Intelligent chili seedling cultivation method based on environmental perception and precision management

Abstract

This invention discloses an intelligent chili seedling cultivation method based on environmental perception and precise management, belonging to the field of smart agriculture IoT technology. The method includes real-time collection of environmental factors related to seedling cultivation via a sensor network; generating a quantitative environmental status assessment matrix and early warning signals through spatiotemporal fusion and anomaly diagnosis; comparing this matrix with pre-set ideal environmental models for different growth stages of chili peppers to generate a multi-dimensional environmental control instruction set containing adjustment type, intensity, and execution sequence, driving the actuators to perform ventilation, supplemental lighting, irrigation, fertilization, and temperature control operations; simultaneously collecting chili seedling growth phenotypic data and establishing an environment-phenotype dynamic response model through correlation analysis with the environmental status assessment matrix, and using this model to optimize the control instruction set. This method achieves overall quantitative assessment of the seedling environment and adaptive precise control based on crop growth feedback, improving the accuracy and efficiency of seedling cultivation.

Description

Technical Field

[0001] This invention belongs to the field of smart agriculture Internet of Things technology, specifically an intelligent chili seedling cultivation method based on environmental perception and precise management. Background Technology

[0002] Traditional chili seedling environmental management relies heavily on manual experience or simple automated control systems. Existing technologies typically involve deploying sensors within the seedling facility to monitor environmental factors such as temperature, humidity, and light at single points or in specific areas, setting fixed thresholds to trigger alarms or individual control actions. These methods can only monitor and respond to discrete parameters independently, failing to provide a holistic assessment and diagnosis of the complex microenvironment formed by the spatiotemporal interaction of multiple environmental factors. Environmental control strategies are often static and pre-set, lacking quantitative criteria to determine how the current overall environmental conditions affect seedling growth. This results in extensive and delayed control measures, failing to meet the refined environmental needs of chili seedlings at different growth stages.

[0003] Most existing automated seedling systems follow an open-loop or simple closed-loop control logic of "sensing-judgment-execution." Their "judgment" phase is based on a fixed rule base, and their execution phase is limited to environmental adjustments. These systems generally lack a mechanism for continuous verification and optimization of the actual effects of regulatory actions. After a regulatory command is issued, the system does not actively and systematically collect data on the crop's growth response and correlate this feedback with historical environmental data. Therefore, the system cannot determine whether a specific environmental combination adjustment truly promotes the expected phenotypic development, nor can it adaptively optimize subsequent regulatory strategies based on the actual growth status of the seedlings. This makes it difficult for environmental management to dynamically align with the actual needs of crop growth, hindering further improvements in seedling quality and efficiency. Summary of the Invention

[0004] This invention aims to solve at least one of the technical problems existing in the prior art;

[0005] Therefore, this invention proposes an intelligent chili seedling cultivation method based on environmental perception and precise management, comprising:

[0006] A set of environmental factors related to the seedling environment is collected in real time through a sensor network deployed in the seedling environment.

[0007] The set of seedling environmental factors is subjected to spatiotemporal fusion and anomaly diagnosis processing to generate an environmental status assessment matrix and early warning signal. The environmental status assessment matrix is ​​used to quantitatively characterize the comprehensive impact of the current environment on the growth of pepper seedlings.

[0008] Based on the environmental state assessment matrix and the pre-set ideal environmental model for different growth stages of chili peppers, a multi-dimensional set of environmental regulation instructions is generated, which includes regulation type, regulation intensity and execution sequence.

[0009] The warning signal and the set of multi-dimensional environmental control instructions are sent to the environmental control execution mechanism to drive the environmental control execution mechanism to perform ventilation, supplemental lighting, irrigation, fertilization and temperature control operations;

[0010] A set of growth phenotypic data of pepper seedlings is collected synchronously. The growth phenotypic data set is correlated with the environmental state assessment matrix to establish an environment-phenotype dynamic response model. Based on the environment-phenotype dynamic response model, the set of multi-dimensional environmental regulation instructions is optimized.

[0011] Furthermore, the spatiotemporal fusion and anomaly diagnosis processing of the set of seedling environmental factors to generate an environmental status assessment matrix and early warning signals includes:

[0012] The set of seedling environmental factors includes air temperature and humidity data, light intensity and spectral distribution data, soil or substrate temperature and humidity data, and nutrient solution composition and concentration data.

[0013] Spatial interpolation calculations are performed on environmental factor data of the same type from different spatial locations to generate a continuous distribution field of the seedling environment in three-dimensional space, which includes a temperature field, a humidity field, and a light field.

[0014] Time series analysis is performed on the continuous distribution field to extract the changing trend characteristics, periodic fluctuation characteristics and abrupt change point information of each environmental factor within a preset period;

[0015] Based on the tolerance thresholds of chili seedlings at different growth stages, dynamic early warning threshold curves for each environmental factor were set.

[0016] The data of each item in the set of seedling environmental factors acquired in real time are compared with the dynamic early warning threshold curve at the corresponding time to identify the data points exceeding the standard and potential risk trends.

[0017] By integrating the changing trend characteristics, the periodic fluctuation characteristics, the mutation point information, and the data points exceeding the standard and potential risk trends, an environmental status assessment matrix indexed by time and space is generated, and the corresponding early warning signal is triggered.

[0018] Furthermore, the environmental state assessment matrix is ​​compared and analyzed with a pre-set ideal environmental model for different growth stages of chili peppers to generate a multi-dimensional set of environmental control instructions, including regulation type, regulation intensity, and execution sequence, comprising:

[0019] Based on the current growth stage of the chili seedlings, the corresponding target environmental parameter range is called from the preset ideal environment model for different growth stages of chili.

[0020] The spatial distribution data of each environmental factor in the current moment in the environmental state assessment matrix are compared point by point with the range of the target environmental parameters to calculate the environmental deviation of each spatial point.

[0021] Based on the spatial distribution pattern of the environmental deviation, the areas that need to be prioritized for regulation and the corresponding regulation types are determined. The regulation types include heating, cooling, humidifying, dehumidifying, increasing light intensity, decreasing light intensity, increasing irrigation, decreasing irrigation, supplementing nutrients, or adjusting pH.

[0022] Based on the magnitude of the environmental deviation, the instantaneous physiological water requirement model of the pepper seedlings, and the light compensation point data, the adjustment intensity for each adjustment is calculated comprehensively.

[0023] By combining the response characteristics of environmental control equipment, the coupling influence between various environmental factors, and the principle of avoiding stress on seedlings, the execution sequence of various control operations is planned to form a serialized set of multi-dimensional environmental control instructions.

[0024] Furthermore, the synchronously collected growth phenotypic data set of chili seedlings includes:

[0025] The growth phenotype dataset includes plant height and stem diameter data, leaf quantity and leaf area data, root development image data, and stem and leaf color characteristic data.

[0026] Top-view and side-view images of pepper seedlings are collected periodically using a visible light imaging device. The canopy projection area and inter-leaf porosity are extracted from the top-view image, and the plant height and stem diameter data are extracted from the side-view image.

[0027] Leaf reflectance spectral data are acquired using multispectral or hyperspectral imaging equipment. Based on the leaf reflectance spectral data, chlorophyll content, carotenoid content, and water content are retrieved, and stem and leaf color feature data are extracted.

[0028] The root development image data is obtained by side-view or bottom-view imaging from a specific angle, or by using micro-root window technology, and the root development image data is processed to quantify the total root length, root surface area and number of root tips.

[0029] The canopy projection area, inter-leaf porosity, plant height and stem diameter, chlorophyll content, water content, total root length, root surface area, and number of root tips are used as the core indicators of the growth phenotype data set.

[0030] Furthermore, the step of performing correlation analysis between the growth phenotype data set and the environmental state assessment matrix to establish an environment-phenotype dynamic response model, and then performing feedback optimization on the multi-dimensional environmental regulation command set based on the environment-phenotype dynamic response model, includes:

[0031] Align the environmental state assessment matrix of the time series with the synchronized growth phenotype data set to establish a corresponding dataset of environmental factor time series and phenotypic index time series;

[0032] Correlation analysis and regression analysis were used to screen out key environmental factors that significantly affect specific phenotypic indicators and their lag time from the corresponding dataset;

[0033] Based on the selected key environmental factors and their lag time, an environment-phenotype dynamic response model is constructed to describe the quantitative relationship between environmental stimuli and phenotypic growth.

[0034] The currently executing set of multi-dimensional environmental regulation instructions is input into the environment-phenotype dynamic response model to predict the changing trend of pepper seedling phenotypic indicators in the future.

[0035] If the predicted trend of change deviates from the expected growth trajectory, the adjustment intensity and execution timing in the multi-dimensional environmental control instruction set are recalculated and adjusted to form an optimized environmental control scheme.

[0036] Furthermore, the method of using correlation analysis and regression analysis to screen out key environmental factors that significantly affect specific phenotypic indicators and their lag time from the corresponding dataset includes:

[0037] Calculate the correlation coefficients between the time series of each environmental factor and the time series of each phenotypic index at different time offsets, and form a correlation coefficient matrix;

[0038] From the correlation coefficient matrix, identify the environmental factor-phenotypic index pairs whose absolute correlation coefficients exceed the significance threshold, and record their corresponding lag time.

[0039] For each selected environmental factor-phenotypic index pair, a regression equation is established with the environmental factor as the independent variable and the phenotypic index as the dependent variable. The goodness of fit and significance of the regression equation are then evaluated.

[0040] The regression equation with high goodness of fit and statistical significance, and its corresponding lag time, are taken as the core components of the environment-phenotype dynamic response model.

[0041] Furthermore, the method also includes:

[0042] After irrigation or fertilization is performed, the concentration data of nutrient solution components and the rate of change of substrate moisture content in the set of seedling environmental factors are monitored in real time.

[0043] The monitored rate of change is compared with the expected absorption or diffusion model to determine whether the nutrient or water supply is being used effectively or whether there is a risk of leaching.

[0044] If it is determined that there is a risk of leaching or that the utilization efficiency is significantly lower than expected, the contents of the multi-dimensional environmental control instruction set that have not yet been executed, concerning irrigation volume, nutrient solution concentration, or fertilization frequency, will be dynamically adjusted.

[0045] Furthermore, the method also includes:

[0046] Collect a complete data chain of multiple batches and varieties of pepper seedlings throughout the entire process. The complete data chain includes a complete time series of the seedling environmental factors, execution records of the multi-dimensional environmental regulation command set, and the final growth phenotype data set.

[0047] The complete data chain is used to iteratively train and optimize the environment-phenotype dynamic response model.

[0048] The optimized environment-phenotype dynamic response model is applied to the initial stage of a new batch or new variety of pepper seedlings as the basis for generating the initial set of multi-dimensional environmental control instructions.

[0049] Furthermore, the method also includes:

[0050] Based on the root development image data and the temperature and humidity data of the soil or substrate, a root microenvironment status assessment sub-model is established.

[0051] The root microenvironment status assessment sub-model is used to evaluate rhizosphere oxygen content, water uniformity, and potential disease risk.

[0052] The evaluation results of the root microenvironment status assessment sub-model are used as a special environmental factor and incorporated into the environmental status assessment matrix for comprehensive analysis, triggering targeted root environment regulation instructions, including oxygen-enhancing irrigation, local water regulation, or application of biological agents.

[0053] Furthermore, the establishment of a root microenvironment status assessment sub-model based on the root development image data and the temperature and humidity data of the soil or substrate includes:

[0054] The spatial distribution density and depth of the root system in the matrix were analyzed from the root development image data.

[0055] By combining the distribution of temperature and humidity data of the soil or substrate at the corresponding depth layers, the actual moisture and temperature conditions in different root zones are assessed.

[0056] Based on the aerobic characteristics of chili roots and the oxygen demand patterns at different growth stages, as well as the influence curves of water and temperature on root respiration and absorption, the oxygen diffusion rate and root activity level in the rhizosphere region were estimated.

[0057] By combining the spatial distribution density, actual moisture and temperature conditions, oxygen diffusion rate, and root activity level, a comprehensive index characterizing the health of the root microenvironment is generated, which serves as the core output of the root microenvironment status assessment sub-model.

[0058] Compared with the prior art, the beneficial effects of the present invention are:

[0059] This technology performs spatiotemporal fusion and anomaly diagnosis on a set of seedling environmental factors from a sensor network to generate a quantitative environmental status assessment matrix. This approach abandons the traditional method of viewing environmental parameters in isolation. By integrating spatiotemporal information and diagnosing anomalies through algorithms, it integrates diverse and heterogeneous environmental data into a numerical matrix that can directly quantify the comprehensive impact of the current environment on pepper seedling growth. This matrix provides a high-order, integrated decision indicator for environmental assessment, enabling the system to transcend single-factor threshold judgments and accurately grasp the "health" and "suitability" of the seedling microenvironment as a whole. This provides a precise quantitative target for implementing multi-factor synergistic regulation.

[0060] A dynamic environment-phenotype response model is dynamically established by simultaneously collecting growth phenotypic data of chili seedlings and correlating this data with an environmental status assessment matrix. Based on this empirical model, the system provides feedback optimization to the set of multi-dimensional environmental control commands being executed. This technology constructs a self-evolving closed loop of "environmental control – phenotypic response – model update – strategy optimization." The system continuously learns the actual correspondence between environmental combinations and growth outcomes, constantly revising and refining its control logic. This transforms environmental management strategies from static presets to dynamic self-optimization, enabling adjustments to intervention measures based on real-time physiological feedback from the crop. This ensures that seedling environmental control truly adapts dynamically to the crop's growth needs, improving the predictability and accuracy of control. Attached Figure Description

[0061] Figure 1 This is a flowchart illustrating the steps of the intelligent chili seedling cultivation method based on environmental perception and precise management described in this invention.

[0062] Figure 2 A flowchart for generating a multi-dimensional set of environmental control instructions;

[0063] Figure 3A dynamic change graph of core phenotypic indicators throughout the entire seedling stage of chili peppers;

[0064] Figure 4 A diagram illustrating environmental control instructions for different growth stages of chili peppers;

[0065] Figure 5 A comparison of temperature and humidity at different depths of the pepper seedling substrate with the optimal values ​​for root systems. Detailed Implementation

[0066] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments.

[0067] Please see Figure 1 This paper presents an intelligent chili seedling cultivation method based on environmental perception and precise management. A sensor network deployed in the seedling environment collects real-time data on environmental factors, including air temperature and humidity, light intensity and spectral distribution, soil or substrate temperature and humidity, and nutrient solution concentration. The method performs spatiotemporal fusion and anomaly diagnosis on this environmental factor set, generating an environmental status assessment matrix and early warning signals. The environmental status assessment matrix quantitatively represents the comprehensive impact of the current environment on chili seedling growth. Based on the environmental status assessment matrix and pre-set ideal environmental models for different chili growth stages, a multi-dimensional set of environmental control instructions is generated, including regulation type, intensity, and execution sequence. The early warning signals and the multi-dimensional environmental control instructions are sent to the environmental control actuators, driving them to perform ventilation, supplemental lighting, irrigation, fertilization, and temperature control operations. Simultaneously, growth phenotypic data of the chili seedlings are collected. This data is correlated with the environmental status assessment matrix to establish an environment-phenotype dynamic response model, which is then used to optimize the multi-dimensional environmental control instructions.

[0068] In one embodiment of the present invention, the seedling environmental factor set includes air temperature and humidity data, light intensity and spectral distribution data, soil or substrate temperature and humidity data, and nutrient solution component concentration data. Spatial interpolation is performed on the same type of environmental factor data from different spatial locations to generate a continuous distribution field of the seedling environment in three-dimensional space. This continuous distribution field includes a temperature field, a humidity field, and a light field. Time series analysis is performed on the continuous distribution field to extract the trend characteristics, periodic fluctuation characteristics, and mutation point information of each environmental factor within a preset period. Based on the tolerance thresholds of different growth stages of pepper seedlings, dynamic early warning threshold curves are set for each environmental factor. The data in the real-time seedling environmental factor set are compared with the corresponding dynamic early warning threshold curves to identify exceeding data points and potential risk trends. The trend characteristics, periodic fluctuation characteristics, mutation point information, and exceeding data points and potential risk trends are integrated to generate an environmental state assessment matrix indexed by time and space, triggering corresponding early warning signals.

[0069] In practical implementation, the set of environmental factors for seedling cultivation includes air temperature and humidity data, light intensity and spectral distribution data, soil or substrate temperature and humidity data, and nutrient solution concentration data. A sensor network is deployed at multiple spatial locations in the seedbed or seedling room, including different heights above the seedling canopy, different depths within the seedling substrate, and key nodes in the nutrient solution circulation pipeline. Each location is equipped with corresponding temperature sensors, humidity sensors, light sensors, and ion-selective electrodes. The sensors continuously collect data at fixed time intervals, forming a time-series raw data set covering the entire seedling space and the entire seedling cycle. In some embodiments, spatial interpolation calculations are performed on the same type of environmental factor data from different spatial locations to generate a continuous distribution field of the seedling environment in three-dimensional space. For example, for temperature data, the temperature value of each three-dimensional grid node is calculated using the geometric dimensions of the seedbed and the sensor coordinates, thereby constructing a temperature field. Similarly, interpolation calculations are performed on humidity and light data respectively to generate humidity and light fields. These continuous distribution fields transform discrete sensor observation point data into a continuous model that reflects the environmental conditions of the entire seedling space.

[0070] In practical implementation, time series analysis is performed on the continuous distribution field to extract the changing trend characteristics, periodic fluctuation characteristics, and abrupt change information of each environmental factor within a preset period. The changing trend characteristics are represented by the slope values ​​obtained from the time series data of the temperature, humidity, and light fields at each grid node through linear regression analysis. The periodic fluctuation characteristics are identified by Fourier transform to determine daily or hourly periodic patterns. Abrupt change information is obtained by using a cumulative sum algorithm to detect the time points of sudden changes in environmental factors. This can be understood as setting dynamic early warning threshold curves for each environmental factor based on the tolerance thresholds of different growth stages of pepper seedlings. For example, during the cotyledon unfolding stage, the temperature early warning threshold curve is set to 25-28℃ during the day and 18-20℃ at night; during the two-leaf-one-heart stage, the temperature early warning threshold curve is adjusted to 22-25℃ during the day and 16-18℃ at night. Each growth stage has corresponding dynamic early warning threshold curves for parameters such as light intensity, humidity, and nutrient solution conductivity.

[0071] In practical implementation, the data from the real-time acquired set of seedling environmental factors are compared with the corresponding dynamic early warning threshold curve to identify data points exceeding the standard and potential risk trends. The system continuously monitors whether the temperature value of each grid node in the temperature field exceeds the current early warning threshold range, recording all locations and times exceeding the threshold as exceeding data points. Simultaneously, the rate of decrease in the humidity field over a recent period is analyzed; if the rate exceeds a preset value, it is marked as a potential risk trend of dryness. Optionally, the system integrates trend characteristics, periodic fluctuation characteristics, abrupt change information, and exceeding data points and potential risk trends to generate an environmental state assessment matrix indexed by time and space. The environmental state assessment matrix is ​​a three-dimensional data structure, where two dimensions represent spatial location (e.g., the planar coordinates of the seedbed), and the third dimension represents a time point. The value of each element in the matrix is ​​calculated by weighted summation of various characteristics at that location and time point. Specifically, the location in the matrix... In time element value Calculated using the following formula:

[0072] ;

[0073] in: Characteristic values ​​representing the normalized temperature change trend; Indicates the deviation of normalized periodic fluctuations; A quantitative value representing the impact of the mutation point; Indicates the severity of the data exceeding the standard; Indicates the intensity of potential risk; to These are the weighting coefficients for each feature. This design allows the environmental state assessment matrix to quantitatively characterize the overall impact of the current environment on the growth of chili seedlings.

[0074] In some embodiments, corresponding early warning signals are triggered based on the numerical changes of specific areas in the environmental status assessment matrix at continuous time points and the spatial clustering of data points exceeding the standard. The early warning signals are divided into different levels; for example, a red high-temperature early warning signal is triggered when a large area of ​​the temperature field exceeds the standard and the environmental status assessment matrix value continues to rise; a yellow humidity early warning signal is triggered when the humidity field shows a drying trend but has not yet reached the exceeding threshold. The early warning signals include specific spatial location information and temporal information, as well as suggested areas of focus. It can be understood that the entire spatiotemporal fusion and anomaly diagnosis process is automatically executed on a computing device. The computing device connects to a sensor network and receives real-time data, performs spatial interpolation, time series analysis, threshold comparison, and feature fusion algorithms, and finally outputs the environmental status assessment matrix and early warning signals. The environmental status assessment matrix and early warning signals provide direct input for the generation of subsequent environmental control instructions.

[0075] See Figure 2 In one embodiment of the present invention, based on the current growth stage of the chili seedlings, the corresponding target environmental parameter range is retrieved from a pre-set ideal environmental model for different growth stages of chili peppers. The spatial distribution data of each environmental factor at the current moment in the environmental state assessment matrix is ​​compared point by point with the target environmental parameter range to calculate the environmental deviation at each spatial point. Based on the spatial distribution pattern of the environmental deviation, the areas requiring priority regulation and the corresponding regulation types are determined. Regulation types include heating, cooling, humidifying, dehumidifying, increasing light intensity, decreasing light intensity, increasing irrigation, decreasing irrigation, supplementing nutrients, or adjusting pH. Based on the magnitude of the environmental deviation, the instantaneous physiological water requirement model of the chili seedlings, and the light compensation point data, the regulation intensity for each regulation is comprehensively calculated. Combining the response characteristics of the environmental control equipment, the coupling influence relationship between various environmental factors, and the principle of avoiding stress on the seedlings, the execution sequence of each regulation operation is planned to form a serialized multi-dimensional environmental control instruction set.

[0076] In practical implementation, based on the current growth stage of the chili seedlings, the system retrieves the corresponding target environmental parameter ranges from pre-set ideal environmental models for different chili growth stages. These ideal environmental models are stored in the system as data tables, defining target ranges for daytime and nighttime temperature, air humidity, light intensity and photoperiod, substrate humidity, and nutrient solution conductivity and pH for each stage from seed germination, cotyledon unfolding, true leaf growth to seedling maturity. For example, when the system determines through image recognition that the seedling population has entered the "two leaves and one heart" growth stage, it automatically retrieves the target environmental parameter ranges for that stage, where the daytime temperature target range is set to 22℃ to 25℃, and the light intensity target range is 300 μmol·m⁻²·s⁻¹ to 400 μmol·m⁻²·s⁻¹. In some embodiments, the spatial distribution data of each environmental factor at the current moment in the environmental state assessment matrix is ​​compared point-by-point with the target environmental parameter ranges to calculate the environmental deviation at each spatial point. The environmental status assessment matrix provides the actual values ​​of factors such as temperature, humidity, and light at each two-dimensional coordinate point (i,j) in the seedbed. The system subtracts these actual values ​​from the median of the target environmental parameter range to obtain the deviation value of each environmental factor at each coordinate point.

[0077] In practical implementation, based on the spatial distribution pattern of environmental deviations, the regions requiring priority regulation and their corresponding regulation types are determined. The system analyzes the environmental deviations at all coordinate points, generating a spatial distribution heatmap of environmental deviations marked with different colors. When the absolute value of the same type of environmental deviation in a continuous region exceeds a set threshold, that region is marked as requiring priority regulation. The regulation type is directly determined based on the sign of the environmental deviation and the factor type. For example, for a continuous high-temperature region (where the temperature environmental deviation is significantly positive), the regulation type is determined as "cooling"; for a continuous low-light region (where the light intensity environmental deviation is significantly negative), the regulation type is determined as "enhancing light". It can be understood that the regulation intensity for each regulation is calculated comprehensively based on the magnitude of the environmental deviation, the instantaneous physiological water requirement model of pepper seedlings, and light compensation point data. The calculation of the regulation intensity is a multi-factor fusion process. Taking irrigation regulation as an example, the system not only considers the environmental deviation of substrate moisture but also combines the transpiration rate of seedlings under current light conditions (from the instantaneous physiological water requirement model) and historical water use efficiency data. For supplemental lighting regulation, the intensity is determined jointly based on the environmental deviation of light intensity and the current light compensation point data of the pepper seedlings, ensuring that the supplemental lighting intensity can both compensate for the deficiency and avoid causing photoinhibition. In a specific calculation, the irrigation regulation intensity for a particular area... The following formula is used to derive:

[0078]

[0079] in: This indicates the environmental deviation in substrate moisture content; This represents the transpiration rate per unit area calculated using the real-time physiological water demand model. This indicates the residual effective water content of the substrate at the current moment; , and It is an empirical coefficient.

[0080] In some embodiments, the execution sequence of various regulatory operations is planned by combining the response characteristics of environmental control equipment, the coupling influence between various environmental factors, and the principle of avoiding stress on seedlings. The system internally stores equipment parameters for all environmental control actuators, including the heating rate of heating equipment, the closing time of shading curtains, and the flow rate of irrigation pumps. When planning the execution sequence, the system prioritizes operations that can alleviate the most urgent environmental stress and have short equipment response times. For example, when both localized high temperatures and overall low humidity require regulation, the system will prioritize activating the evaporative cooling system for cooling and humidification, as this operation can positively influence both factors simultaneously, and cooling is more urgent for alleviating heat stress in seedlings. Optionally, for operations with coupling conflicts, the system will stagger the timing; for example, before preparing for foliar fertilization or pesticide application, planned irrigation operations will be postponed or canceled to avoid rinsing off droplets. Finally, the system packages the type of regulation to be executed, the intensity of regulation for a specific area, and the execution time instructions accurate to the minute, forming a serialized multi-dimensional set of environmental control instructions. It is understandable that the multi-dimensional environmental control command set is generated in the form of a structured command list. Each command in the list clearly includes the execution device number, action type, action intensity parameters, and a predetermined execution timestamp. This command set is directly sent to the central controller of the environmental control execution mechanism, driving ventilation, supplemental lighting, irrigation, fertilization, and temperature control equipment to operate according to predetermined sequences and parameters.

[0081] In one embodiment of the present invention, the growth phenotypic data set includes plant height and stem diameter data, leaf number and leaf area data, root development image data, and stem and leaf color feature data. Top-view and side-view images of pepper seedlings are periodically acquired using a visible light imaging device. The canopy projection area and inter-leaf porosity are extracted from the top-view image, and plant height and stem diameter data are extracted from the side-view image. Leaf reflectance spectral data are acquired using a multispectral or hyperspectral imaging device. Chlorophyll content, carotenoid content, and water content are inverted based on the leaf reflectance spectral data, and stem and leaf color feature data are extracted. Root development image data is obtained through side-view or bottom-view imaging at a specific angle, or by using micro-root window technology. The root development image data is then processed to quantify root length, root surface area, and root tip number. The core indicators of the growth phenotypic data set were canopy projection area, inter-leaf porosity, plant height and stem diameter, chlorophyll content, water content, total root length, root surface area, and root tip number. The environmental state assessment matrix of the time series was aligned with the synchronous growth phenotypic data set to establish a corresponding dataset of environmental factor time series and phenotypic index time series. Correlation and regression analyses were used to screen key environmental factors that significantly affected specific phenotypic indices and their lag times from the corresponding datasets. The correlation coefficients between the time series of each environmental factor and the time series of each phenotypic index at different time offsets were calculated to form a correlation coefficient matrix. Environmental factor-phenotypic index pairs with absolute correlation coefficients exceeding the significance threshold were identified from the correlation coefficient matrix, and their corresponding lag times were recorded. For each selected environmental factor-phenotypic index pair, a regression equation was established with the environmental factor as the independent variable and the phenotypic index as the dependent variable. The goodness of fit and significance of the regression equations were evaluated. Regression equations with high goodness of fit and statistical significance, along with their corresponding lag times, were used as the core components of the environment-phenotypic dynamic response model. Based on the selected key environmental factors and their lag time, an environment-phenotype dynamic response model is constructed to describe the quantitative relationship between environmental stimuli and phenotypic growth. The set of multi-dimensional environmental regulation instructions currently being executed is input into the environment-phenotype dynamic response model to predict the changing trend of pepper seedling phenotypic indicators in the future. If the predicted changing trend deviates from the expected growth trajectory, the regulation intensity and execution sequence in the set of multi-dimensional environmental regulation instructions are recalculated and adjusted to form an optimized environmental regulation scheme.

[0082] In practice, the growth phenotypic dataset includes plant height and stem diameter data, leaf quantity and leaf area data, root development image data, and stem and leaf color characteristic data. Visible light imaging equipment is used to periodically collect top-view and side-view images of pepper seedlings. The visible light imaging equipment is fixed on a guide rail directly above and to the side of the seedbed, and images are taken daily at fixed times according to a preset program. From the top-view images, image segmentation algorithms are used to extract the canopy projection area and inter-leaf porosity. From the side-view images, plant height and stem diameter data are extracted using calibration references and edge detection algorithms. Multispectral or hyperspectral imaging equipment is used to collect leaf reflectance spectral data. Multispectral imaging equipment has multiple specific wavelength filters in the visible and near-infrared bands. Based on the leaf reflectance spectral data, chlorophyll content, carotenoid content, and water content are calculated using a pre-established reflectance spectrum and biochemical parameter inversion model, and stem and leaf color characteristic data are extracted. Root development image data is acquired through side-view or bottom-view imaging from specific angles, or by using micro-root window technology. Micro-root window technology involves installing a transparent observation surface on one side of the seedling container, with the root growth close to the observation surface. A high-resolution camera is used to periodically photograph the observation surface. The acquired root development image data is then processed through binarization, skeletonization, and topological analysis to quantify root length, root surface area, and root tip number. Canopy projection area, inter-leaf porosity, plant height and stem diameter, chlorophyll content, water content, total root length, root surface area, and root tip number are used as core indicators of the growth phenotypic data set, and a time-series archive of phenotypic data for each individual seedling is established.

[0083] In practice, the environmental state assessment matrix of the time series is aligned with the synchronous growth phenotypic data set to establish a corresponding dataset of environmental factor time series and phenotypic index time series. The alignment operation uses a unified timestamp as a benchmark. The environmental state assessment matrix provides spatially averaged environmental data at hourly intervals, while the growth phenotypic data set provides phenotypic index data at daily intervals. The corresponding dataset is a structured table, with each row recording the observed phenotypic index (e.g., average plant height) for a specific date and specific seedling area, along with environmental factor data (e.g., daily average temperature, daily cumulative light) that were synchronous with the observed phenotypic index for several consecutive days prior, but with different lag days. Correlation and regression analysis methods are used to screen out key environmental factors that significantly affect specific phenotypic indices and their lag times from the corresponding dataset. The correlation coefficients between the time series of each environmental factor and the time series of each phenotypic index at different time offsets are calculated, forming a correlation coefficient matrix. For example, the correlation coefficients between yesterday's average temperature and today's plant height, the correlation coefficients between the average temperature of the day before yesterday and today's plant height are calculated, and so on, until the time offset with the largest absolute value of the correlation coefficient is found. From the correlation coefficient matrix, identify environmental factor-phenotypic indicator pairs whose absolute correlation coefficients exceed a pre-set significance threshold (e.g., 0.7) and record their corresponding lag times. For each selected environmental factor-phenotypic indicator pair, establish a regression equation with the environmental factor as the independent variable and the phenotypic indicator as the dependent variable, and evaluate the goodness of fit and significance of the regression equation. It can be understood that regression equations with high goodness of fit and statistical significance, along with their corresponding lag times, are considered core components of the environmental-phenotypic dynamic response model. For example, a regression equation included in the model might be in the form of:

[0084] ;

[0085] in: Indicates the first Plant height in the sky; The intercept; and These are the regression coefficients; Indicates the first The average daily air temperature of the day; Indicates the first Heaven to the Di The cumulative photosynthetically active radiation over the previous two days. This equation shows that plant height growth is significantly correlated with the average temperature of the previous two days and the cumulative light intensity of the previous three days.

[0086] In some embodiments, an environment-phenotypic dynamic response model is constructed based on the selected key environmental factors and their lag times, describing the quantitative relationship between environmental stimuli and phenotypic growth. The environment-phenotypic dynamic response model consists of multiple regression equations similar to those described above, collectively predicting the changing trends of pepper seedling phenotypic indicators over a future period. The currently executing set of multi-dimensional environmental control instructions is input into the environment-phenotypic dynamic response model to predict the changing trends of pepper seedling phenotypic indicators over a future period. During prediction, the system uses planned future environmental parameters (such as planned temperature and light intensity for the next three days) from the multi-dimensional environmental control instruction set as input variables into the model to calculate the predicted values ​​of key phenotypic indicators (such as plant height and leaf area) for the next few days. If the predicted trend deviates from the expected growth trajectory, the regulatory intensity and execution sequence in the multi-dimensional environmental control instruction set are recalculated and adjusted to form an optimized environmental control scheme. For example, if the model predicts that the daily increase in plant height will exceed the predetermined range in the next five days under the current control instructions, the system will automatically adjust the temperature setpoint or light duration for the following days, generating a new set of multi-dimensional environmental control instructions aimed at bringing the growth rate back to the expected trajectory, and replacing some of the instructions in the original plan that have not yet been executed. Optionally, this feedback optimization process is triggered every time a new set of environmental control instructions is generated or new phenotypic observation data is received, achieving closed-loop control.

[0087] See Figure 3 This is a dynamic chart showing the changes in core phenotypic indicators throughout the entire seedling stage of chili peppers. It visually presents the trends in plant height, stem diameter, chlorophyll content, and total root length across four growth stages. The growth rate of total root length is significantly higher than that of the above-ground parts, increasing approximately 14 times from germination to hardening-off, indicating that rapid root development during the seedling stage is the core foundation for robust plant growth. Plant height, stem diameter, and chlorophyll content show a steady upward trend, with plant height increasing approximately 6.3 times, stem diameter approximately 3.8 times, and chlorophyll content approximately 2.5 times, reflecting the synergistic growth relationship between the above-ground and below-ground parts. All indicators show the greatest increase during the hardening-off period, a critical stage for seedlings transitioning from a greenhouse environment to a field environment. The rapid expansion of the root system and accumulation of chlorophyll lay the foundation for post-transplant stress resistance.

[0088] In one embodiment of the present invention, after irrigation or fertilization is performed, the rate of change of nutrient solution concentration data and substrate moisture content in the seedling environmental factor set is monitored in real time. The monitored rate of change is compared with the expected absorption or diffusion model to determine whether the nutrient or water supply is effectively utilized or whether there is a risk of leaching. If a risk of leaching is determined or the utilization efficiency is significantly lower than expected, the unexecuted instructions regarding irrigation volume, nutrient solution concentration, or fertilization frequency in the multi-dimensional environmental control instruction set are dynamically adjusted. A complete data chain is collected for the entire process of multiple batches and varieties of pepper seedlings. The complete data chain includes a complete time series of the seedling environmental factor set, execution records of the multi-dimensional environmental control instruction set, and the final growth phenotypic data set. The complete data chain is used to iteratively train and optimize the environment-phenotype dynamic response model. The optimized environment-phenotype dynamic response model is applied to the initial stage of a new batch or new variety of pepper seedlings as the basis for generating the initial multi-dimensional environmental control instruction set.

[0089] In practice, after irrigation or fertilization, the system monitors the rate of change in nutrient solution concentration and substrate moisture content in the seedling environment in real time. The system continuously collects nutrient solution concentration data using conductivity and pH sensors deployed in the nutrient solution circulation pipeline, and collects substrate moisture content data using moisture sensors buried at different depths in the seedling substrate. Within 30 minutes of completing a quantitative irrigation operation, the system records the process of the nutrient solution conductivity value decreasing from 1.8 mS / cm to 1.5 mS / cm at a frequency of once per minute, while simultaneously recording the curve of substrate moisture content rising from 45% to 60% and then slowly decreasing. The monitored rates of change are compared with the expected absorption or diffusion model, which is based on the historical average nutrient absorption rate and substrate moisture diffusion characteristics of pepper seedlings at the current growth stage (e.g., four-leaf stage). The system calculates the ratio of the actual rate of decrease in nutrient solution conductivity to the model-predicted rate, and also calculates the ratio of the rate of decrease after the substrate moisture content reaches its peak to the model-predicted transpiration absorption rate. The system determines whether nutrient or water supply is being effectively utilized or whether there is a risk of leaching. For example, if the rate of decrease in nutrient solution conductivity is significantly faster than predicted by the model while the substrate moisture content decreases slowly, the system determines that there is a risk of nutrient leaching. If the rate of decrease in substrate moisture content is much lower than the transpiration rate predicted by the model, the system determines that water use efficiency is lower than expected. In some embodiments, if a risk of leaching or significantly lower-than-expected utilization efficiency is determined, the system dynamically adjusts the unexecuted instructions regarding irrigation volume, nutrient solution concentration, or fertilization frequency in the multi-dimensional environmental control instruction set. The system retrieves irrigation and fertilization instructions planned for execution within the next 24 hours, modifying the original single irrigation volume from 200 ml per plant to 150 ml, adjusting the nutrient solution stock solution addition ratio from 1:100 to 1:120, or postponing the next fertilization time from the originally planned six hours later to twelve hours later. These adjustments aim to make the nutrient supply more closely match the actual absorption capacity of the seedlings.

[0090] In practice, a complete data chain was collected for the entire process of multi-batch, multi-variety pepper seedling cultivation. This complete data chain includes a complete time series of seedling environmental factors, execution records of multi-dimensional environmental control commands, and the final growth phenotypic data set. The time series of seedling environmental factors records data such as air temperature, humidity, light intensity, substrate moisture, and nutrient solution concentration at minute or hour intervals. The execution records of the multi-dimensional environmental control commands detail the issuance time, executing equipment, action parameters, and actual completion status of each control command. The final growth phenotypic data set summarizes indicators such as plant height, stem diameter, leaf area, and root length for each seedling unit at the end of the seedling cycle. The complete data chain is stored in a structured database table format, with each batch of seedling data associated with a specific variety identifier and seedling time interval. See Table 1 for a simplified table of the complete data chain elements.

[0091] Table 1: Complete Data Link Element Table

[0092]

[0093] A complete data chain is used to iteratively train and optimize the parameters of an environment-phenotype dynamic response model. The training process uses the time series of environmental factors from the complete data chain as input features and the growth phenotype dataset as the target label. The model's internal parameters are adjusted by minimizing the error between the predicted and actual phenotypes. The parameter optimization employs a gradient-based method, with the core update formula expressed as:

[0094] ;

[0095] in: This represents the updated model parameter vector; This represents the model parameter vector before the update; Indicates the learning rate; Represents the loss function Regarding parameters The gradient; This represents the set of growth phenotypic data actually observed. This represents the growth phenotypic data predicted by the environment-phenotype dynamic response model. Each iteration uses a complete batch of data for calculation. After multiple iterations, the model parameters converge, and the accuracy of the model's prediction of phenotypic indicators is improved. The optimized environment-phenotype dynamic response model is applied to the initial stage of a new batch or variety of pepper seedlings as the basis for generating the initial set of multi-dimensional environmental control instructions. When formulating the first week's seedling plan for a new batch of "Pepper Variety A," the system no longer relies entirely on the general ideal environment model, but instead inputs the known varietal characteristic parameters of "Pepper Variety A" into the optimized environment-phenotype dynamic response model. Optionally, based on the parameters of the new variety and data of historically similar varieties, the model infers a more precise temperature and humidity range required by the variety during the germination period. Based on this inference result, the system generates initial daily temperature control instructions and humidity maintenance instructions, thereby providing a more personalized initial environmental control strategy. In some embodiments, for new varieties lacking historical data, the system uses some more general parameters from the optimized model as a basis, combined with limited known biological characteristics, to generate a relatively conservative set of initial environmental regulation instructions, and then rapidly fine-tunes the model through real-time data collection after the seedling stage begins.

[0096] See Figure 4This is a chart analyzing environmental regulation commands for chili peppers at different growth stages, clearly demonstrating the dynamic adjustment strategies for irrigation frequency, temperature, humidity, light, and nutrient regulation at different growth stages. Irrigation frequency peaks at the four-leaf stage (4 times / day) and then declines before transplanting, reflecting the maximum water demand of seedlings during their rapid growth period. Appropriate water control before transplanting promotes root lignification and improves transplant survival rate. Temperature regulation intensity is highest during germination (2.5), as seed germination is most sensitive to temperature and requires precise temperature control. As seedlings develop, temperature regulation intensity gradually decreases, reflecting the increased stress resistance of the plant. Light and nutrient regulation intensity continuously increases from germination to the four-leaf stage, reaching a peak at the four-leaf stage (light 2.5, nutrient 1.5), perfectly aligning with the physiological pattern of increased photosynthesis and nutrient demand after true leaf unfolding. Humidity regulation intensity remains at a moderate level of 1.5-1.8 from the cotyledon stage to the four-leaf stage, avoiding diseases caused by high humidity while ensuring the transpiration needs of seedlings.

[0097] In one embodiment of the present invention, a root microenvironment status assessment sub-model is established based on root development image data and soil or substrate temperature and humidity data. The spatial distribution density and depth of roots in the substrate are analyzed from the root development image data. Combined with the distribution of soil or substrate temperature and humidity data at corresponding depth layers, the actual moisture and temperature conditions in different root zones are assessed. Based on the aerobic characteristics of pepper roots, the oxygen demand patterns at different growth stages, and the influence curves of water and temperature on root respiration and absorption functions, the oxygen diffusion rate and root activity level in the rhizosphere are calculated. A comprehensive index characterizing the health of the root microenvironment is generated by integrating spatial distribution density, actual moisture and temperature conditions, oxygen diffusion rate, and root activity level, serving as the core output of the root microenvironment status assessment sub-model. Through the root microenvironment status assessment sub-model, rhizosphere oxygen content, moisture uniformity, and potential disease risk are assessed. The assessment results of the root microenvironment status assessment sub-model are treated as a special environmental factor and incorporated into the environmental status assessment matrix for comprehensive analysis, triggering targeted root environment regulation commands, including oxygen-enhancing irrigation, localized water regulation, or application of biological agents.

[0098] In practical implementation, a sub-model for assessing the root microenvironment is established based on root development image data and soil or substrate temperature and humidity data. Root development image data is acquired through micro-root window technology or a bottom-view imaging system, displaying the two-dimensional distribution morphology of roots in the substrate. The spatial distribution density and depth of roots in the substrate are analyzed from the root development image data. Image processing algorithms are used to identify and quantify the root skeleton, calculating the pixel length of roots per unit area as the distribution density, and statistically analyzing the vertical coordinate range of the main root distribution as the depth. For example, in one analysis, image processing results showed that the root distribution density was highest in the middle layer of the seedling container (depth 5cm to 10cm), reaching 15cm root length per square centimeter, while the density was lower in the surface and bottom layers. Combining the distribution of soil or substrate temperature and humidity data at corresponding depths, the actual moisture and temperature conditions in different root zones are assessed. The soil or substrate temperature and humidity data are obtained from sensor arrays buried at different depths (e.g., 3cm, 8cm, 13cm) in the substrate. The system matches the root depth distribution data with the sensor data of the same depth layer. For the 5cm to 10cm layer with dense roots, the average value of the temperature sensor for this layer is 22℃, and the average value of the humidity sensor is 58% volume water content.

[0099] In some embodiments, based on the aerobic characteristics of chili pepper roots and the oxygen demand patterns at different growth stages, as well as the influence curves of water and temperature on root respiration and absorption functions, the oxygen diffusion rate and root activity level in the rhizosphere are calculated. The aerobic characteristics of chili pepper roots are represented parametrically within the system; for example, the root oxygen demand at the "four-leaf stage" is set to 0.5 mg of oxygen per gram of root dry weight per hour. The influence curves of water and temperature on root respiration and absorption functions are stored in a lookup table, indicating that the root respiration rate is at 85% of its optimal range under conditions of 22℃ and 58% water content. When calculating the oxygen diffusion rate, the system uses a physical model based on matrix porosity, water content, and temperature to calculate the rate of oxygen diffusion from the matrix air phase to the root surface. It can be understood that by integrating spatial distribution density, actual water and temperature conditions, oxygen diffusion rate, and root activity level, a comprehensive index characterizing the health of the root microenvironment is generated as the core output of the root microenvironment status assessment sub-model. The calculation of the comprehensive index integrates the above-mentioned multiple factors; for example, a specific calculation formula is:

[0100]

[0101] in: This represents the comprehensive health index of the root microenvironment; Represents the spatial distribution density of the root system (unit: ); Indicates the actual root zone temperature (unit: °C); This indicates the optimal temperature for chili pepper roots at this growth stage (unit: °C). Indicates the effective temperature range (unit: °C); This represents the estimated oxygen diffusion rate (unit: ); This indicates the actual volumetric water content; Indicates the saturated water content of the matrix; and This is the adjustment factor. This formula quantifies the effects of density, temperature suitability, oxygen supply, and supersaturated water stress on root health.

[0102] In practice, a root microenvironment status assessment sub-model is used to evaluate rhizosphere oxygen content, water uniformity, and potential disease risk. Rhizosphere oxygen content is calculated based on the oxygen diffusion rate. Indirect representation, when When the value is below a set threshold, the system assesses insufficient rhizosphere oxygen content. Moisture uniformity is assessed by analyzing the standard deviation of humidity sensor data at different depths. For example, if the readings of three sensors are 52%, 58%, and 49% respectively, the standard deviation is relatively large, and the system assesses poor moisture uniformity. Potential disease risk is assessed using an empirical function. The function's inputs include the duration of sustained high humidity (>70%) in the root zone, the temperature range, and a comprehensive root health index. The recent downward trend. The assessment results of the root microenvironment status assessment sub-model are treated as a special environmental factor and incorporated into the environmental status assessment matrix for comprehensive analysis. The system creates a new environmental factor dimension called "root microenvironment status," the value of which is determined by a comprehensive index. After normalization, this dimension is added to the original dimensions such as temperature, humidity, and light, forming an expanded environmental state assessment matrix. In some embodiments, it triggers targeted root environment regulation commands, including aeration irrigation, local water regulation, or application of biological agents. For example, when the root microenvironment health comprehensive index... When root zone oxygen levels remain consistently low and are assessed as insufficient, the system generates an "oxygen-enhancing irrigation" command, controlling the irrigation system to mix in water with higher dissolved oxygen content or briefly inject air bubbles during the next irrigation. Optionally, when water uniformity is assessed as poor, the system generates a "local water regulation" command, adjusting the water output or start / stop time of the drip irrigation heads for specific depth layers that are shown to be too dry or too wet by sensors. For areas assessed as having a high risk of potential disease growth, the system may generate a "bio-inoculant application" command, adding specified beneficial microbial agents to the irrigation solution. These root environment control commands, like those for conventional environmental control commands targeting air temperature, humidity, and light, are integrated into a multi-dimensional environmental control command set and executed by a unified central control system.

[0103] See Figure 5 This is a comparison chart of temperature and humidity at different depths in the pepper seedling substrate and their optimal values ​​for root development, showing the matching of actual temperature and humidity with the optimal values ​​for pepper roots in the surface, middle, and bottom layers. The middle layer is the core area for root growth: the actual temperature (22.0℃) in the middle layer (5-10cm) is closest to the optimal value for roots (22.5℃), perfectly matching the observation results of dense root distribution, making it the core root zone supporting seedling growth. The temperature decreases gradually from the surface to the bottom layer, with the surface layer being too hot (+2.0℃) and the bottom layer being too cold (-2.0℃), which may lead to excessive respiration in the surface roots and insufficient activity in the bottom roots. The moisture content decreases from the middle layer to the top and bottom layers, with the middle layer being too wet (+3.0%) and the bottom layer being too dry (-6.0%), which can easily lead to the risk of oxygen deficiency in the middle layer roots and water stress in the bottom roots. The middle layer has a high moisture content (58%) and the temperature is close to the optimal value. High humidity will reduce substrate porosity, restrict oxygen diffusion, and may exacerbate the risk of rhizosphere hypoxia.

Claims

1. An intelligent chili seedling cultivation method based on environmental perception and precise management, characterized in that, The method includes: A set of environmental factors related to the seedling environment is collected in real time through a sensor network deployed in the seedling environment. The set of seedling environmental factors is subjected to spatiotemporal fusion and anomaly diagnosis processing to generate an environmental status assessment matrix and early warning signal. The environmental status assessment matrix is ​​used to quantitatively characterize the comprehensive impact of the current environment on the growth of pepper seedlings. Based on the environmental state assessment matrix and the pre-set ideal environmental model for different growth stages of chili peppers, a multi-dimensional set of environmental regulation instructions is generated, which includes regulation type, regulation intensity and execution sequence. The warning signal and the set of multi-dimensional environmental control instructions are sent to the environmental control execution mechanism to drive the environmental control execution mechanism to perform ventilation, supplemental lighting, irrigation, fertilization and temperature control operations; A set of growth phenotypic data of pepper seedlings is collected synchronously. The set of growth phenotypic data is correlated with the environmental state assessment matrix to establish an environment-phenotype dynamic response model. The set of multi-dimensional environmental regulation instructions is then optimized based on the environment-phenotype dynamic response model. The process of performing spatiotemporal fusion and anomaly diagnosis on the set of seedling environmental factors to generate an environmental status assessment matrix and early warning signals includes: The set of seedling environmental factors includes air temperature and humidity data, light intensity and spectral distribution data, soil or substrate temperature and humidity data, and nutrient solution composition and concentration data. Spatial interpolation calculations are performed on environmental factor data of the same type from different spatial locations to generate a continuous distribution field of the seedling environment in three-dimensional space, which includes a temperature field, a humidity field, and a light field. Time series analysis is performed on the continuous distribution field to extract the changing trend characteristics, periodic fluctuation characteristics and abrupt change point information of each environmental factor within a preset period; Based on the tolerance thresholds of chili seedlings at different growth stages, dynamic early warning threshold curves for each environmental factor were set. The data of each item in the set of seedling environmental factors acquired in real time are compared with the dynamic early warning threshold curve at the corresponding time to identify the data points exceeding the standard and potential risk trends. By integrating the changing trend characteristics, the periodic fluctuation characteristics, the mutation point information, and the data points exceeding the standard and potential risk trends, an environmental status assessment matrix indexed by time and space is generated, and the corresponding early warning signal is triggered. The environmental state assessment matrix is ​​compared and analyzed with a pre-set ideal environmental model for different growth stages of chili peppers to generate a multi-dimensional set of environmental control instructions, including regulation type, regulation intensity, and execution sequence, including: Based on the current growth stage of the chili seedlings, the corresponding target environmental parameter range is called from the preset ideal environment model for different growth stages of chili. The spatial distribution data of each environmental factor in the current moment in the environmental state assessment matrix are compared point by point with the range of the target environmental parameters to calculate the environmental deviation of each spatial point. Based on the spatial distribution pattern of the environmental deviation, the areas that need to be prioritized for regulation and the corresponding regulation types are determined. The regulation types include heating, cooling, humidifying, dehumidifying, increasing light intensity, decreasing light intensity, increasing irrigation, decreasing irrigation, supplementing nutrients, or adjusting pH. Based on the magnitude of the environmental deviation, the instantaneous physiological water requirement model of the pepper seedlings, and the light compensation point data, the adjustment intensity for each adjustment is calculated comprehensively. Based on the response characteristics of environmental control equipment, the coupling influence between various environmental factors, and the principle of avoiding stress on seedlings, the execution sequence of various control operations is planned to form a serialized set of multi-dimensional environmental control instructions. The step of performing correlation analysis between the growth phenotype data set and the environmental state assessment matrix to establish an environment-phenotype dynamic response model, and then performing feedback optimization on the multi-dimensional environmental regulation command set based on the environment-phenotype dynamic response model, includes: Align the environmental state assessment matrix of the time series with the synchronized growth phenotype data set to establish a corresponding dataset of environmental factor time series and phenotypic index time series; Correlation analysis and regression analysis were used to screen out key environmental factors that significantly affect specific phenotypic indicators and their lag time from the corresponding dataset; Based on the selected key environmental factors and their lag time, an environment-phenotype dynamic response model is constructed to describe the quantitative relationship between environmental stimuli and phenotypic growth. The currently executing set of multi-dimensional environmental regulation instructions is input into the environment-phenotype dynamic response model to predict the changing trend of pepper seedling phenotypic indicators in the future. If the predicted trend of change deviates from the expected growth trajectory, the adjustment intensity and execution timing in the multi-dimensional environmental control instruction set are recalculated and adjusted to form an optimized environmental control scheme.

2. The intelligent chili seedling cultivation method based on environmental perception and precise management according to claim 1, characterized in that, The synchronously collected growth phenotypic data set of pepper seedlings includes: The growth phenotype dataset includes plant height and stem diameter data, leaf quantity and leaf area data, root development image data, and stem and leaf color characteristic data. Top-view and side-view images of pepper seedlings are collected periodically using a visible light imaging device. The canopy projection area and inter-leaf porosity are extracted from the top-view image, and the plant height and stem diameter data are extracted from the side-view image. Leaf reflectance spectral data are acquired using multispectral or hyperspectral imaging equipment. Based on the leaf reflectance spectral data, chlorophyll content, carotenoid content, and water content are retrieved, and stem and leaf color feature data are extracted. The root development image data is obtained by side-view or bottom-view imaging from a specific angle, or by using micro-root window technology, and the root development image data is processed to quantify the total root length, root surface area and number of root tips. The canopy projection area, inter-leaf porosity, plant height and stem diameter, chlorophyll content, water content, total root length, root surface area, and number of root tips are used as the core indicators of the growth phenotype data set.

3. The intelligent chili seedling cultivation method based on environmental perception and precise management according to claim 2, characterized in that, The method employs correlation and regression analysis to screen key environmental factors that significantly influence specific phenotypic indicators and their lag times from the corresponding dataset, including: Calculate the correlation coefficients between the time series of each environmental factor and the time series of each phenotypic index at different time offsets, and form a correlation coefficient matrix; From the correlation coefficient matrix, identify the environmental factor-phenotypic index pairs whose absolute correlation coefficients exceed the significance threshold, and record their corresponding lag time. For each selected environmental factor-phenotypic index pair, a regression equation is established with the environmental factor as the independent variable and the phenotypic index as the dependent variable. The goodness of fit and significance of the regression equation are then evaluated. The regression equation with high goodness of fit and statistical significance, and its corresponding lag time, are taken as the core components of the environment-phenotype dynamic response model.

4. The intelligent chili seedling cultivation method based on environmental perception and precise management according to claim 1, characterized in that, The method further includes: After irrigation or fertilization is performed, the concentration data of nutrient solution components and the rate of change of substrate moisture content in the set of seedling environmental factors are monitored in real time. The monitored rate of change is compared with the expected absorption or diffusion model to determine whether the nutrient or water supply is being used effectively or whether there is a risk of leaching. If it is determined that there is a risk of leaching or that the utilization efficiency is significantly lower than expected, the contents of the multi-dimensional environmental control instruction set that have not yet been executed, concerning irrigation volume, nutrient solution concentration, or fertilization frequency, will be dynamically adjusted.

5. The intelligent chili seedling cultivation method based on environmental perception and precise management according to claim 1, characterized in that, The method further includes: Collect a complete data chain of multiple batches and varieties of pepper seedlings throughout the entire process. The complete data chain includes a complete time series of the seedling environmental factors, execution records of the multi-dimensional environmental regulation command set, and the final growth phenotype data set. The complete data chain is used to iteratively train and optimize the environment-phenotype dynamic response model. The optimized environment-phenotype dynamic response model is applied to the initial stage of a new batch or new variety of pepper seedlings as the basis for generating the initial set of multi-dimensional environmental control instructions.

6. The intelligent chili seedling cultivation method based on environmental perception and precise management according to claim 1, characterized in that, The method further includes: Based on the root development image data and the temperature and humidity data of the soil or substrate, a root microenvironment status assessment sub-model is established. The root microenvironment status assessment sub-model is used to evaluate rhizosphere oxygen content, water uniformity, and potential disease risk. The evaluation results of the root microenvironment status assessment sub-model are used as a special environmental factor and incorporated into the environmental status assessment matrix for comprehensive analysis, triggering targeted root environment regulation instructions, including oxygen-enhancing irrigation, local water regulation, or application of biological agents.

7. The intelligent chili seedling cultivation method based on environmental perception and precise management according to claim 6, characterized in that, The sub-model for assessing the root microenvironment state, based on the root development image data and the temperature and humidity data of the soil or substrate, includes: The spatial distribution density and depth of the root system in the matrix were analyzed from the root development image data. By combining the distribution of temperature and humidity data of the soil or substrate at the corresponding depth layers, the actual moisture and temperature conditions in different root zones are assessed. Based on the aerobic characteristics of chili roots and the oxygen demand patterns at different growth stages, as well as the influence curves of water and temperature on root respiration and absorption, the oxygen diffusion rate and root activity level in the rhizosphere region were estimated. By combining the spatial distribution density, actual moisture and temperature conditions, oxygen diffusion rate, and root activity level, a comprehensive index characterizing the health of the root microenvironment is generated, which serves as the core output of the root microenvironment status assessment sub-model.

Images