Seed cultivation process optimization method based on discrete element theory

Abstract

The invention discloses a seed cultivation process optimization method based on a discrete element theory, and belongs to the technical field of seed cultivation. The method comprises the following steps: establishing a mechanical interaction model between soil particles and seeds according to an acquired physical characteristic parameter set by utilizing a discrete element method; acquiring a target physical characteristic parameter set, inputting the target physical characteristic parameter set into a mechanical interaction model, and performing discrete element analogue simulation output under a dynamic simulation condition to obtain a change curve of contact force along with seed growth displacement; and judging whether the contact force in the change curve meets a preset threshold value in each stage of seed growth, if not, continuously self-optimizing the target physical characteristic parameter set, and ending optimization until the judgment result is yes. According to the method, a seed optimization design scheme can be generated, improvement strategies for different soil types can be formulated, and a reliable basis is provided for agricultural machinery design optimization and seeding efficiency improvement.

Description

Technical Field

[0001] This invention belongs to the field of seed breeding technology, specifically relating to a method for optimizing the seed breeding process based on discrete element theory. Background Technology

[0002] The Discrete Element Method (DEM) has made significant progress in the study of seed emergence mechanics and soil dynamics, especially in seed-soil interaction modeling and agricultural machinery optimization design. However, existing research still faces some key challenges, mainly focusing on dynamic environment simulation, parameter calibration methods, multi-scale modeling, and the integration of complex environmental factors.

[0003] Research on dynamic environments is still insufficient. Although some studies have analyzed the effects of shovel-type rolling contact components on soil resistance and micro-topography preparation at different operating speeds using the DEM-MBD coupled model and verified the accuracy of the model through experiments, the influence of dynamic conditions such as changes in soil moisture content and external disturbances on seed breaking-out mechanical behavior still lacks systematic discussion.

[0004] Limitations in parameter calibration and validation remain a significant challenge in current research. Existing work primarily calibrates seed-soil interaction parameters through collision tests, static friction tests, and rolling friction tests, yielding relatively accurate results. For the contact process between maize seeds and soils with varying moisture contents, researchers optimized the parameters using a central composite design method, achieving an error control within 5.4%. However, experimental validation and parameter calibration methods are still insufficient for modeling soil particle bonding characteristics and particle breakage behavior, making it difficult to fully reflect the actual physical and mechanical properties of soil.

[0005] From the perspective of scale modeling, the coupling mechanism between microscopic and macroscopic mechanical responses has not been fully elucidated. Current research mainly focuses on single-scale modeling and analysis, failing to effectively integrate microscopic particle behavior with macroscopic mechanical properties. Existing studies have emphasized the importance of soil particle shape and bonding characteristics in modeling accuracy, proposing the construction of contact mechanics models that better reflect actual particle morphology. Meanwhile, preliminary results on the competition mechanism between root penetration and soil pressure indicate a significant negative correlation between soil stiffness and root penetration depth, but a systematic analysis of the multi-scale coupling behavior during seed emergence remains lacking.

[0006] Regarding the integration of computational efficiency and environmental factors, the high accuracy advantage of DEM simulation is limited by issues such as long computation time and complex parameter optimization. Some studies have proposed introducing machine learning algorithms to improve the efficiency of parameter calibration and developing extended particle models and post-processing techniques to enhance the computational performance of large-scale simulations. Furthermore, the impact of environmental factors such as air, moisture, and carbon storage on soil dynamics has not been thoroughly considered; further refinement of 3D DEM models provides direction for future in-depth research. Summary of the Invention

[0007] In view of this, the purpose of this invention is to provide a seed breeding process optimization method based on discrete element theory to solve the above-mentioned technical problems.

[0008] To achieve the above objectives, the present invention provides the following technical solution: An optimization method for seed breeding process based on discrete element theory, comprising: Using the discrete element method, a mechanical interaction model between soil particles and seeds is established based on the collected set of physical characteristic parameters. The set of physical characteristic parameters includes: soil physical characteristic parameters under different soil environments, seed physical characteristic parameters of different seed types, and mechanical behavior data of any seed type during the breaking process in different soil environments. The target physical property parameter set is obtained and input into the mechanical interaction model. Under dynamic simulation conditions, the output of the contact force variation curve with seed growth displacement is obtained through discrete element simulation. The target physical property parameter set includes: soil physical property parameters under the target soil environment and seed physical property parameters of the target seed category. Determine whether the contact force in the change curve meets the preset threshold at each stage of seed growth. If the determination result is no, continue to self-optimize the target physical characteristic parameter set until the determination result is yes, and then end the optimization.

[0009] Furthermore, the soil physical property parameters include: the soil's shear strength, internal friction angle, and compressibility measured by direct shear test and triaxial compression test, as well as the soil particle size distribution, porosity, and permeability measured by particle size analyzer and permeability test device. Seed physical property parameters include: external geometric properties, internal tissue structure, and internal structural mechanical properties of seeds, measured using microscopic imaging equipment and mechanical testing equipment. Among them, the internal structural mechanical properties include the seed's elastic modulus, seed coat thickness, and rupture pressure. The process of acquiring mechanical behavior data during seed emergence includes: designing emergence experiments under different soil conditions, using a loading device to measure the emergence pressure and radicle extension force of different seed types, recording the seed emergence process using a high-speed camera, and analyzing the recorded results to obtain mechanical behavior data during seed emergence; among which, the soil environment includes compaction parameters and moisture parameters.

[0010] Furthermore, a comprehensive soil mechanical behavior model is generated by using soil physical property parameters; A seed model is constructed by using the physical property parameters of the seeds; By integrating the mechanical behavior model, the seed model, and the mechanical behavior data of seeds during the seed emergence process corresponding to the soil environment and seed type, the discrete element method is used to establish the corresponding mechanical interaction model between soil particles and seeds.

[0011] Furthermore, in the integrated soil mechanical behavior model, the particle size distribution of soil particles is determined by the particle size distribution curve of the actual soil, and different sizes of particles need to be generated according to the actual proportion of different soil types to simulate the real soil structure during modeling; among them, the normal contact force between soil particles is described by the Hertz-Mundling contact model, and the tangential contact force is described by the Coulomb friction model.

[0012] Furthermore, during discrete element simulation, GPU acceleration and distributed computing technologies are used to distribute the entire discrete element simulation computation task across multiple computing nodes on the GPU platform, and data communication and collaborative computation between computing nodes are achieved through message passing interfaces.

[0013] Furthermore, the continuously self-optimizing target physical property parameter set includes design schemes for optimizing seed physical property parameters and improvement strategies for optimizing soil physical property parameters; Among them, the optimization logic of the design scheme and improvement strategy in the optimization process is related to the parameter sensitivity of seed physical property parameters and soil physical property parameters.

[0014] Furthermore, a prediction model based on support vector machine or random forest algorithm is constructed, taking seed physical property parameters and soil physical property parameters as inputs and seed germination performance index as output; among which, seed germination performance index is related to germination time and germination force. By modifying the input parameters, different seed emergence performance indices are obtained, which are then used in conjunction with the Sobol' global sensitivity analysis method to evaluate the parameter sensitivity of any parameter among the seed physical property parameters and soil physical property parameters.

[0015] Furthermore, the process of constructing the mechanical interaction model includes constructing the initial mechanical interaction model and calibrating the final mechanical interaction model; The final mechanical interaction model obtained through calibration includes: Discrete element simulation is performed using the initial mechanical interaction model. The simulation data is then compared with the actual experimental data. Based on the comparison results, the simulation model parameters of the initial mechanical interaction model are calibrated until the comparison results meet the preset calibration conditions, thus obtaining the final mechanical interaction model.

[0016] Furthermore, the continuously self-optimizing target physical property parameter set includes design schemes for optimizing seed physical property parameters and improvement strategies for optimizing soil physical property parameters, including: Obtain the parameter sensitivity of all optimizable parameters, and the parameter modification cost coefficient of each specific parameter; where the sum of the parameter modification cost coefficients of all specific parameters is 1. Obtain the parameter matching degree for each specific modified parameter; Wherein, parameter matching degree = parameter sensitivity * (1 - parameter modification cost coefficient); All specific modified parameters are arranged into a scheduling sequence from largest to smallest according to the parameter matching degree; Build a design scheme pool by obtaining specific parameters from the target physical property parameter set; The specific parameters are scheduled sequentially according to the scheduling sequence. Data of the same type in the design scheme pool are updated. It is determined whether the contact force in the change curve generated after each update meets the preset threshold at each stage of seed growth after discrete element simulation. The optimization ends when the threshold is met. All specific parameters in the latest design scheme pool are integrated as a new set of target physical property parameters to generate optimized seed design schemes and soil improvement strategies.

[0017] Furthermore, when multiple specific modified parameters are scheduled from the scheduling sequence each time, a low-cost parameter rollback mechanism is triggered, which specifically includes the following steps: Step 1: Store all the specific modified parameters for this scheduling in the temporary storage area. In the temporary storage area, perform a search for parameters of the same type for all the specific modified parameters to obtain the search results. Step 2: When the search results show that there are parameters of the same type, compare the parameter matching degree of all parameters of the same type, mark the parameters of the same type with low matching degree as rollback, and extract all the specific modified parameters that do not have rollback marks in the temporary storage area to generate the update file. Step 3: Update the data of the same type in the design scheme pool according to the update file, and then input the updated data into the mechanical interaction model for discrete element simulation to obtain the simulation results. At the same time, record the update number as 1. The simulation results are the curves of contact force changing with seed growth displacement. Step 4: If all stages of the simulation results meet the preset threshold, proceed to step 5. If none of the stages of the simulation results meet the preset threshold and the update count is 1, proceed to step 6. If none of the stages of the simulation results meet the preset threshold and the update count is not 1, proceed to step 7. Step 5: Extract the specific modified parameters that have rollback flags and the highest parameter matching degree from the temporary storage area, generate a new update file, set a rollback mechanism for the design scheme pool, and perform an update count increment operation, then jump to step 3. Step 6: Reset the temporary storage area and update count, perform the next scheduling according to the scheduling sequence, and jump to step 1; Step 7: Based on the rollback mechanism in Step 5, rollback and update the design scheme pool. Use all the specific parameters in the updated design scheme pool as the new target physical property parameter set to generate the optimized seed design scheme and soil improvement strategy.

[0018] The beneficial effects of this invention are as follows: This invention proposes a seed cultivation process optimization method based on discrete element theory, focusing on the mechanical behavior of seed-soil interaction under dynamic conditions. It combines high-precision experiments with machine learning methods to optimize parameter calibration technology, breaking through the bottleneck of coupled modeling at the micro and macro scales. At the same time, it integrates environmental factors such as air and water to comprehensively reveal the mechanical mechanism of seed emergence, providing theoretical basis and practical guidance for agricultural machinery design optimization and sowing efficiency improvement.

[0019] Other advantages, objectives, and features of the invention will be set forth in the following description and will be apparent to those skilled in the art in some respects, or may be learned by practice of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0020] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0021] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart of a seed cultivation process optimization method based on discrete element theory in an embodiment of the present invention. Figure 2 This is a flowchart of a low-cost optimization method in a seed breeding process optimization method based on discrete element theory, as described in an embodiment of the present invention. Figure 3This is a flowchart of a low-cost parameter rollback mechanism in a seed breeding process optimization method based on discrete element theory, as described in an embodiment of the present invention. Detailed Implementation

[0022] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0023] like Figure 1 As shown, this invention proposes a seed cultivation process optimization method based on discrete element theory, comprising: S101. Using the discrete element method, a mechanical interaction model between soil particles and seeds is established based on the collected set of physical characteristic parameters. The set of physical characteristic parameters includes: soil physical characteristic parameters under different soil environments, seed physical characteristic parameters of different seed types, and mechanical behavior data of any seed type during the breaking process in different soil environments. S102. Obtain the target physical property parameter set, input it into the mechanical interaction model, and output the curve of contact force changing with seed growth displacement through discrete element simulation under dynamic simulation conditions; wherein, the target physical property parameter set includes: soil physical property parameters under the target soil environment and seed physical property parameters of the target seed type; S103. Determine whether the contact force in the change curve meets the preset threshold at each stage of seed growth. If the determination result is no, continue to self-optimize the target physical characteristic parameter set until the determination result is yes and then end the optimization. The working principle of the above technical solution is as follows: The seed cultivation process optimization method implemented in this application is mainly divided into three steps, including the construction of the discrete element model, the application of the discrete element model, and the optimization of the soil breaking strategy; The first step specifically includes seed structure and morphology analysis, soil physical structure and mechanical parameters research, seed-soil interaction mechanism analysis, and model construction based on the above analysis results using the discrete element method. The second step specifically includes acquiring target parameters, setting dynamic simulation conditions, and discrete element simulation; the dynamic simulation conditions include, but are not limited to, setting different humidity, rainfall, and mechanical disturbance conditions. The third step specifically includes judging the simulation results and optimizing the parameters based on the judgment results; Specifically, firstly; After collecting a set of physical property parameters that meet the target requirements and quantity, a mechanical interaction model between soil particles and seeds is established using the Discrete Element Method (DEM). This model simulates the force distribution of seeds under different environmental conditions, the flow of soil particles, and the pressure transmission law, revealing the changes in key mechanical parameters during seed emergence. During model establishment, parameter inputs include soil compaction, porosity, particle size distribution, and friction coefficient, as well as seed elastic modulus, seed coat thickness, and fracture strength, calibrated experimentally. This ensures the accuracy of the model inputs and the reliability of the simulation results. Similarly, dynamic simulation conditions need to be set during the simulation model construction process to ensure the simulation accuracy when using the model for subsequent simulations. After completing the training of the mechanical interaction model, the target physical property parameter set is obtained and input into the mechanical interaction model. Under dynamic simulation conditions, the output of the contact force variation curve with seed growth displacement is obtained through discrete element simulation. The parameter acquisition method in the target physical property parameter set corresponds to the parameter acquisition method used in the aforementioned model training. Soil physical property parameters can be obtained by methods such as direct shear test and triaxial compression test, while seed physical property parameters can be obtained by methods such as microscopic imaging and mechanical testing equipment observation. It is worth noting that the curve of contact force change with seed growth displacement is mainly used to reflect the force relationship between the seed and soil particles during the seed emergence process. For example, during the elongation of the seed radicle, the normal contact force between the radicle and the surrounding soil particles will change continuously. This change is related to the growth rate of the radicle and the compaction of the soil. Meanwhile, the pressure distribution and mechanical behavior of seeds under different soil compaction, friction coefficient, and humidity conditions were analyzed to further reveal the fluidity of soil particles. Combined with dynamic simulation and experiments, the effects of environmental variables such as rainfall and mechanical disturbance on soil particle fluidity, compaction, and friction characteristics were studied, and the dynamic effect of humidity changes on seed emergence behavior was quantified. Through artificial rainfall and mechanical disturbance experiments, the dynamic response of seeds during the emergence process was captured, revealing the adaptive laws of seed-soil interaction under complex environments, and providing theoretical support for solving agricultural planting problems under dynamic conditions. Secondly, the advantages of using the discrete element method are: in soil, force chains are formed between particles; the growth of seeds will interfere with and disrupt the surrounding force chains, and the discrete element method can simulate the formation and change process of force chains; for example, when the radicle of a seed grows, it will break the original soil force chain, and the process of breaking and rebuilding the force chain has an important impact on the resistance and energy consumption of seed breaking through the soil. After obtaining the simulation results, the contact force in the variation curve at each stage of seed growth was analyzed to obtain the analysis results based on seed structure and soil physical properties. Methods for optimizing seed shape, surface characteristics and internal structure design for different soil types were proposed. At the same time, specific countermeasures to improve the soil environment (such as controlling humidity and reducing density) were proposed to improve seed breaking efficiency and germination success rate, providing a scientific basis for the development of modern agricultural planting technology. The beneficial effects of the above technical solutions are as follows: 1. By simulating and experimenting, the comprehensive influence of dynamic environments such as rainfall and mechanical disturbance on soil particle mobility and seed mechanical properties is studied, quantifying the regulatory role of environmental variables on seed germination efficiency and solving the problem of insufficient seed germination adaptability in complex environments; 2. For the stress concentration problem during seed germination, the stress distribution and crack evolution law between seeds and soil are studied, a stress distribution model (mechanical interaction model) of the seed-soil system is constructed, and the key mechanical factors in the germination process are revealed; 3. Based on the comprehensive consideration of seed morphology, mechanical properties and soil physical properties, a design scheme for optimizing seed shape, hardness and surface properties is proposed, and improvement strategies for different soil types are formulated, including controlling humidity and reducing density, to improve seed germination rate and growth efficiency, providing technical support for agricultural production practices.

[0024] In one embodiment, the soil physical property parameters include: the soil's shear strength, internal friction angle, and compressibility as measured by direct shear test and triaxial compression test, and the soil particle size distribution, porosity, and permeability as measured by particle size analyzer and permeability test device. Seed physical property parameters include: external geometric properties, internal tissue structure, and internal structural mechanical properties of seeds, measured using microscopic imaging equipment and mechanical testing equipment. Among them, the internal structural mechanical properties include the seed's elastic modulus, seed coat thickness, and rupture pressure. The process of acquiring mechanical behavior data during seed emergence includes: designing emergence experiments under different soil conditions, using a loading device to measure the emergence pressure and radicle extension force of different seed types, recording the seed emergence process using a high-speed camera, and analyzing the mechanical behavior data during seed emergence based on the recorded results; among which, the soil environment includes compaction parameters and moisture parameters. The working principle of the above technical solution is as follows: It studies the physical properties of different soil types, such as particle morphology, particle size distribution, porosity, friction coefficient, and cohesion; it obtains the soil's shear strength, compressibility, and permeability through particle size analyzers and direct shear tests; it systematically observes the external morphology of seeds, including size, shape, color, and surface structure characteristics, and obtains seed surface geometric parameters using three-dimensional scanning technology (3D scanner); it microscopically observes the internal tissue structure of seeds (seed coat, embryo, endosperm) and their mechanical action during seed emergence; it measures the elastic modulus, breaking strength, and thickness of the seed coat through compression and tensile tests, analyzes the seed's adaptability to different soil physical environments and seed emergence strategies, and provides a basis for numerical simulation and experimental research on seed mechanical properties; when microscopically observing the mechanical action of seeds during seed emergence, it also designs seed emergence experiments under different soil environments, uses a loading device to measure the seed emergence pressure and radicle extension force of different seed types, and records the seed emergence process using a high-speed camera; it analyzes the recorded results to obtain mechanical behavior data of seeds during seed emergence, where the soil environment is related to the aforementioned dynamic simulation conditions; The beneficial effects of the above technical solution are: through reasonable data collection, it can provide reliable data support for subsequent numerical simulation and optimization of the seed breeding process.

[0025] In one embodiment, a comprehensive soil mechanical behavior model is generated by using soil physical property parameters; A seed model is constructed by using the physical property parameters of the seeds; By integrating the mechanical behavior model, the seed model, and the mechanical behavior data of seeds during the seed emergence process corresponding to the soil environment and seed type, the discrete element method is used to establish the corresponding mechanical interaction model between soil particles and seeds. The working principle of the above technical solution is as follows: The external morphology of seeds, including seed size, shape, color, and surface structure features, is obtained through systematic observation. Three-dimensional scanning technology is used to acquire the geometric parameters of the seed surface, and the external morphology is integrated to generate external seed data. The internal tissue structure of the seed is obtained through microscopic observation, including the seed coat, embryo, and endosperm. Based on the external seed data and internal tissue structure, a seed model is constructed and assigned corresponding physical properties. Through multi-point testing of stratified soil samples, the stress change law of soil under external dynamic influences in the natural environment is analyzed, and a comprehensive soil mechanical behavior model is constructed. Finally, using the comprehensive mechanical behavior model, the seed model, and the mechanical behavior data of seeds during the germination process corresponding to the soil environment and seed type, the discrete element method is used to establish a corresponding mechanical interaction model between soil particles and seeds. The beneficial effects of the above technical solution are as follows: by establishing the above technical solution, a corresponding mechanical interaction model between soil particles and seeds is obtained, which is beneficial to providing a reliable simulation model for subsequent discrete element simulation.

[0026] In one embodiment, the particle size distribution of soil particles in the integrated soil mechanical behavior model is determined by the particle size distribution curve of the actual soil, and different sizes of particles need to be generated according to the actual proportion of different soil types to simulate the real soil structure during modeling; wherein, the normal contact force between soil particles is described by the Hertz-Mundling contact model, and the tangential contact force is described by the Coulomb friction model. The working principle and beneficial effects of the above technical solution are as follows: Soil is a granular material composed of particles of different sizes. The particle size distribution of soil particles in the discrete element model should be determined according to the particle size distribution curve of the actual soil. For example, for loam, which contains a certain proportion of sand, silt and clay, different sizes of particles should be generated according to their actual proportions to simulate the real soil structure during modeling. The Hertz-Mindlin contact model is usually used to describe the normal contact force between particles, taking into account the elastic deformation of the particles. For tangential contact force, the Coulomb friction model can be used to set the friction coefficient between soil particles. This coefficient is related to the texture of the soil. For example, the friction coefficient of sand is relatively small, while the friction coefficient of clay is large.

[0027] In one embodiment, when performing discrete element simulation, GPU acceleration technology and distributed computing technology are used to distribute the entire discrete element simulation computation task to multiple computing nodes on the GPU platform, and data communication and collaborative computing between computing nodes are realized through message passing interface. The working principle and beneficial effects of the above technical solution are as follows: The study investigates the influence of physical properties such as soil particle distribution, density, porosity, and friction coefficient on seed germination efficiency. The seed germination process involves a large number of soil particles and complex interactions, resulting in a huge computational load for discrete element simulation. Graphics processing unit (GPU) acceleration technology is employed. GPUs possess powerful parallel computing capabilities, allowing the discrete element calculation task to be ported to the GPU platform. By optimizing the parallelism of the algorithm, rapid simulation of large-scale particle systems can be achieved. Meanwhile, by combining distributed computing technology, the entire simulation task is distributed across multiple computing nodes (such as computer clusters), and data communication and collaborative computing between nodes are realized through message passing interface (MPI), further improving computing efficiency; and revealing the interaction patterns between seeds and soil under different soil conditions; Furthermore, to help those skilled in the art better understand the proposed soil structure optimization scheme and the idea of ​​improving seed adaptability and germination efficiency, this scheme provides some simulation functions for discrete element simulation, mainly including geometric model creation, particle generation and distribution, parameter setting, and result analysis and visualization; among which, Geometric model creation includes: PFC software provides a wealth of tools and commands that can help users easily create various complex geometric models, such as the arrangement and combination of particles, the setting of boundary conditions, etc., so as to accurately simulate the geometric shape of the actual physical system and lay the foundation for subsequent discrete element analysis.

[0028] Particle generation and distribution include: rapidly generating a large number of discrete particles according to user-defined parameters, such as particle size, shape, and density, and realizing their random or regular distribution in the model space to simulate different types of particulate matter systems, such as soil particles, grain particles, etc.

[0029] Parameter settings include contact model selection and material parameter assignment. Specifically, Contact model selection includes: the software has built-in a variety of contact models, such as elastic contact model, plastic contact model, viscous contact model, etc. Users can select the appropriate contact model to describe the interaction between particles and between particles and boundaries according to the physical characteristics of the simulated object and the actual working conditions, so as to more accurately reflect the mechanical behavior of the system. Material parameter assignment includes: allowing users to assign various material property parameters to particles and contacts, such as elastic modulus, Poisson's ratio, coefficient of friction, and bond strength. By accurately setting these parameters, the discrete element model can better approximate the mechanical properties of real materials, thereby improving the reliability of simulation results.

[0030] Results analysis and visualization include data output and processing, as well as visualization presentation. Specifically, Data output and processing include: outputting rich simulation results data, such as particle position, velocity, force, stress tensor, etc. Users can further process and analyze this data as needed to extract valuable information, such as the average velocity of particles, stress distribution law, energy dissipation, etc., to provide data support for in-depth research on the mechanical behavior of particle systems. Visualization includes: displaying simulation results in two-dimensional or three-dimensional graphics through an intuitive graphical interface and powerful visualization functions, such as particle motion trajectories, stress cloud diagrams, velocity vector diagrams, etc., enabling users to more intuitively observe and understand the evolution process and mechanical state of the particle system, which helps to discover patterns, explain phenomena, and verify theories.

[0031] In one embodiment, the continuously self-optimizing target physical property parameter set includes a design scheme for optimizing seed physical property parameters and an improvement strategy for optimizing soil physical property parameters. Among them, the optimization logic of the design scheme and improvement strategy in the optimization process is related to the parameter sensitivity of seed physical property parameters and soil physical property parameters; The methods for determining the parameter sensitivity of seed physical property parameters and soil physical property parameters include: A prediction model based on support vector machine or random forest algorithm is constructed, taking seed physical property parameters and soil physical property parameters as inputs and seed germination performance index as output; among which, the seed germination performance index is related to germination time and germination force. By modifying the input parameters, different seed emergence performance indices are obtained, which are then used in conjunction with the Sobol's global sensitivity analysis method to evaluate the parameter sensitivity of any parameter among the seed physical property parameters and soil physical property parameters; where, the higher the parameter sensitivity, the more important it is to the seed emergence performance. The working principle and beneficial effects of the above technical solution are as follows: There are numerous physical and mechanical parameters of seeds and soil, and there are complex coupling relationships between them. Feature selection algorithms in machine learning (such as recursive feature elimination) are used to screen parameters that have an important impact on seed germination performance. A prediction model based on algorithms such as support vector machine (SVM) or random forest (RF) is constructed. Seed and soil parameters are used as inputs, and seed germination performance indicators (such as germination time and germination force) are used as outputs. The sensitivity of parameters is analyzed by evaluating the feature importance of the model. This method can effectively handle the nonlinear relationship between parameters and improve the accuracy of sensitivity analysis. Global sensitivity analysis methods such as Sobol' are used to comprehensively evaluate the impact of parameters on seed emergence performance across the entire value space, considering the range and distribution of parameter values. By calculating the first-order and total-order sensitivity indices of the parameters, the contribution of each parameter and the interaction between parameters to seed emergence performance is determined, providing a basis for model simplification and parameter optimization. Single-factor sensitivity analysis: Only one parameter (such as the seed's elastic modulus or the soil's friction coefficient) is changed each time, while other parameters remain constant. The effect of this parameter change on the seed germination process (such as germination time and germination force) is observed. For example, by gradually increasing the soil friction coefficient, the changing trends of seed germination force and germination time are analyzed to determine the sensitivity of this parameter. Multifactor sensitivity analysis: Considering the combined impact of multiple parameters changing simultaneously on seed emergence, orthogonal design or Latin hypercube sampling can be used to combine and change multiple parameters to determine the relative importance and interaction relationship of each parameter.

[0032] In one embodiment, the process of constructing a mechanical interaction model includes constructing an initial mechanical interaction model and calibrating a final mechanical interaction model. The final mechanical interaction model obtained through calibration includes: Discrete element simulation is performed using the initial mechanical interaction model. The simulation data is compared with the real experimental data. Based on the comparison results, the simulation model parameters of the initial mechanical interaction model are calibrated until the comparison results meet the preset calibration conditions, and the final mechanical interaction model is obtained. The working principle and beneficial effects of the above technical solution are as follows: Before model verification, tools such as MATLAB, Python, or ORIGIN are used to clean and statistically analyze the experimental and simulation results, establishing a quantitative relationship between seed emergence efficiency and soil physical parameters; then, the simulation model parameters are calibrated using experimental results, adjusting the simulation accuracy of soil particle-seed contact characteristics, and verifying the applicability of the model under various environmental conditions, i.e., comparing the discrete element simulation results with experimental data, comparing key parameters such as seed growth status and force magnitude; if there are differences between the simulation results and experimental data, the reasons for the differences are analyzed, which may be unreasonable model assumptions, inaccurate parameters, etc.; by continuously adjusting the model and parameters, the simulation results are made to better match the experimental results, thereby verifying the effectiveness of the discrete element model.

[0033] like Figure 2 As shown, in one embodiment, the continuously self-optimizing target physical property parameter set includes a design scheme for optimizing seed physical property parameters and an improvement strategy for optimizing soil physical property parameters, including: S201. Obtain the parameter sensitivity of all optimizable parameters, and the parameter modification cost coefficient of each specific parameter; wherein, the sum of the parameter modification cost coefficients of all specific parameters is 1. S202. Obtain the parameter matching degree for each specific modified parameter; Wherein, parameter matching degree = parameter sensitivity * (1 - parameter modification cost coefficient); S203. Arrange all the specific modified parameters into a scheduling sequence from largest to smallest according to the parameter matching degree; S204. Obtain the specific parameters from the target physical property parameter set to construct a design scheme pool; S205. Schedule the specific modified parameters in sequence according to the scheduling sequence, and update the data of the same type in the design scheme pool; S206. Determine whether the contact force in the generated change curve after each update of the data meets the preset threshold at each stage of seed growth, until the optimization ends when the threshold is met. S207. Integrate all specific parameters from the latest design scheme pool as a new set of target physical property parameters to generate optimized seed design schemes and soil improvement strategies; The working principle of the above technical solution is as follows: When optimizing parameters, there are usually three methods. The first is to optimize only the seed parameters to meet the requirements for germination. The second is to optimize only the soil parameters to meet the requirements for germination. However, the above two methods often result in over-optimization. To address this, the existing method is to use the third method to comprehensively optimize both seed and soil parameters simultaneously, so that the optimized solution can perfectly match the actual germination requirements. However, the above methods are often only available in the laboratory, that is, experiments are conducted with a small number of samples as the standard, without considering the cost issues of optimization when facing a large number of seeds or soil in actual applications. Based on this, this application proposes a low-cost optimization method for situations involving a large number of seeds or soil, which comprehensively improves the planting efficiency of the agricultural industry. Specifically, the method proposed in this application includes: Obtain the parameter sensitivity of all optimizable parameters; where parameters refer to the categories of seed physical property parameters and soil physical property parameters stored in the database beforehand, and optimizable parameters refer to the categories of parameters among seed physical property parameters and soil physical property parameters that can be modified manually. Obtain the parameter modification cost coefficient for the specific modified parameter; where the parameter here refers to one of the optimizable parameters that can be modified manually, the specific modified parameter refers to the parameter involving the specific modified value, and the parameter modification cost coefficient refers to the weight of the modification cost corresponding to the current specific modified parameter relative to the modification cost of all specific modified parameters. It is worth noting that, for the convenience of calculation and subsequent sequence construction, the parameter modification cost coefficient of all specific modified parameters is summed to 1. After obtaining the above two data points, the parameter matching degree is calculated for each specific modified parameter. The calculation method is: Parameter matching degree = Parameter sensitivity * (1 - Parameter modification cost coefficient). It is worth noting that the calculation here is dimensionless. Since the higher the parameter sensitivity, the higher the importance of the parameter category in the simulation, and the lower the parameter modification cost coefficient, the lower the cost of optimizing the specific parameter, this scheme sets the parameter sensitivity and parameter modification cost coefficient to be of equal importance in order to simplify the calculation process. Therefore, the above calculation process is used to determine the parameter matching degree. However, this scheme is mainly used to limit the definition of parameter matching degree. The calculation of parameter matching degree is not limited to the above calculation method. When the importance of the parameter modification cost coefficient is lower than that of the parameter sensitivity, an exponential operation method can also be used for calculation, such as: Parameter matching degree = Parameter sensitivity * e^(-Parameter modification cost coefficient). After determining the parameter matching degree, all specific modified parameters are arranged into a scheduling sequence from largest to smallest according to the parameter matching degree; Simultaneously, specific parameters from the target physical property parameter set are used to construct a design scheme pool; The specific parameters are scheduled sequentially according to the scheduling sequence. Data of the same type in the design scheme pool are updated. There is no limit to the number of data to be scheduled in each update. It is usually set to complete the data scheduling in an exponential decreasing trend. For example, 8 data are scheduled in the first update, 4 in the second, 2 in the third, and 1 in each subsequent update. This is to reduce the number of simulation calculations while meeting the design requirements of the lowest cost optimization. After each update, it is also necessary to determine whether the contact force in the generated change curve meets the preset threshold at each stage of seed growth. The optimization ends when the threshold is met. After optimization, all specific parameters in the latest design pool are integrated as a new set of target physical property parameters to generate optimized seed design schemes and soil improvement strategies, and to provide visual feedback. The beneficial effects of the above technical solution are as follows: By designing a scheduling sequence based on parameter matching degree, the design scheme pool is updated sequentially, so that parameter adjustment has clear priority and logic, avoiding the randomness and blindness of parameter optimization, ensuring that each step of modification is directed towards the core optimization goal, which is conducive to improving the efficiency of cost optimization, and reducing the cost of optimized seed design scheme and soil improvement strategy with minimal error, making it more in line with the actual planting situation.

[0034] like Figure 3 As shown, in one embodiment, when multiple specific modified parameters are scheduled from the scheduling sequence each time, a low-cost parameter rollback mechanism is triggered, which specifically includes the following steps: Step 1: Store all the specific modified parameters for this scheduling in the temporary storage area. In the temporary storage area, perform a search for parameters of the same type for all the specific modified parameters to obtain the search results. Step 2: When the search results show that there are parameters of the same type, compare the parameter matching degree of all parameters of the same type, mark the parameters of the same type with low matching degree as rollback, and extract all the specific modified parameters that do not have rollback marks in the temporary storage area to generate the update file. Step 3: Update the data of the same type in the design scheme pool according to the update file, and then input the updated data into the mechanical interaction model for discrete element simulation to obtain the simulation results. At the same time, record the update number as 1. The simulation results are the curves of contact force changing with seed growth displacement. Step 4: If all stages of the simulation results meet the preset threshold, proceed to step 5. If none of the stages of the simulation results meet the preset threshold and the update count is 1, proceed to step 6. If none of the stages of the simulation results meet the preset threshold and the update count is not 1, proceed to step 7. Step 5: Extract the specific modified parameters that have rollback flags and the highest parameter matching degree from the temporary storage area, generate a new update file, set a rollback mechanism for the design scheme pool, and perform an update count increment operation, then jump to step 3. Step 6: Reset the temporary storage area and update count, perform the next scheduling according to the scheduling sequence, and jump to step 1; Step 7: Based on the rollback mechanism in Step 5, rollback update the design scheme pool, and use all the specific parameters in the updated design scheme pool as the new target physical property parameter set to generate the optimized seed design scheme and soil improvement strategy. The working principle and beneficial effects of the above technical solution are as follows: In order to reduce the number of simulations and further reduce the computational cost, when this application is applied to the actual optimization process, the method of scheduling multiple specific modified parameters simultaneously is usually adopted in the early stage of optimization. Based on this, in order to prevent parameter conflicts caused by simultaneous calls of multiple specific modified parameters and the generation of design schemes with non-low optimization costs, this application sets a low-cost parameter rollback mechanism to be triggered when multiple specific modified parameters are scheduled from the scheduling sequence each time, in order to solve the above problems. Specifically, firstly, when multiple specific parameter modification calls exist, all specific parameter modification calls for this scheduling are stored in a temporary storage area. Then, a search for parameters of the same type is performed on all specific parameter modification calls in the temporary storage area to obtain the search results. By searching for parameters of the same type in advance through the temporary storage area and handling conflicts (retaining parameters with high matching degree), the mutual interference or redundant adjustments that may be caused by simultaneous modification of parameters of the same type are avoided from the source. This reduces the risk of data chaos in the design scheme pool and ensures the stability of the parameter optimization process. When similar parameters are found in the search results, parameter matching is compared among all similar parameters. It's worth noting that this comparison only considers similar parameters. Parameters with lower matching scores are marked for rollback. All modified parameters not marked for rollback are extracted from the temporary storage area to generate update files. It's also worth noting that modified parameters with higher matching scores among similar parameters usually indicate higher modification costs. Common sense dictates that modified parameters with higher modification costs generally have better adjustment effects. Therefore, we choose to prioritize high adjustment effects over low adjustment effects, marking similar parameters with lower matching scores for rollback. If multiple similar parameters with lower matching scores exist, all similar parameters are marked for rollback. Furthermore, if multiple types of similar parameters exist, they are still marked for rollback. By searching for similar parameters in the temporary storage area and filtering based on matching scores (reflecting parameter sensitivity and modification value), parameters with high matching scores are retained, while those with low matching scores are marked for rollback. This ensures that parameters contributing more to optimization are updated first, reducing the ineffective consumption caused by low-value parameter modifications and improving the optimization efficiency of a single scheduling operation. The data of the same type in the design scheme pool are updated according to the update file, and then input into the mechanical interaction model for discrete element simulation to obtain the simulation results. At the same time, the update number is recorded as 1. The simulation results referred to here are the curves of contact force changing with seed growth displacement as described above. Furthermore, in order to ensure the integrity of the updated design scheme, the integrity of the generated updated design scheme should also be verified. If the simulation results meet the preset thresholds at each stage, it indicates that the optimization meets the conditions. Further verification is needed to confirm whether it meets the low-cost optimization conditions. Extract the specific modified parameters that have rollback flags and the highest parameter matching degree from the temporary storage area, and generate a new update file. It is worth noting that there are no restrictions on the specific modified parameter types here. Only the parameter matching degree of the data in the temporary storage area is compared in order to obtain the modified parameters with the best adjustment effect and the lowest adjustment cost. At the same time, in order to facilitate the subsequent acquisition of the previous design scheme after the update, a rollback mechanism is set for the current design scheme pool, and in order to distinguish it from the next scheduling, an update count increment function is set. After a new update file is generated, the design scheme pool is updated again, and the simulation is repeated until the non-uniformity of each stage of the simulation results meets the preset threshold. At this point, the design scheme pool from the previous simulation should be the optimal design scheme. The design scheme pool is updated using the aforementioned rollback mechanism to restore the previous design scheme, which serves as a new set of target physical property parameters for generating optimized seed design schemes and soil improvement strategies. The rollback mechanism aims to minimize improvement costs while meeting the soil breaking conditions. If the number of updates is 1 and the simulation results do not all meet the preset threshold (i.e., the judgment result described above is negative), then the temporary storage area and the number of updates are reset, and the next scheduling is performed according to the scheduling sequence. If the number of the next scheduling is still multiple, the above operation is repeated.

[0035] Finally, it should be noted that the above preferred embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail through the above preferred embodiments, those skilled in the art should understand that various changes can be made to it in form and detail without departing from the scope defined by the claims of the present invention.

Claims

1. A method for optimizing seed breeding processes based on discrete element theory, characterized in that, include: Using the discrete element method, a mechanical interaction model between soil particles and seeds is established based on the collected set of physical characteristic parameters. The set of physical characteristic parameters includes: soil physical characteristic parameters under different soil environments, seed physical characteristic parameters of different seed types, and mechanical behavior data of any seed type during the breaking process in different soil environments. The target physical property parameter set is obtained and input into the mechanical interaction model. Under dynamic simulation conditions, the output of the contact force variation curve with seed growth displacement is obtained through discrete element simulation. The target physical property parameter set includes: soil physical property parameters under the target soil environment and seed physical property parameters of the target seed category. Determine whether the contact force in the change curve meets the preset threshold at each stage of seed growth. If the determination result is no, continue to self-optimize the target physical characteristic parameter set until the determination result is yes, and then end the optimization.

2. The seed breeding process optimization method based on discrete element theory according to claim 1, characterized in that: Soil physical property parameters include: soil shear strength, internal friction angle and compressibility measured by direct shear test and triaxial compression test, and soil particle size distribution, porosity and permeability measured by particle size analyzer and permeability test device. Seed physical property parameters include: external geometric properties, internal tissue structure, and internal structural mechanical properties of seeds, measured using microscopic imaging equipment and mechanical testing equipment. Among them, the internal structural mechanical properties include the seed's elastic modulus, seed coat thickness, and rupture pressure. The process of acquiring mechanical behavior data during seed emergence includes: designing emergence experiments under different soil conditions, using a loading device to measure the emergence pressure and radicle extension force of different seed types, recording the seed emergence process using a high-speed camera, and analyzing the recorded results to obtain mechanical behavior data during seed emergence; among which, the soil environment includes compaction parameters and moisture parameters.

3. The seed cultivation process optimization method based on discrete element theory according to claim 1 or 2, characterized in that: A comprehensive soil mechanical behavior model is generated by constructing soil physical property parameters; A seed model is constructed by using the physical property parameters of the seeds; By integrating the mechanical behavior model, the seed model, and the mechanical behavior data of seeds during the seed emergence process corresponding to the soil environment and seed type, the discrete element method is used to establish the corresponding mechanical interaction model between soil particles and seeds.

4. The seed cultivation process optimization method based on discrete element theory according to claim 3, characterized in that: In the comprehensive soil mechanical behavior model, the particle size distribution of soil particles is determined by the particle size distribution curve of the actual soil. During modeling, particles of different sizes need to be generated according to the actual proportion of different soil types to simulate the real soil structure. The normal contact force between soil particles is described by the Hertz-Mundling contact model, and the tangential contact force is described by the Coulomb friction model.

5. The seed breeding process optimization method based on discrete element theory according to claim 1, characterized in that: When performing discrete element simulation, GPU acceleration technology and distributed computing technology are used to distribute the entire discrete element simulation computation task to multiple computing nodes on the GPU platform, and data communication and collaborative computing between computing nodes are realized through message passing interface.

6. The seed breeding process optimization method based on discrete element theory according to claim 1, characterized in that: The continuously self-optimizing target physical property parameter set includes design schemes for optimizing seed physical property parameters and improvement strategies for optimizing soil physical property parameters; Among them, the optimization logic of the design scheme and improvement strategy in the optimization process is related to the parameter sensitivity of seed physical property parameters and soil physical property parameters.

7. The seed breeding process optimization method based on discrete element theory according to claim 6, characterized in that: A prediction model based on support vector machine or random forest algorithm is constructed, taking seed physical property parameters and soil physical property parameters as inputs and seed germination performance index as output; among which, the seed germination performance index is related to germination time and germination force. By modifying the input parameters, different seed emergence performance indices are obtained, which are then used in conjunction with the Sobol' global sensitivity analysis method to evaluate the parameter sensitivity of any parameter among the seed physical property parameters and soil physical property parameters.

8. The seed breeding process optimization method based on discrete element theory according to claim 1, characterized in that: The process of constructing a mechanical interaction model includes constructing an initial mechanical interaction model and calibrating it to obtain a final mechanical interaction model. The final mechanical interaction model obtained through calibration includes: Discrete element simulation is performed using the initial mechanical interaction model. The simulation data is then compared with the actual experimental data. Based on the comparison results, the simulation model parameters of the initial mechanical interaction model are calibrated until the comparison results meet the preset calibration conditions, thus obtaining the final mechanical interaction model.

9. The seed breeding process optimization method based on discrete element theory according to claim 6, characterized in that, The continuously self-optimizing target physical property parameter set includes design schemes for optimizing seed physical property parameters and improvement strategies for optimizing soil physical property parameters, including: Obtain the parameter sensitivity of all optimizable parameters, and the parameter modification cost coefficient of each specific parameter; where the sum of the parameter modification cost coefficients of all specific parameters is 1. Obtain the parameter matching degree for each specific modified parameter; Wherein, parameter matching degree = parameter sensitivity * (1 - parameter modification cost coefficient); All specific modified parameters are arranged into a scheduling sequence from largest to smallest according to the parameter matching degree; Build a design scheme pool by obtaining specific parameters from the target physical property parameter set; The specific parameters are scheduled sequentially according to the scheduling sequence. Data of the same type in the design scheme pool are updated. It is determined whether the contact force in the change curve generated after each update meets the preset threshold at each stage of seed growth after discrete element simulation. The optimization ends when the threshold is met. All specific parameters in the latest design scheme pool are integrated as a new set of target physical property parameters to generate optimized seed design schemes and soil improvement strategies.

10. The seed breeding process optimization method based on discrete element theory according to claim 9, characterized in that, When multiple modified parameters are scheduled from the scheduling sequence each time, a low-cost parameter rollback mechanism is triggered, which includes the following steps: Step 1: Store all the specific modified parameters for this scheduling in the temporary storage area. In the temporary storage area, perform a search for parameters of the same type for all the specific modified parameters to obtain the search results. Step 2: When the search results show that there are parameters of the same type, compare the parameter matching degree of all parameters of the same type, mark the parameters of the same type with low matching degree as rollback, and extract all the specific modified parameters that do not have rollback marks in the temporary storage area to generate the update file. Step 3: Update the data of the same type in the design scheme pool according to the update file, and then input the updated data into the mechanical interaction model for discrete element simulation to obtain the simulation results. At the same time, record the update number as 1. The simulation results are the curves of contact force changing with seed growth displacement. Step 4: If all stages of the simulation results meet the preset threshold, proceed to step 5. If none of the stages of the simulation results meet the preset threshold and the update count is 1, proceed to step 6. If none of the stages of the simulation results meet the preset threshold and the update count is not 1, proceed to step 7. Step 5: Extract the specific modified parameters that have rollback flags and the highest parameter matching degree from the temporary storage area, generate a new update file, set a rollback mechanism for the design scheme pool, and perform an update count increment operation, then jump to step 3. Step 6: Reset the temporary storage area and update count, perform the next scheduling according to the scheduling sequence, and jump to step 1; Step 7: Based on the rollback mechanism in Step 5, rollback and update the design scheme pool. Use all the specific parameters in the updated design scheme pool as the new target physical property parameter set to generate the optimized seed design scheme and soil improvement strategy.

Images