A flue-cured tobacco monitoring management method and system for applying fertilizer according to real-time data

Abstract

This invention discloses a method and system for monitoring and controlling flue-cured tobacco fertilization based on real-time data. The method includes: acquiring multi-dimensional sensing data under flue-cured tobacco planting conditions, including environmental physicochemical parameters, soil biological activity parameters, and crop phenotypic physiological parameters; calibrating the state vector in a pre-constructed digital physiological model based on the multi-dimensional sensing data using a parameter calibration algorithm; performing collaborative planning based on the calibrated digital physiological model, flue-cured tobacco transpiration rate, and comprehensive stress index to generate a final control command; and controlling the ratio and total amount of various control liquids according to the final control command, and triggering pest and disease control operations based on the command. This invention can accurately sense and correct prediction deviations caused by environmental disturbances or individual differences, ensuring that the generated control command meets the actual physiological and metabolic needs of the crop at the present time.

Description

Technical Field

[0001] This invention relates to a method and system for monitoring and controlling the application of fertilizers to flue-cured tobacco based on real-time data, belonging to the field of tobacco processing technology. Background Technology

[0002] As a high-tech intensive industry in modern agriculture, flue-cured tobacco cultivation demands extremely high timeliness and precision in water and fertilizer management, environmental monitoring, and pest and disease control. With the popularization of the Internet of Things (IoT) and smart agriculture technologies, data-driven automated monitoring and control systems have gradually become an important means to improve tobacco yield and quality. These systems typically integrate environmental sensing, data transmission, and automated control functions, aiming to replace traditional manual field inspections and experience-based fertilization with digital means, thereby achieving standardized and intelligent management of the flue-cured tobacco cultivation process.

[0003] In existing technologies, the application of flue-cured tobacco monitoring and control systems mainly relies on various physical sensors and image acquisition devices deployed in the fields. The system collects basic physicochemical data of the environment and soil in real time through temperature sensors, humidity sensors, and soil electrical conductivity (EC) sensors, and combines this with meteorological information obtained from weather stations. Some systems also utilize industrial cameras to photograph plants to assist in manual observation. In terms of control logic, existing systems typically have pre-set general agronomic fertilization expert systems or fixed water and fertilizer formulas. When monitored environmental parameters (such as soil moisture) fall below a set threshold, the system will drive the actuators to irrigate or fertilize according to preset rules and logic, using fixed ratios and dosages to maintain the basic environmental conditions required for crop growth.

[0004] However, a major shortcoming of existing technologies is that their decision-making logic is mostly based on static preset models or fixed empirical thresholds, which cannot adapt to the highly nonlinear and dynamic time-varying physiological characteristics of flue-cured tobacco under complex and ever-changing natural environments. Due to the lack of a mechanism for online adaptive calibration of the internal control model based on real-time collected multidimensional phenotypic and physicochemical data, existing systems struggle to accurately detect and correct prediction biases caused by environmental disturbances or individual differences. This results in generated control commands often being misaligned with the crop's actual physiological and metabolic needs, making it difficult to achieve optimal growth adaptation and stress resistance while simultaneously meeting preset yield targets. Therefore, in order to solve the above-mentioned technical problems, there is an urgent need for a method and system for monitoring and controlling the application of fertilizers to flue-cured tobacco based on real-time data. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a method and system for monitoring and controlling the application of fertilizer to flue-cured tobacco based on real-time data. This system can accurately sense and correct prediction deviations caused by environmental disturbances or individual differences, so that the generated control instructions can meet the actual physiological and metabolic needs of the crop at the moment.

[0006] To achieve the above objectives, the present invention is implemented using the following technical solution: In a first aspect, the present invention provides a method for monitoring and controlling the application of fertilizer to flue-cured tobacco based on real-time data, comprising: Acquire multidimensional sensing data in flue-cured tobacco planting scenarios, including environmental physicochemical parameters, soil biological activity parameters, and crop phenotypic and physiological parameters. Based on the multidimensional sensing data, the state vectors in the pre-constructed digital physiological model are calibrated using a parameter calibration algorithm; Based on the calibrated digital physiological model and multidimensional sensing data, the transpiration rate of flue-cured tobacco, which reflects the instantaneous water consumption of crops, and the comprehensive stress index, which characterizes the degree of environmental stress, were calculated. Based on the calibrated digital physiological model, flue-cured tobacco transpiration rate and comprehensive stress index, collaborative planning is carried out to generate the final regulatory instructions. According to the final control instructions, the ratio and total amount of various control solutions are controlled, and pest and disease control operations are triggered according to the instructions.

[0007] Furthermore, the collaborative planning includes long-term planning and short-term planning, wherein, The long-term planning aims to maximize the closeness between the flue-cured tobacco growth status and the yield target curve within a future preset period. It uses reinforcement learning to make sequential decisions in the intelligent decision space to determine the macro-fertilization plan, and uses the flue-cured tobacco transpiration rate as a feedback signal to dynamically adjust the nutrient formulation in the macro-fertilization plan. The short-cycle planning uses crop phenotypic physiological parameters as real-time feedback signals to generate micro-control instructions for dynamically adjusting the conductivity and pH of the water and fertilizer solution during local application. The environmental state vector of the intelligent decision space includes environmental physicochemical parameters, tobacco transpiration rate, crop phenotypic and physiological parameters, preset yield target, and future growth trend predicted by the calibrated digital physiological model.

[0008] Furthermore, when the comprehensive stress index is greater than the preset warning threshold, the optimal input determined by the reinforcement learning strategy is blocked, an instruction to apply the preset stress buffer formula is generated, and the preset reward function of the reinforcement learning strategy is switched from the maximum output mode to the minimum physiological stress mode. The preset reward function in the minimized physiological stress mode is the reward value in the maximized output mode minus the product of the preset penalty coefficient and the comprehensive stress index.

[0009] Furthermore, the digital physiological model is defined as a discrete-time nonlinear state-space system, and the state vector of the digital physiological model includes the soil conductivity coefficient and photosynthetic efficiency. Based on the multidimensional sensing data, the state vectors in the pre-constructed digital physiological model are calibrated using a parameter calibration algorithm, including: The prior state estimate at the current time is predicted using the posterior estimate from the previous time step, and the prior error covariance matrix at the current time step is predicted using the posterior error covariance matrix from the previous time step. The Kalman gain matrix is ​​calculated by combining the prior error covariance matrix, the preset measurement noise covariance matrix, and the observation matrix; wherein, the observation matrix is ​​used to establish the mapping relationship from the state space to the observation space, the soil biological activity parameter is used as the observation mapping of the soil conductivity coefficient, and the crop phenotypic physiological parameter is used as the observation mapping of photosynthetic efficiency. The calibrated state vector is obtained by calibrating the Kalman gain matrix, the actual observed data vector, and the prior state estimate.

[0010] Furthermore, the environmental physicochemical parameters include various environmental indicators during crop growth, including air temperature, wind speed, and light duration; the crop phenotypic and physiological parameters include the degree of linear polarization. and leaf temperature suppression value The soil biological activity parameters include microbial respiration intensity. ,in, The formula for calculating the degree of linear polarization is: ; in, , , , These represent the light radiation intensity values ​​received when the polarizer is rotated to 0°, 90°, 45°, and 135°, respectively. The formula for calculating the blade temperature suppression value is as follows: ; in, The average temperature of the blade surface; Air temperature; This is the wet-bulb temperature, representing the lowest theoretical temperature the leaf can reach under the current humidity conditions. The value is dimensionless. The larger the value, the weaker the leaf's ability to dissipate heat through transpiration, and the stronger the physiological inhibition it experiences. The formula for calculating the microbial respiration intensity is as follows: ; in, The response current value is measured in real time by the sensor; Baseline current noise in sterile soil matrix; The number of electrons transferred in the redox reaction; It is Faraday's constant; This represents the effective surface area of ​​the working electrode of the sensor. This is a correction factor for the effective diffusion coefficient related to soil porosity and moisture content.

[0011] Furthermore, the formula for calculating the transpiration rate of the flue-cured tobacco is as follows: ; in, This refers to the instantaneous evaporation rate of flue-cured tobacco. The slope of the saturated water vapor pressure-temperature curve; Net radiation of the canopy; air density; The specific heat of air at constant pressure; The saturated water vapor pressure difference is calculated from the ambient temperature and humidity. For aerodynamic conductivity; The latent heat of vaporization of water; This is the constant of the wet / dry meter; It concerns the leaf temperature inhibition value. The stomatal conductance function is defined as follows: , The maximum porosity is associated with the state vector. The attenuation coefficient; The formula for calculating the comprehensive stress index is as follows: ; in, The preset weighting coefficients satisfy... This represents the normalized degree of linear polarization. The leaf lesion area is extracted using RGB data. The total area of ​​the blades. Represents disease and stress; The current temperature, The optimal temperature for flue-cured tobacco growth is used to characterize thermal environmental stress. The nutrient distribution in the macro-fertilization plan, dynamically adjusted based on the transpiration rate of flue-cured tobacco as feedback signal, is as follows: ; ; in, This is the revised single-control feed-liquid ratio. This refers to the original ratio in the macro-fertilization plan. It is the correction factor for the fertilizer ratio; It is the lower limit of the correction factor; It is a correction of the sensitivity coefficient. This refers to the instantaneous evaporation rate of flue-cured tobacco in real time. The modified formula is based on the preset healthy threshold for instantaneous evaporation rate. This ensures that fertilization is highly synchronized with the plant's real-time physiological and metabolic state.

[0012] Furthermore, the calculation formula for dynamically adjusting the conductivity and pH of the water-fertilizer solution during local application is as follows: ; in, represent Temporal conductivity or pH The incremental adjustment; for Leaf temperature suppression value at any time Relative to the median of the target range deviation, that is , These are the proportional, integral, and differential coefficients, respectively. This represents the deviation from the previous moment; , This represents the upper limit of the target range. This is the lower limit of the target range.

[0013] Secondly, the present invention provides a monitoring and control system for flue-cured tobacco that applies fertilizer based on real-time data, comprising: The multidimensional data acquisition module is used to acquire multidimensional sensing data in the tobacco planting scenario. The multidimensional sensing data includes environmental physicochemical parameters, soil biological activity parameters, and crop phenotypic and physiological parameters. The model adaptive calibration module is used to calibrate the state vector in the pre-constructed digital physiological model based on the multidimensional sensing data and a parameter calibration algorithm. Based on the calibrated digital physiological model and multidimensional sensing data, the transpiration rate of flue-cured tobacco, which reflects the instantaneous water consumption of crops, and the comprehensive stress index, which characterizes the degree of environmental stress, were calculated. The dual-layer intelligent control module is used to perform collaborative planning based on the calibrated digital physiological model, the transpiration rate of flue-cured tobacco, and the comprehensive stress index, and generate the final control command. The precision execution module is used to control the ratio and total amount of various control solutions according to the final control instructions, and to trigger pest and disease control operations based on the instructions.

[0014] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of any of the methods described above.

[0015] Fourthly, the present invention provides a computer device, comprising: Memory, used to store computer programs / instructions; A processor for executing the computer program / instructions to implement the steps of any of the methods described above.

[0016] Fifthly, the present invention provides a computer program product, including a computer program / instructions that, when executed by a processor, implement the steps of any of the methods described above.

[0017] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: This invention provides a method and system for monitoring and controlling flue-cured tobacco fertilization based on real-time data. By integrating multi-dimensional environmental sensing and industrial-grade optical monitoring technologies, a closed-loop control system for flue-cured tobacco growth based on real-time data is established. The system utilizes a sensor array to collect real-time data on ambient temperature, soil moisture, and soil conductivity, and coordinates with plant shape and leaf color information captured by an industrial-grade color camera. This achieves a comprehensive digital mapping of the crop's growth environment and physiological phenotype. Based on a pre-set yield target value, the system uses a multi-dimensional data fusion analysis algorithm to accurately quantify the actual nutrient deficit of the crop under the current environment, providing reliable data support for subsequent precision fertilization and ensuring accurate transmission and scientific quantification from information perception to agronomic decision-making.

[0018] This invention provides a method and system for monitoring and controlling the application of fertilizers to flue-cured tobacco based on real-time data. By constructing an intelligent fertilizer formulation and control algorithm oriented towards target yield, the system achieves dynamic optimization and on-demand allocation of nutrient supply. Based on real-time meteorological data and deviation analysis results of crop growth status, the system intelligently calculates and generates the optimal mixing ratio and total application amount instructions for various control liquids. This algorithm linkage mechanism overcomes the lag and blindness of traditional fixed-formula fertilization. It can flexibly adjust the application strategies of various nutrient solutions, such as liquid solution one and liquid solution two, according to the actual physiological needs of flue-cured tobacco at different growth stages. While improving fertilizer utilization efficiency, it effectively ensures a high degree of consistency between the final yield and the preset target.

[0019] This invention provides a method and system for monitoring and controlling flue-cured tobacco fertilization based on real-time data. By incorporating intelligent pest and disease identification and emergency response control logic, the system enhances the risk prevention and control capabilities and automated error correction of the planting system. While conducting routine growth monitoring, the system continuously scans for abnormal leaf color and plant shape distortion using image analysis algorithms. Once pest and disease characteristics or growth stress signals are identified, a graded alarm is immediately triggered and a targeted prevention and control plan is automatically generated. The system can coordinate the spraying of insecticides and the supplementary application of specific functional liquid fertilizers, realizing fully automated handling from abnormal monitoring and automatic early warning to targeted application of pesticides, effectively reducing the potential losses to flue-cured tobacco quality and yield caused by biological stress. Attached Figure Description

[0020] Figure 1 This is a flowchart of a method for monitoring and controlling the application of fertilizer to flue-cured tobacco based on real-time data, provided by an embodiment of the present invention; Figure 2 This is a schematic diagram of a flue-cured tobacco monitoring and control system for applying fertilizer based on real-time data, provided in an embodiment of the present invention. Detailed Implementation

[0021] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the embodiments and specific features in the embodiments are detailed descriptions of the technical solution of the present application, rather than limitations thereof. In the absence of conflict, the embodiments and technical features in the embodiments can be combined with each other.

[0022] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article generally indicates that the preceding and following related objects have an "or" relationship. Example 1

[0023] Figure 1 This is a flowchart of a method for monitoring and controlling flue-cured tobacco application based on real-time data, according to Embodiment 1 of the present invention. The method for monitoring and controlling flue-cured tobacco application based on real-time data provided in this embodiment can be applied to a terminal and can be executed by a flue-cured tobacco monitoring and control system that applies fertilizer based on real-time data. This system can be implemented by software and / or hardware and can be integrated into the terminal, such as any smartphone, tablet, or computer device with communication capabilities. See also... Figure 1 The method implemented in this way specifically includes the following steps: Acquire multidimensional sensing data in flue-cured tobacco planting scenarios, including environmental physicochemical parameters, soil biological activity parameters, and crop phenotypic and physiological parameters. Based on the multidimensional sensing data, the state vectors in the pre-constructed digital physiological model are calibrated using a parameter calibration algorithm; Based on the calibrated digital physiological model and multidimensional sensing data, the transpiration rate of flue-cured tobacco, which reflects the instantaneous water consumption of crops, and the comprehensive stress index, which characterizes the degree of environmental stress, were calculated. Based on the calibrated digital physiological model, flue-cured tobacco transpiration rate and comprehensive stress index, collaborative planning is carried out to generate the final regulatory instructions. According to the final control instructions, the ratio and total amount of various control solutions are controlled, and pest and disease control operations are triggered according to the instructions.

[0024] The method for monitoring and controlling the application of fertilizers to flue-cured tobacco based on real-time data provided in this embodiment involves the following steps in its application process: The collaborative planning includes long-term planning and short-term planning, wherein, The long-term planning aims to maximize the closeness between the flue-cured tobacco growth status and the yield target curve within a future preset period. It uses reinforcement learning to make sequential decisions in the intelligent decision space to determine the macro-fertilization plan, and uses the flue-cured tobacco transpiration rate as a feedback signal to dynamically adjust the nutrient formulation in the macro-fertilization plan. The short-cycle planning uses crop phenotypic physiological parameters as real-time feedback signals to generate micro-control instructions for dynamically adjusting the conductivity and pH of the water and fertilizer solution during local application. The environmental state vector of the intelligent decision space includes environmental physicochemical parameters, tobacco transpiration rate, crop phenotypic and physiological parameters, preset yield target, and future growth trend predicted by the calibrated digital physiological model.

[0025] Furthermore, when the comprehensive stress index is greater than the preset warning threshold, the optimal input determined by the reinforcement learning strategy is blocked, an instruction to apply the preset stress buffer formula is generated, and the preset reward function of the reinforcement learning strategy is switched from the maximum output mode to the minimum physiological stress mode. The preset reward function in the minimized physiological stress mode is the reward value in the maximized output mode minus the product of the preset penalty coefficient and the comprehensive stress index.

[0026] Furthermore, the digital physiological model is defined as a discrete-time nonlinear state-space system, and the state vector of the digital physiological model includes the soil conductivity coefficient and photosynthetic efficiency. Based on the multidimensional sensing data, the state vectors in the pre-constructed digital physiological model are calibrated using a parameter calibration algorithm, including: The prior state estimate at the current time is predicted using the posterior estimate from the previous time step, and the prior error covariance matrix at the current time step is predicted using the posterior error covariance matrix from the previous time step. The Kalman gain matrix is ​​calculated by combining the prior error covariance matrix, the preset measurement noise covariance matrix, and the observation matrix; wherein, the observation matrix is ​​used to establish the mapping relationship from the state space to the observation space, the soil biological activity parameter is used as the observation mapping of the soil conductivity coefficient, and the crop phenotypic physiological parameter is used as the observation mapping of photosynthetic efficiency. The calibrated state vector is obtained by calibrating the Kalman gain matrix, the actual observed data vector, and the prior state estimate.

[0027] The environmental physicochemical parameters include various environmental indicators during crop growth, including air temperature, wind speed, and light duration. The crop phenotypic and physiological parameters include the degree of linear polarization. and leaf temperature suppression value The soil biological activity parameters include microbial respiration intensity. ,in, The formula for calculating the degree of linear polarization is: ; in, , , , These represent the light radiation intensity values ​​received when the polarizer is rotated to 0°, 90°, 45°, and 135°, respectively. The formula for calculating the blade temperature suppression value is as follows: ; in, The average temperature of the blade surface; Air temperature; This is the wet-bulb temperature, representing the lowest theoretical temperature the leaf can reach under the current humidity conditions. The value is dimensionless. The larger the value, the weaker the leaf's ability to dissipate heat through transpiration, and the stronger the physiological inhibition it experiences. The formula for calculating the microbial respiration intensity is as follows: ; in, The response current value is measured in real time; Baseline current noise in sterile soil matrix; The number of electrons transferred in the redox reaction; It is Faraday's constant; This represents the effective surface area of ​​the working electrode of the sensor. This is a correction factor for the effective diffusion coefficient related to soil porosity and moisture content.

[0028] The formula for calculating the transpiration rate of the flue-cured tobacco is as follows: ; in, This refers to the instantaneous evaporation rate of flue-cured tobacco. The slope of the saturated water vapor pressure-temperature curve; Net radiation of the canopy; air density; The specific heat of air at constant pressure; The saturated water vapor pressure difference is calculated from the ambient temperature and humidity. For aerodynamic conductivity; The latent heat of vaporization of water; This is the constant of the wet / dry meter; It concerns the leaf temperature inhibition value. The stomatal conductance function is defined as follows: , The maximum porosity is associated with the state vector. The attenuation coefficient; The formula for calculating the comprehensive stress index is as follows: ; in, The preset weighting coefficients satisfy... This represents the normalized degree of linear polarization. The leaf lesion area is extracted using RGB data. The total area of ​​the blades. Represents disease and stress; The current temperature, The optimal temperature for flue-cured tobacco growth is used to characterize thermal environmental stress. The nutrient distribution in the macro-fertilization plan, dynamically adjusted based on the transpiration rate of flue-cured tobacco as feedback signal, is as follows: ; ; in, This is the revised single-control feed-liquid ratio. This refers to the original ratio in the macro-fertilization plan. It is the correction factor for the fertilizer ratio; It is the lower limit of the correction factor; It is a correction of the sensitivity coefficient. This refers to the instantaneous evaporation rate of flue-cured tobacco in real time. The modified formula is based on the preset healthy threshold for instantaneous evaporation rate. This ensures that fertilization is highly synchronized with the plant's real-time physiological and metabolic state.

[0029] Furthermore, the calculation formula for dynamically adjusting the conductivity and pH of the water-fertilizer solution during local application is as follows: ; in, represent Temporal conductivity or pH The incremental adjustment; for Leaf temperature suppression value at any time Relative to the median of the target range deviation, that is , These are the proportional, integral, and differential coefficients, respectively. This represents the deviation from the previous moment; , This represents the upper limit of the target range. This is the lower limit of the target range.

[0030] The following describes in detail the contents involved in the above embodiments with reference to a preferred embodiment.

[0031] To acquire multidimensional sensing data in the context of flue-cured tobacco cultivation, the multidimensional data acquisition module serves as the system's sensing front-end. Its core task is to convert analog signals from the physical world into digital multidimensional sensing data that can be processed by a computer. Specifically, this includes: 1. Optical polarization acquisition and processing of crop phenotypic and physiological parameters Using a configured industrial-grade polarization camera, the radiation intensity images of flue-cured tobacco leaves at different polarization angles are acquired within preset specific wavelength bands, such as the visible light band at 550nm or the near-infrared band at 850nm. Specifically, the camera acquires light intensity images at four polarization directions: 0°, 45°, 90°, and 135°, denoted as [images of light intensity at different polarization angles]. , , and Based on these four images, the module first calculates the Stokes vector. and The components are calculated using the following formula: , ; In the formula, , , , These represent the light radiation intensity values ​​received by the sensor when the polarizer is rotated to 0°, 90°, 45°, and 135°, respectively.

[0032] Furthermore, based on and The optical scattering characteristics of the blade surface microstructure are calculated from the component analysis. This information is specifically characterized by the degree of linear polarization (DoLP), and the calculation formula is as follows: ; The calculated DoLP value directly reflects the integrity of the leaf epidermal wax layer and the microstructure of the cell wall. When the cell wall is damaged or the wax layer is shed, the DoLP value will change significantly, serving as an input for subsequent stress index calculations.

[0033] 2. Thermal Infrared Acquisition and Calculation of Leaf Temperature Inhibition Values The system utilizes a microscopic thermal infrared sensor array on the blades to acquire real-time images of the thermal distribution of the tobacco canopy and extract the absolute temperature of the blade surface. To eliminate the interference of ambient background temperature and accurately characterize the warming effect caused by stomatal closure in crops, the module needs to calculate the leaf temperature suppression value. This calculation relies on synchronously collected ambient air temperature. and the preset lower limit reference temperature for crops. This refers to the theoretical leaf temperature at which crop stomata are fully open and transpiration is at its maximum under current environmental conditions. The calculation formula is: ; in, The average surface temperature of the blade is measured by a thermal infrared sensor array. The air temperature measured by the ambient temperature sensor; This is the wet-bulb temperature, representing the lowest theoretical temperature the leaf can reach under the current humidity conditions. The value is dimensionless. The larger the value, the weaker the leaf's ability to dissipate heat through transpiration, and the stronger the physiological inhibition it experiences. 3. Electrochemical acquisition and transformation of soil biological activity parameters The system monitors the redox reactions of specific microbial metabolites in the soil using current analysis by inserting an electrochemical sensor into the root zone of the soil. The raw signal output by the sensor is a microampere-level current. In order to obtain the microbial respiration intensity among soil biological activity parameters The module is transformed using a modified form of the Cottrell equation: ; in, The response current value measured by the sensor in real time; Baseline current noise in sterile soil matrix; The number of electrons transferred in the redox reaction; is the Faraday constant; S is the effective surface area of ​​the working electrode of the sensor; This is a correction factor for the effective diffusion coefficient related to soil porosity and water content. The calculated... It characterizes the metabolic activity of the soil microbial community per unit time.

[0034] Model adaptive calibration module: As the computational core of the system, it is responsible for maintaining the digital physiological model and using the Kalman filter to correct the model parameters in real time to ensure that the model predictions are consistent with the actual growth state. 1. Construction and State Space Definition of Digital Physiological Models The digital physiological model is defined as a discrete-time nonlinear state-space system, whose input is the actual observation data transmitted from the multidimensional data acquisition module, i.e., the observation vector. The output is the plant's most accurate physical state at this moment, determined by the machine after processing and calibration; that is, the state vector. The system's observation equations are defined as follows: ; in, yes The vector of observed data at each moment; It is the observation matrix, used to establish the mapping relationship from the state space to the observation space; It is the measurement noise vector, representing the random error generated by the sensor during the measurement process.

[0035] The system's state vector It includes two core parameters to be calibrated: soil conductivity. And crop photosynthetic efficiency The state equations are defined as follows:

[0036] in, yes The state vector at any given time; It is the state vector from the previous moment; It is a process noise vector with a mean of 0 and a covariance of . The equation assumes a Gaussian distribution. It also assumes that, under conditions of no strong external disturbance, the soil conductivity and photosynthetic efficiency remain relatively stable over a short period.

[0037] 2. Parameter Adaptive Calibration Algorithm The module receives the observation vector input from the multidimensional data acquisition module. This vector contains the measured soil biological activity parameters at the current moment as... Observational mapping and crop phenotypic physiological parameters as The observation mapping and calibration process consists of two steps: prediction and update. Step A: Time update, using the posterior estimate from the previous time step. Predict the prior state at the current moment :

[0038] in, Let A be the prior estimate of the state vector at time k; A is the state transition matrix, which is the identity matrix I in this model. The prior error covariance matrix; Let be the posterior error covariance matrix of the previous time step; Q is the process noise covariance matrix.

[0039] Step B: Measurement update, calculate Kalman gain And update the state vector:

[0040] in, H is the Kalman gain matrix, which determines the weight of the observation data in correcting the model parameters; H is the observation matrix, used to map the state space to the observation space; R is the measurement noise covariance matrix. This is the vector of actual observed data at the current moment; This is the final calibrated state vector at time k, i.e., the updated soil conductivity and photosynthetic efficiency; The superscript in the formula represents the updated error covariance matrix. This indicates that the matrix is ​​transposed. Through this step, the digital physiological model completes adaptive parameter adjustments in response to changes in the current environment.

[0041] 3. Calculation of the instantaneous transpiration rate of flue-cured tobacco, a key physiological indicator Based on the calibrated model parameters and real-time sensing data, the module calculates key physiological indicators, including the instantaneous transpiration rate of flue-cured tobacco. This calculation incorporates the leaf temperature suppression value. A simplified variant of the Penman-Montes formula:

[0042] in, This refers to the instantaneous evaporation rate; The slope of the saturated water vapor pressure-temperature curve; Net radiation of the canopy; air density; The specific heat of air at constant pressure; The saturated water vapor pressure difference is calculated from the ambient temperature and humidity. For aerodynamic conductivity; The latent heat of vaporization of water; This is the constant of the wet / dry meter; It concerns the leaf temperature inhibition value. The stomatal conductance function is usually defined as That is, the porosity decreases exponentially with the increase of the temperature inhibition value. For maximum porosity, The attenuation coefficient; In the model, the maximum porosity It is not a fixed constant, but rather determined by the aforementioned calibrated model state vector (i.e., soil conductivity coefficient). and photosynthetic efficiency A dynamically determined function. Specifically, The baseline value will be based on Characterized water supply capacity and The system adaptively adjusts its carbon assimilation requirements in real time; when root-soil conductivity is good and leaf photosynthetic efficiency is high, the system will increase [the required level]. The input value is set to a certain value, and vice versa. Thus, the calibrated internal physiological parameters were successfully introduced and dominated the calculation of the instantaneous transpiration rate of flue-cured tobacco.

[0043] The specific formula for calculating dynamic correlation is as follows:

[0044] This represents the maximum porosity after real-time calibration. This is the reference constant for the maximum porosity conductance of this variety of flue-cured tobacco under ideal conditions; The soil conductivity coefficient after calibration using a digital model; The photosynthetic efficiency is calibrated using a digital model. and These are the preset weighting coefficients for water supply and carbon assimilation demand (both are positive numbers).

[0045] 4. Construction of the Comprehensive Stress Index The module uses a multidimensional weighted algorithm to calculate a comprehensive stress index that characterizes crop health risk. This index integrates physical structural damage and environmental meteorological risks, and its calculation formula is as follows:

[0046] in, The preset weighting coefficients satisfy... This represents the normalized degree of linear polarization; a lower value indicates more severe damage to the cell structure. Characterizing structural stress; The leaf lesion area is extracted using RGB data. The ratio of the two values ​​represents the total leaf area and disease stress. The current temperature, This is the optimal temperature for flue-cured tobacco growth; this factor characterizes thermal environmental stress, and the final output is... The value is between 0 and 1 and is directly input into the dual-layer intelligent control module to trigger the emergency interlock logic.

[0047] Dual-layer intelligent control module: The dual-layer intelligent control module will output the calibrated digital physiological model from the model adaptive calibration module. Intelligent decision-making state space Key physiological indicators and comprehensive stress index As input, the final control command is generated through long-term planning and short-term adjustment. ; 1. Long-term planning function The long-term planning function employs a Deep Q-Network (DQN) or Soft Actor-Critic (SAC) strategy in reinforcement learning, within the intelligent decision-making state space. Sequential decision-making is performed to determine the macro-fertilization plan. The plan includes adjusting the ratio of feed liquid. and total application rate Intelligent decision-makers and calibrated digital physiological models Conduct interactive simulations and select the preset reward function. Cumulative fertilization actions reward function The setting goal is to maximize the future preset period. Predicted growth status of flue-cured tobacco using internal digital physiological models Compared with the preset production target curve The degree of closeness is measured using a quadratic penalty term to determine the deviation. Reward function. The calculation formula is:

[0048] in, It provides basic positive rewards to ensure the stability of RL training; These are weighting coefficients used to penalize bias. ; It is predicted by the calibrated digital physiological model. Indicators of flue-cured tobacco growth status, such as leaf area index or biomass, are monitored at all times. yes The ideal target growth state value preset at all times; It is the planned future preset cycle length, representing the time span of decision-making, maximizing This means optimizing the long-term growth path.

[0049] 2. Dynamic formulation correction logic based on key physiological indicators The dynamic formula correction logic uses real-time monitoring of the instantaneous evaporation rate of flue-cured tobacco. As the core feedback signal, the macro-fertilization plan generated by the long-term planning function... The nutrient formula is adjusted in real time to generate a revised formula. ,when Below the preset health threshold When this occurs, it indicates that the plant is under water stress or its physiological metabolism is impaired, and the module will initiate a correction: reducing growth-promoting elements such as phosphorus. potassium Weighting in the formula and At the same time, the proportion of stress buffering substances such as calcium, magnesium, and amino acids is increased. The revised single-control feed-liquid ratio Through the original proportions Multiply by Correction coefficient for positive correlation Obtain the correction coefficient. The calculation formula is:

[0050] in, It is the correction factor for the fertilizer ratio; It is the lower limit of the correction coefficient, used to prevent the formula weight from dropping to zero; It is a correction of the sensitivity coefficient. This refers to the instantaneous evaporation rate of flue-cured tobacco in real time. The modified formula is based on the preset healthy threshold for instantaneous evaporation rate. This ensures that fertilization is highly synchronized with the real-time physiological and metabolic state of the plants; 3. Emergency Interlock Logic Based on Comprehensive Stress Index The emergency interlock logic is responsible for the system's security protection, based on the tobacco stress index. As a trigger signal, it forces intervention in the system's decision-making. Real-time logical comparison. With the preset warning threshold ,when Upon this, the system immediately performs an interlock operation, which includes two parts: first, it disables the fertilization action currently output by the reinforcement learning policy. Directly generate instructions to apply a preset stress buffer formulation. Secondly, the preset reward function of the reinforcement learning strategy will be changed from the output-maximizing mode. Switch to Minimize Physiological Stress Mode The calculation formula is as follows:

[0051] in, It is the value of the reward function that minimizes the physiological stress mode; It is the reward function value under the original maximum output model; It is a preset penalty coefficient. Furthermore, this coefficient is a real number, used to quantify the negative impact of the current stress index on the total system reward, in order to guide the RL strategy to prioritize reducing it in the next planning cycle. value.

[0052] 4. Short-cycle regulation function and generation of final regulation command The short-cycle regulation function uses the leaf temperature suppression value collected by the microscopic thermal infrared sensor array on the leaf to control temperature. As a high-frequency feedback signal, the locally applied water and fertilizer solution is fine-tuned at the millisecond level, generating micro-control instructions. The fine-tuning target is the conductivity of the water-fertilizer solution. and pH Fine-tuning employs a proportional-integral-derivative (PID) control algorithm to maintain... Within the preset optimal physiological range The target is the preset optimal physiological range. The upper and lower threshold values ​​are determined based on historical planting experiment data of this flue-cured tobacco variety at specific growth stages, as well as suitable growth indicators preset by the agronomic expert system, to ensure that the plant is in the most efficient photosynthetic and transpiration state. The real-time control output of the PID controller... and The calculation formula is:

[0053] in, represent time or The incremental adjustment; for time The deviation relative to the median of the target range, i.e. , These are the proportional, integral, and differential coefficients, respectively. This addresses the deviation from the previous moment. The dual-layer intelligent control module ultimately integrates the macro-fertilization plan, formula correction, interlock commands, and micro-control commands into a single final control command. And send it to the precision execution module, the aforementioned proportional coefficient Integral coefficient and differential coefficients The value is automatically tuned by the system during the initial operation phase using the Ziegler-Nichols step response method. Specifically, the system inputs a step control signal to the precision execution module and records feedback indicators (such as...) in real time. The response curve of the system is obtained, and the pure time delay of the system response is extracted from it. and time constant .

[0054] Based on the extracted feature parameters, the formulas for determining the three control coefficients are as follows:

[0055] The system calculates the optimal control coefficient matching the current water and fertilizer fluid characteristics and the response delay of flue-cured tobacco using the above formula, and solidifies it into the short-cycle control function. This avoids the blindness of manual experience-based parameter adjustment and ensures the speed and stability of fine-tuning control. During the initial system tuning phase, the dual-layer intelligent control module disconnects the closed-loop control and sends a constant step control signal to the precision execution module, which suddenly changes and maintains a certain opening degree of the water and fertilizer valve. Simultaneously, the system records the response indicators fed back by the sensors at high frequency (such as leaf temperature suppression values). The dynamic data that changes over time is used to plot the system's step response curve. This curve typically presents as an S-shaped trajectory with a delay.

[0056] The system uses an automatic optimization algorithm to locate the inflection point with the steepest slope on the S-curve and draws a tangent line through this inflection point. This tangent line intersects the horizontal baseline of the system's initial state, and the time difference between the intersection point and the starting point of the applied step signal is extracted as the system's pure time delay. Subsequently, the tangent line intersects the horizontal asymptote of the system's final steady state. The time span between these two intersection points is extracted as the system's time constant. .

[0057] IV. Specific Implementation and Operation of the Precision Execution Module The precision execution module is responsible for receiving the final control commands generated by the dual-layer intelligent control module. And translate it into actual physical operations; 1. Precise execution module for controlling the ratio and application rate of the liquid feed. The weighting of various regulating solutions included, such as nitrogen fertilizer, phosphorus fertilizer, potassium fertilizer, and stress buffer. and total application rate The module drives a high-precision metering pump and solenoid valve, monitors the output in real time via a flow sensor, and performs mass flow control-based mixing to ensure the quality of the mixed water-fertilizer solution. Actual proportions and instructions In Deviation does not exceed preset tolerance For arbitrary adjustment of feed liquid The execution control formula is:

[0058] in, yes Time of material liquid The instantaneous application volume; yes The total instantaneous volume of all liquid materials applied at any given time; It is the proportion and weight required by the instruction; It is a preset mixing tolerance, used to ensure mixing accuracy.

[0059] 2. Triggering of pest and disease control operations When the final control order Includes due to When the stress index exceeds the limit, triggering an emergency interlock logic instruction, the precision execution module uses the specific pest and disease control code attached to the instruction. Control the actuator, for example, if Indicating a risk of fungal infection, the module will drive a drone or spraying system to apply biocontrol agents or low-toxicity pesticides according to preset dosages and areas. Operation trigger status. For a Boolean function:

[0060] in, This indicates the operation is triggered; 1 indicates triggered, and 0 indicates not triggered. It is the final regulatory instruction; This is an emergency interlock command; It is the pest and disease control code contained in the instruction.

[0061] Example 2: This example provides a monitoring and control system for flue-cured tobacco that applies fertilizer based on real-time data, including: The multidimensional data acquisition module is used to acquire multidimensional sensing data in the tobacco planting scenario. The multidimensional sensing data includes environmental physicochemical parameters, soil biological activity parameters, and crop phenotypic and physiological parameters. The model adaptive calibration module is used to calibrate the state vector in the pre-constructed digital physiological model based on the multidimensional sensing data and a parameter calibration algorithm. Based on the calibrated digital physiological model and multidimensional sensing data, the transpiration rate of flue-cured tobacco, which reflects the instantaneous water consumption of crops, and the comprehensive stress index, which characterizes the degree of environmental stress, were calculated. The dual-layer intelligent control module is used to perform collaborative planning based on the calibrated digital physiological model, the transpiration rate of flue-cured tobacco, and the comprehensive stress index, and generate the final control command. The precision execution module is used to control the ratio and total amount of various control solutions according to the final control instructions, and to trigger pest and disease control operations based on the instructions.

[0062] The specific functions of each module described above are explained in the relevant content of the method in Embodiment 1, and will not be repeated here.

[0063] Example 3: This example provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in any of Examples 1.

[0064] Example 4: This example provides a computer device, including: Memory, used to store computer programs / instructions; A processor for executing the computer program / instructions to implement the steps of the method described in any of Embodiment 1.

[0065] Example 5: This example provides a computer program product, including a computer program / instructions, which, when executed by a processor, implement the steps of the method described in any of Examples 1.

[0066] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

[0067] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0068] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0069] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0070] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0071] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure and not to limit its protection scope. Although this disclosure has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading this disclosure, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the protection scope of the pending claims.

Claims

1. A method for monitoring and controlling the application of fertilizer to flue-cured tobacco based on real-time data, characterized in that, include: Acquire multidimensional sensing data in flue-cured tobacco planting scenarios, including environmental physicochemical parameters, soil biological activity parameters, and crop phenotypic and physiological parameters. Based on the multidimensional sensing data, the state vectors in the pre-constructed digital physiological model are calibrated using a parameter calibration algorithm; Based on the calibrated digital physiological model and multidimensional sensing data, the transpiration rate of flue-cured tobacco, which reflects the instantaneous water consumption of crops, and the comprehensive stress index, which characterizes the degree of environmental stress, were calculated. Based on the calibrated digital physiological model, flue-cured tobacco transpiration rate and comprehensive stress index, collaborative planning is carried out to generate the final regulatory instructions. According to the final control instructions, the ratio and total amount of various control solutions are controlled, and pest and disease control operations are triggered according to the instructions.

2. The method for monitoring and controlling flue-cured tobacco application based on real-time data according to claim 1, characterized in that, The collaborative planning includes long-term planning and short-term planning, wherein, The long-term planning aims to maximize the closeness between the flue-cured tobacco growth status and the yield target curve within a future preset period. It uses reinforcement learning to make sequential decisions in the intelligent decision space to determine the macro-fertilization plan, and uses the flue-cured tobacco transpiration rate as a feedback signal to dynamically adjust the nutrient formulation in the macro-fertilization plan. The short-cycle planning uses crop phenotypic physiological parameters as real-time feedback signals to generate micro-control instructions for dynamically adjusting the conductivity and pH of the water and fertilizer solution during local application. The environmental state vector of the intelligent decision space includes environmental physicochemical parameters, tobacco transpiration rate, crop phenotypic and physiological parameters, preset yield target, and future growth trend predicted by the calibrated digital physiological model.

3. The method for monitoring and controlling flue-cured tobacco application based on real-time data according to claim 2, characterized in that, When the comprehensive stress index exceeds the preset warning threshold, the optimal input determined by the reinforcement learning strategy is blocked, an instruction to apply the preset stress buffer formula is generated, and the preset reward function of the reinforcement learning strategy is switched from the maximum output mode to the minimum physiological stress mode. The preset reward function in the minimized physiological stress mode is the reward value in the maximized output mode minus the product of the preset penalty coefficient and the comprehensive stress index.

4. The method for monitoring and controlling flue-cured tobacco application based on real-time data according to claim 1, characterized in that, The digital physiological model is defined as a discrete-time nonlinear state-space system, and the state vector of the digital physiological model includes the soil conductivity coefficient and photosynthetic efficiency. Based on the multidimensional sensing data, the state vectors in the pre-constructed digital physiological model are calibrated using a parameter calibration algorithm, including: The prior state estimate at the current time is predicted using the posterior estimate from the previous time step, and the prior error covariance matrix at the current time step is predicted using the posterior error covariance matrix from the previous time step. The Kalman gain matrix is ​​calculated by combining the prior error covariance matrix, the preset measurement noise covariance matrix, and the observation matrix; wherein, the observation matrix is ​​used to establish the mapping relationship from the state space to the observation space, the soil biological activity parameter is used as the observation mapping of the soil conductivity coefficient, and the crop phenotypic physiological parameter is used as the observation mapping of photosynthetic efficiency. The prior state estimate is calibrated based on the Kalman gain matrix and the actual observed data vector to obtain the calibrated state vector.

5. The method for monitoring and controlling flue-cured tobacco application based on real-time data according to claim 1, characterized in that, The environmental physicochemical parameters include various environmental indicators during crop growth, including air temperature, wind speed, and light duration. The crop phenotypic and physiological parameters include the degree of linear polarization. and leaf temperature suppression value The soil biological activity parameters include microbial respiration intensity. ,in, The formula for calculating the degree of linear polarization is: ； in, , , , These represent the light radiation intensity values ​​received when the polarizer is rotated to 0°, 90°, 45°, and 135°, respectively. The formula for calculating the blade temperature suppression value is as follows: ； in, The average temperature of the blade surface; Air temperature; This is the wet-bulb temperature, representing the lowest theoretical temperature the leaf can reach under the current humidity conditions. The value is dimensionless. The larger the value, the weaker the leaf's ability to dissipate heat through transpiration, and the stronger the physiological inhibition it experiences. The formula for calculating the microbial respiration intensity is as follows: ； in, The response current value is measured in real time by the sensor; Baseline current noise in sterile soil matrix; The number of electrons transferred in the redox reaction; It is Faraday's constant; This represents the effective surface area of ​​the working electrode of the sensor. This is a correction factor for the effective diffusion coefficient related to soil porosity and moisture content.

6. The method for monitoring and controlling flue-cured tobacco application based on real-time data according to claim 1, characterized in that, The formula for calculating the transpiration rate of the flue-cured tobacco is as follows: ； in, This refers to the instantaneous evaporation rate of flue-cured tobacco. The slope of the saturated water vapor pressure-temperature curve; Net radiation of the canopy; air density; The specific heat of air at constant pressure; The saturated water vapor pressure difference is calculated from the ambient temperature and humidity. For aerodynamic conductivity; The latent heat of vaporization of water; This is the constant of the wet / dry meter; It concerns the leaf temperature inhibition value. The stomatal conductance function is defined as follows: , The maximum porosity is associated with the state vector. The attenuation coefficient; The formula for calculating the comprehensive stress index is as follows: ； in, The preset weighting coefficients satisfy... This represents the normalized degree of linear polarization. The leaf lesion area is extracted using RGB data. The total area of ​​the blades. Represents disease and stress; The current temperature, The optimal temperature for flue-cured tobacco growth is used to characterize thermal environmental stress. The nutrient distribution in the macro-fertilization plan, dynamically adjusted based on the transpiration rate of flue-cured tobacco as feedback signal, is as follows: ； ； in, This is the revised single-control feed-liquid ratio. This refers to the original ratio in the macro-fertilization plan. It is the correction factor for the fertilizer ratio; It is the lower limit of the correction factor; It is a correction of the sensitivity coefficient. This refers to the instantaneous evaporation rate of flue-cured tobacco in real time. The modified formula is based on the preset healthy threshold for instantaneous evaporation rate. This ensures that fertilization is highly synchronized with the plant's real-time physiological and metabolic state.

7. The method for monitoring and controlling flue-cured tobacco application based on real-time data according to claim 2, characterized in that, The formula for dynamically adjusting the conductivity and pH of the water-fertilizer solution during local application is as follows: ； in, represent Temporal conductivity or pH The incremental adjustment; for Leaf temperature suppression value at any time Relative to the median of the target range deviation, that is , These are the proportional, integral, and differential coefficients, respectively. This represents the deviation from the previous moment; , This represents the upper limit of the target range. This is the lower limit of the target range.

8. A monitoring and control system for flue-cured tobacco applying fertilizer based on real-time data, characterized in that, include: The multidimensional data acquisition module is used to acquire multidimensional sensing data in the tobacco planting scenario. The multidimensional sensing data includes environmental physicochemical parameters, soil biological activity parameters, and crop phenotypic and physiological parameters. The model adaptive calibration module is used to calibrate the state vector in the pre-constructed digital physiological model based on the multidimensional sensing data and a parameter calibration algorithm. Based on the calibrated digital physiological model and multidimensional sensing data, the transpiration rate of flue-cured tobacco, which reflects the instantaneous water consumption of crops, and the comprehensive stress index, which characterizes the degree of environmental stress, were calculated. The dual-layer intelligent control module is used to perform collaborative planning based on the calibrated digital physiological model, the transpiration rate of flue-cured tobacco, and the comprehensive stress index, and generate the final control command. The precision execution module is used to control the ratio and total amount of various control solutions according to the final control instructions, and to trigger pest and disease control operations based on the instructions.

9. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the steps of the method described in any one of claims 1-7.

10. A computer device, characterized in that, include: Memory, used to store computer programs / instructions; A processor for executing the computer program / instructions to implement the steps of the method according to any one of claims 1-7.

Images