achieved. This not only provides farmers with scientific yield forecasts to assist in decision-making, but also optimizes field management measures such as irrigation by analyzing the relationship between water stress and growth and development, thereby effectively improving water resource utilization efficiency and promoting the transformation from traditional agriculture to data-driven precision agriculture. It has important value in both production guidance and scientific research.

[0033] Specifically, the process of calculating the daily thermal time characterization value of the meteorological index parameters includes: Extract temperature parameters from the target agricultural meteorological experimental station over a historical period; If the temperature parameter is less than the predetermined lower temperature threshold or greater than the predetermined upper temperature threshold, then the daily thermal time characterization value is determined to be the predetermined minimum value. If the temperature parameter is greater than or equal to the predetermined lower temperature threshold and less than or equal to the predetermined upper temperature threshold, then the daily thermal time characterization value is determined to be the product of the difference between the predetermined suitable temperature threshold and the predetermined lower temperature threshold and the predetermined temperature function.

[0034] In this embodiment, the predetermined minimum value, predetermined lower temperature threshold, predetermined upper temperature threshold, predetermined suitable lower temperature threshold, and predetermined suitable upper temperature threshold are all predetermined. The predetermined minimum value is selected in the range of [0.00, 0.15], and preferably 0.00 in this embodiment; the predetermined lower temperature threshold is selected in the range of [9.95, 10.15], and preferably 10.0℃ in this embodiment; the predetermined upper temperature threshold is selected in the range of [39.95, 40.15], and preferably 40.0℃ in this embodiment; the predetermined suitable lower temperature threshold is selected in the range of [20.95, 21.15], and preferably 21.0℃ in this embodiment; the predetermined suitable upper temperature threshold is selected in the range of [29.95, 31.15], and preferably 30.0℃ in this embodiment.

[0035] In this embodiment, the formula for calculating the daily thermal time characterization value is as follows: ; The method for calculating f(t) is as follows: When T < T b Or T > T cd When f(t) = 0; When T b ≤T≤T cd hour, ; in, DTU represents the daily thermal time characterization value; T represents the daily average temperature; T b The predetermined lower limit temperature threshold; T opt The predetermined suitable temperature threshold; T cd This is the predetermined upper limit temperature threshold.

[0036] Please see Figure 4 As shown, this is a logic diagram for determining the developmental stage of a target sample based on daily thermal time characteristics, according to an embodiment of the present invention. The present invention provides a logic method for determining the developmental stage of a target sample based on daily thermal time characteristics, including: If the daily thermal time characterization value is less than or equal to the first thermal time characterization threshold, then the target sample is determined to be from sowing to emergence stage; If the daily thermal time characterization value is greater than the predetermined first thermal time characterization threshold and less than or equal to the predetermined second thermal time characterization threshold, then the target sample is determined to be in the stage from seedling emergence to the start of seed growth. If the daily thermal time characterization value is greater than the predetermined second thermal time characterization threshold and less than or equal to the predetermined third thermal time characterization threshold, then the target sample is determined to be the stage from seed growth to leaf cessation. If the daily thermal time characterization value is greater than the predetermined third thermal time characterization threshold and less than or equal to the predetermined fourth thermal time characterization threshold, then the target sample is determined to be the leaf cessation to maturity stage. If the daily thermal time characterization value is greater than the predetermined total thermal time characterization threshold, the target sample is determined to be in the mature stage.

[0037] In this embodiment, the predetermined first thermal time characterization threshold, second thermal time characterization threshold, third thermal time characterization threshold, fourth thermal time characterization threshold, and total thermal time characterization threshold are all obtained in advance. The predetermined first thermal time characterization threshold is selected in the range [146.90, 147.15], and preferably 147.10 in this embodiment; the predetermined second thermal time characterization threshold is selected in the range [864.05, 865.15], and preferably 864.7 in this embodiment; the predetermined third thermal time characterization threshold is selected in the range [212.95, 213.15], and preferably 213.0 in this embodiment; the predetermined fourth thermal time characterization threshold is selected in the range [675.95, 676.15], and preferably 676.0 in this embodiment; the predetermined total thermal time characterization threshold is selected in the range [1899.95, 1901.15], and preferably 1900.8 in this embodiment.

[0038] In this embodiment, by setting continuous thermal time thresholds, the peanut growth cycle is precisely divided into five key developmental stages: from sowing to emergence, from emergence to seed growth, from seed growth to leaf cessation, from leaf cessation to maturity, and maturity. This constructs a phenological quantitative monitoring system based on physiological time. This method uses daily cumulative thermal time as a unified scale, effectively avoiding the developmental period prediction bias caused by interannual climate fluctuations in traditional calendar age models. It achieves dynamic, objective, and quantitative tracking of crop growth and development processes. Its clearly defined stage transition thresholds not only provide crucial stage switching signals for subsequent processes such as leaf area index calculation and dry matter distribution, ensuring the temporal accuracy of different physiological process simulations, but also intuitively generate crop development stage time-series diagrams, providing accurate time node forecasts for agricultural operations. Ultimately, this significantly improves the simulation accuracy of peanut yield formation and the timeliness and targeting of field management.

[0039] Specifically, the process of determining the leaf area index characterization value of the target sample at the developmental stage includes: If the target sample is from sowing to emergence, then the leaf area index is determined to be the predetermined minimum value. If the target sample is from seedling emergence to the start of seed growth, then the leaf area index is determined to be the sum of the daily increase in leaf area index and the leaf area index value of the previous day. If the target sample is from the beginning of seed growth to the leaf cessation stage, then the leaf area index is determined to be the product of the amount of total dry matter allocated to the leaf and the specific leaf area. If the target sample is the leaf growth cessation stage to maturity, then the leaf area index characterization value is determined to be a function of the daily thermal time characterization value and the mature leaf area index characterization value.

[0040] In practice, the predetermined minimum value is obtained in advance. The minimum value of the leaf area index characterization value within the historical period is calculated, and its average value is calculated. The predetermined minimum value is selected in the range of [0.00, 0.05]. In this embodiment, it is preferably 0.00.

[0041] In this embodiment, if the target sample is in the initial growth stage of seedlings, the formula for calculating the daily increase in leaf area index is: ; LAI on that day n =LAI n-1 +GLAI Wherein, LAI is the leaf area index; GLAI is the daily growth rate of leaf area index; PDEN is the sowing density parameter; and WSFL is the moisture correction coefficient.

[0042] In this embodiment, if the target sample is the stage from the start of seed growth to the cessation of leaf growth, the formula for calculating the daily increase in leaf area index is: ; Wherein, GLAI is the daily increase in leaf area index; GLF is the daily dry matter distribution to leaves; and SLA is the specific leaf area.

[0043] In this embodiment, if the target sample is the leaf growth cessation to maturity stage, the formula for calculating the daily increase in leaf area index is: ; Wherein, DLAI represents the daily decline in leaf area index; DTU represents the daily heat time characterization value; tuMAT–tuTSG represents the total heat time from leaf cessation to seed cessation; LAI MAX Maximum leaf area index; LAI MAT This represents the leaf area index of peanuts at maturity.

[0044] Specifically, the process of calculating daily solar radiation using the meteorological index parameters includes: Extract sunshine parameters from the target agricultural meteorological experimental station over a historical period; The daily solar radiation is calculated from the sunshine duration based on the evapotranspiration model.

[0045] In this embodiment, by utilizing historical sunshine duration data and converting solar radiation based on an evapotranspiration correlation model, not only is the data continuity of energy input factors in the agricultural hydrological model guaranteed, providing a stable and reliable data foundation for precision agricultural water management and drought assessment, but it also demonstrates the significant advantages of making full use of existing observation resources, having mature and reliable methods, and having calculation results that can directly serve core applications.

[0046] Specifically, the process of calculating the light interception rate characterization value from the extinction coefficient and the target sample leaf area index characterization value includes: Extract the characterization values ​​of the light emission coefficient and the leaf area index of the target sample; The product of the extinction coefficient and the leaf area index of the target sample is calculated as the data characterization value; The difference between the predetermined maximum value and the negative data representation value raised to the power of the natural constant is the light interception rate representation value.

[0047] In this embodiment, the formula for calculating the light interception rate is as follows: ; Among them, FINT is the light interception rate; KPAR is the extinction coefficient; and LAI is the leaf area index.

[0048] Specifically, the process of calculating the temperature correction coefficient for solar energy utilization efficiency based on the meteorological index parameters includes: Extract temperature parameters from the target agricultural meteorological experimental station over a historical period; If the temperature parameter is less than or equal to the predetermined lower temperature threshold or greater than or equal to the predetermined upper temperature threshold, then the temperature correction factor for light energy utilization is determined to be the predetermined minimum value. If the temperature parameter is greater than the predetermined lower temperature threshold and less than the predetermined suitable lower temperature threshold, then the light energy utilization temperature correction coefficient is determined as the ratio of the difference between the temperature parameter and the predetermined lower temperature threshold to the difference between the predetermined suitable lower temperature threshold and the predetermined lower temperature threshold. If the temperature parameter is greater than the predetermined upper limit threshold of suitable temperature but less than the predetermined upper limit threshold of temperature, then the temperature correction factor for light energy utilization is determined to be the ratio of the difference between the predetermined upper limit threshold of suitable temperature and the temperature parameter to the difference between the predetermined upper limit threshold of temperature and the predetermined upper limit threshold of suitable temperature. If the temperature parameter is greater than or equal to the predetermined lower limit of the suitable temperature and less than or equal to the predetermined upper limit of the suitable temperature, then the temperature correction factor for light energy utilization is determined to be the predetermined maximum value.

[0049] In this embodiment, the soil moisture stress coefficient is calculated using field management parameters, and the light energy utilization rate, temperature, and moisture correction coefficients are further derived. This cleverly couples the moisture status of the soil, plants, and atmosphere continuum into the biomass accumulation process, significantly improving the model's adaptability and accuracy in variable environments.

[0050] Specifically, the process of calculating the soil moisture stress coefficient characterization value based on field management parameters and meteorological index parameters, and calculating the light energy use efficiency moisture correction coefficient based on the soil moisture stress coefficient characterization value, includes: The soil moisture stress coefficient is equal to the light energy utilization rate moisture correction coefficient.

[0051] In this embodiment, the formula for calculating daily soil moisture content is as follows: ; Where W is the soil moisture content of the day; W0 is the soil moisture content of the previous day; P is the precipitation; G is the irrigation amount; ET is the evapotranspiration; Q is the deep infiltration; and I is the amount of precipitation intercepted by the crop canopy.

[0052] Calculation methods for each component Rainfall interception I is related to rainfall intensity and vegetation cover, and the calculation formula is: ; Among them, F c The formula for calculating vegetation cover is: ; The infiltration rate Q is calculated using an empirical formula: ; Where a and d are empirical parameters obtained from model calibration; W f W is the field holding capacity. c It is the critical water storage capacity for water exchange at the bottom of the root zone, and is related to factors such as soil water holding capacity and groundwater depth. This represents the length of the time period.

[0053] The formula for calculating evapotranspiration ET using the single-crop coefficient method is as follows: ; Where ET0 is the reference crop evapotranspiration, calculated according to the Penman-Monteith formula; K c K represents the crop coefficient, representing the seedling emergence to the beginning of seed growth stage. c The value is 0.4, indicating the seed begins to grow and leaves stop growing (K). c The value is 1.15, representing the leaf cessation and maturity stage (K). c It is 0.6.

[0054] K sThe soil moisture stress coefficient is calculated using the following formula: ; Where W represents the actual soil moisture content; W f W is the field holding capacity. p denoted as wilting humidity; b is an empirical coefficient of 0.55.

[0055] Since most items in the soil moisture balance equation are measured in millimeters, soil moisture needs to be converted to water layer thickness in millimeters. The conversion method is as follows: ; In the formula, W h Let be the water storage capacity, h be the soil layer thickness (taken as 10 cm), d be the soil bulk density of that layer, and W be the soil moisture content. The water storage capacity of each soil layer from 0 to 50 cm is calculated, and then summed to obtain the total water storage capacity of the five soil layers. Based on this dataset, the daily soil water storage capacity from seedling emergence to maturity is simulated.

[0056] In this embodiment, the formula for calculating the moisture correction factor for light energy utilization is: If W≤W p If WSFG is 0, then WSFG is 0. If 1.2W f ≥W>W p Then W S For W and W f The ratio; If W S If <0.6, then WSFG is W and W p The difference is 0.6W f and W p The ratio of the differences; If W S ≥0.6 and W S If ≤0.8, then WSFG is 1; If W S >0.8 and W S ≤1.0, WSFG is 1.2W f The difference between W and 1.2W f and 0.8W f The ratio of the differences; If W > 1.2W f When soil moisture exceeds 1.2 times field capacity, it is considered that the soil is in a saturated state, and WSFG is 0. When WSFL equals WSFG, it is assumed that the growth influence coefficient of water on leaves and biomass is the same. Where W is the moisture coefficient; WSFG is the soil moisture stress coefficient; WSFL is the temperature correction coefficient for light energy use efficiency; W p For wilting humidity; WS relative humidity of the soil; W f It refers to field water holding capacity.

[0057] Specifically, the process of calculating the total biomass characterization value based on meteorological index characterization values ​​includes: Extract daily solar radiation, light interception rate, temperature correction coefficient for light energy utilization, and moisture correction coefficient for light energy utilization. The total biomass characterization value is determined by summing the product of the daily solar radiation, light interception rate characterization value, light energy utilization rate temperature correction coefficient, and light energy utilization rate moisture correction coefficient with the predetermined biomass threshold and the daily total biomass characterization value per square meter.

[0058] In this embodiment, the formula for calculating the total biomass characterization value is as follows: ; Among them, DDMP is the total biomass per square meter per day; SRAD is the daily solar radiation; FINT is the light interception rate; and RUE is the light energy utilization rate.

[0059] In this embodiment, the leaf area index and light interception rate were calculated to accurately quantify the canopy structure and its photosynthetic capacity, providing core input for biomass simulation. The light energy utilization rate temperature correction coefficient was calculated using meteorological data, and the soil moisture stress coefficient was calculated using field management parameters, and the light energy utilization rate moisture correction coefficient was further derived. This cleverly couples the water status of the soil, plants and atmosphere continuum into the biomass accumulation process, significantly improving the model's adaptability and accuracy in variable environments.

[0060] Specifically, the process of calculating the daily biomass transfer characterization value from the daily thermal time characterization value and the total biomass characterization value includes: The ratio of the daily thermal time characterization value to the predetermined total thermal time characterization threshold is determined as the daily biomass transfer characterization value.

[0061] In this embodiment, a simulation system with a clear mechanism and accurate feedback was constructed by systematically calculating peanut growth and development parameters. The calculated meteorological index characterization value provides the energy basis for driving biophysical processes in the model. The daily thermal time characterization value was calculated and the development stage was determined accordingly, realizing the dynamic quantitative tracking of crop phenological processes and enabling the model to switch algorithms according to the growth and development stage.

[0062] Specifically, the daily seed growth characterization value is calculated based on the total biomass characterization value, the daily biomass transfer characterization value, and the food conversion coefficient. The process of calculating the final yield of the target sample using the cumulative seed dry weight characterization value and the daily seed growth characterization value includes: Extract total biomass characterization values, daily biomass transfer characterization values, and food conversion coefficient; The summation of the total biomass characterization value, the daily biomass transfer characterization value, and the food conversion coefficient is determined as the daily seed growth characterization value. The cumulative seed dry weight characterization value and the daily seed growth characterization value are summed to determine the final yield of the target sample.

[0063] In this embodiment, the formula for calculating the daily seed growth characterization value from the start of seed growth to the cessation of leaf growth is as follows: ; SGR represents the daily seed growth characterization value; DDMP represents the total biomass characterization value.

[0064] The formula for calculating the daily seed growth characterization value from leaf cessation to maturity is as follows: ; Among them, SGR is the daily seed growth characterization value; DDMP is the total biomass characterization value; TRANSL is the daily transfer characterization value, which is determined by the total biomass when leaves stop growing and the ratio of the daily thermal time characterization value to the predetermined total thermal time characterization threshold; GCC is the food conversion coefficient, which is determined to be 1.5.

[0065] In this embodiment, the formula for calculating the final yield of the target sample is: ; Among them, WGRN is the cumulative seed dry weight characterization value; SGR is the daily seed growth characterization value.

[0066] In this embodiment, the total biomass characterization value and the allocation of dry matter according to the developmental stage are used to accumulate the final yield, thus completely reproducing the physiological and ecological chain of photosynthetic product formation, allocation, and economic yield formation. The chain calculation and organic integration of parameters together realize the mechanistic, dynamic, and precise simulation of crop growth and yield formation process, providing strong decision support for yield forecasting, disaster assessment, and management optimization.

[0067] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for constructing a peanut yield simulation model based on reinforcement learning, characterized in that, include: Collect meteorological index parameters of the target agricultural meteorological experimental station within a historical period and field management parameters of the target sample within a historical period; Calculate the daily thermal time characterization value based on the aforementioned meteorological index parameters; The developmental stage of the target sample is determined based on the daily thermal time characterization values; The leaf area index characterization value of the target sample is determined based on the developmental stage of the target sample. Calculate the daily solar radiation based on the aforementioned meteorological parameters; The light interception rate was calculated based on the extinction coefficient and the leaf area index of the target sample. Calculate the temperature correction coefficient for light energy utilization based on the aforementioned meteorological index parameters; The soil moisture stress coefficient characterization value was calculated based on the meteorological index parameters and field management parameters. Calculate the light energy utilization rate moisture correction coefficient based on the soil moisture stress coefficient characterization value; The total biomass characterization value is calculated based on the daily solar radiation, light interception rate characterization value, light energy utilization rate temperature correction coefficient, and light energy utilization rate water correction coefficient. Calculate the daily biomass transfer characterization value based on the aforementioned meteorological index parameters; The daily seed growth characterization value is calculated based on the total biomass characterization value, the daily biomass transfer characterization value, and the food conversion coefficient. When the target sample is in the mature stage, the final yield of the target sample is calculated based on the cumulative seed dry weight characterization value and the daily seed growth characterization value. The meteorological parameters include temperature, sunshine, precipitation, wind speed, and relative humidity. The field management parameters include planting density, sowing date, and irrigation amount.

2. The method for constructing a peanut yield simulation model based on reinforcement learning according to claim 1, characterized in that, The process of calculating the daily thermal time characterization value based on the aforementioned meteorological index parameters includes: Extract temperature parameters from the target agricultural meteorological experimental station over a historical period; If the temperature parameter is less than the predetermined lower temperature threshold or greater than the predetermined upper temperature threshold, then the daily thermal time characterization value is determined to be the predetermined minimum value. If the temperature parameter is greater than or equal to the predetermined lower temperature threshold and less than or equal to the predetermined upper temperature threshold, then the daily thermal time characterization value is determined to be the product of the difference between the predetermined suitable temperature threshold and the predetermined lower temperature threshold and the predetermined temperature function.

3. The method for constructing a peanut yield simulation model based on reinforcement learning according to claim 2, characterized in that, The process of determining the developmental stage of the target sample based on the daily thermal time characterization values ​​includes: If the daily thermal time characterization value is less than or equal to the first thermal time characterization threshold, then the target sample is determined to be from sowing to emergence stage; If the daily thermal time characterization value is greater than the predetermined first thermal time characterization threshold and less than or equal to the predetermined second thermal time characterization threshold, then the target sample is determined to be in the stage from seedling emergence to the start of seed growth. If the daily thermal time characterization value is greater than the predetermined second thermal time characterization threshold and less than or equal to the predetermined third thermal time characterization threshold, then the target sample is determined to be the stage from the start of seed growth to the cessation of leaf growth. If the daily thermal time characterization value is greater than the predetermined third thermal time characterization threshold and less than or equal to the predetermined fourth thermal time characterization threshold, then the target sample is determined to be the leaf cessation to maturity stage. If the daily thermal time characterization value is greater than the predetermined total thermal time characterization threshold, the target sample is determined to be in the mature stage.

4. The method for constructing a peanut yield simulation model based on reinforcement learning according to claim 3, characterized in that, The process of determining the leaf area index characterization value of the target sample based on the developmental stage of the target sample includes: If the target sample is from sowing to emergence, then the leaf area index is determined to be the predetermined minimum value. If the target sample is from seedling emergence to the start of seed growth, then the leaf area index is determined to be the sum of the daily increase in leaf area index and the leaf area index value of the previous day. If the target sample is from the beginning of seed growth to the leaf cessation stage, then the leaf area index is determined to be the product of the amount of total dry matter allocated to the leaf and the specific leaf area. If the target sample is the leaf growth cessation stage to maturity, then the leaf area index characterization value is determined to be a function of the daily thermal time characterization value and the mature leaf area index characterization value.

5. The method for constructing a peanut yield simulation model based on reinforcement learning according to claim 4, characterized in that, The process of calculating daily solar radiation based on the aforementioned meteorological parameters includes: Extract sunshine parameters from the target agricultural meteorological experimental station over a historical period; The daily solar radiation is calculated from the sunshine duration based on the evapotranspiration model.

6. The method for constructing a peanut yield simulation model based on reinforcement learning according to claim 5, characterized in that, The process of calculating the light interception rate characterization value based on the extinction coefficient and the leaf area index characterization value of the target sample includes: Extract the characterization values ​​of the light emission coefficient and the leaf area index of the target sample; The product of the extinction coefficient and the leaf area index of the target sample is calculated as the data characterization value; The difference between the predetermined maximum value and the negative data representation value raised to the power of the natural constant is the light interception rate representation value.

7. The method for constructing a peanut yield simulation model based on reinforcement learning according to claim 6, characterized in that, The process of calculating the temperature correction coefficient for light energy utilization based on the aforementioned meteorological index parameters includes: Extract temperature parameters from the target agricultural meteorological experimental station over a historical period; If the temperature parameter is less than or equal to the predetermined lower temperature threshold or greater than or equal to the predetermined upper temperature threshold, then the temperature correction factor for light energy utilization is determined to be the predetermined minimum value. If the temperature parameter is greater than the predetermined lower temperature threshold and less than the predetermined suitable lower temperature threshold, then the light energy utilization temperature correction coefficient is determined as the ratio of the difference between the temperature parameter and the predetermined lower temperature threshold to the difference between the predetermined suitable lower temperature threshold and the predetermined lower temperature threshold. If the temperature parameter is greater than the predetermined upper limit threshold of suitable temperature but less than the predetermined upper limit threshold of temperature, then the temperature correction factor for light energy utilization is determined to be the ratio of the difference between the predetermined upper limit threshold of suitable temperature and the temperature parameter to the difference between the predetermined upper limit threshold of temperature and the predetermined upper limit threshold of suitable temperature. If the temperature parameter is greater than or equal to the predetermined lower limit of the suitable temperature and less than or equal to the predetermined upper limit of the suitable temperature, then the temperature correction factor for light energy utilization is determined to be the predetermined maximum value.

8. The method for constructing a peanut yield simulation model based on reinforcement learning according to claim 7, characterized in that, The process of calculating the soil moisture stress coefficient characterization value based on field management parameters and meteorological index parameters, and calculating the light energy use efficiency moisture correction coefficient based on the soil moisture stress coefficient characterization value, includes: The soil moisture stress coefficient is equal to the light energy utilization rate moisture correction coefficient.

9. The method for constructing a peanut yield simulation model based on reinforcement learning according to claim 8, characterized in that, The process of calculating the total biomass characterization value based on meteorological index characterization values ​​includes: Extract daily solar radiation, light interception rate, temperature correction coefficient for light energy utilization, and moisture correction coefficient for light energy utilization. The total biomass characterization value is determined by summing the products of the daily solar radiation, light interception rate characterization value, light energy utilization rate temperature correction coefficient, and light energy utilization rate water correction coefficient with the predetermined biomass threshold and the daily total biomass characterization value per square meter. The process of calculating the daily biomass transfer characterization value based on the daily thermal time characterization value and the total biomass characterization value includes: The ratio of the daily thermal time characterization value to the predetermined total thermal time characterization threshold is determined as the daily biomass transfer characterization value.

10. The method for constructing a peanut yield simulation model based on reinforcement learning according to claim 9, characterized in that, The process of calculating the daily seed growth characterization value based on the total biomass characterization value, the daily biomass transfer characterization value, and the food conversion coefficient, and then calculating the final yield of the target sample based on the cumulative seed dry weight characterization value and the daily seed growth characterization value includes: Extract total biomass characterization values, daily biomass transfer characterization values, and food conversion coefficient; The summation of the total biomass characterization value, the daily biomass transfer characterization value, and the food conversion coefficient is determined as the daily seed growth characterization value. The cumulative seed dry weight characterization value and the daily seed growth characterization value are summed to determine the final yield of the target sample.

Images