Greenhouse environment regulation method and system for plant cultivation

Abstract

The present application relates to the technical field of agricultural facility control, and discloses a greenhouse environment regulation method and system for plant cultivation. The method initializes an environment parameter benchmark based on a plant growth model, the plant growth model being predefined according to plant types and growth cycles; multi-source environment sensor data inside the greenhouse is collected in real time; the multi-source environment sensor data is subjected to time series smoothing processing to obtain smoothed environment data sequences; a deviation degree of the smoothed environment data sequences from the environment parameter benchmark is calculated, the deviation degree being a weighted sum of parameter deviations; a potential environment anomaly is identified when the deviation degree exceeds an adaptive threshold; the duration and change trend of the potential environment anomaly are analyzed to confirm a real environment anomaly event; the occurrence frequency and influence range of the real environment anomaly event are counted to calculate an environment risk index; and the environment monitoring frequency and control strategy are dynamically adjusted according to the environment risk index.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of agricultural facility control, in particular to a greenhouse environment regulation method and system for plant cultivation. BACKGROUND

[0002] Current greenhouse environment regulation mainly adopts fixed threshold alarm and manual intervention. The existing technology relies on single sensor independent detection for environmental parameter monitoring, lacks multi-parameter collaborative analysis capability. The benchmark setting method is static and fixed, which cannot meet the specific needs of different plant species and growth stages. The data processing process is simple, and sensor noise and transient fluctuation interference are not effectively filtered out. The abnormal judgment mechanism is rigid, usually based on fixed threshold binary judgment, which cannot identify gradual environmental deterioration. The response strategy lacks adaptability, and the monitoring frequency and control intensity cannot be dynamically adjusted according to the risk level. The existing method needs to solve the key technical problems of multi-source data fusion, dynamic benchmark setting, intelligent abnormal identification and adaptive regulation.

[0003] Traditional greenhouse environment regulation system has obvious shortcomings in intelligence level and precise control. The environmental parameter benchmark setting is empirical, and no direct correlation with the plant growth model is established. The data acquisition frequency is fixed, which cannot balance the monitoring accuracy and system energy consumption. The smoothing filter algorithm is simple, which is difficult to effectively separate the real environmental change and the measurement noise. The deviation calculation weight is fixed, which does not consider the different influences of different environmental factors on plant growth. The adaptive threshold setting lacks theoretical guidance, and the false alarm and missed alarm risk is high. The abnormal confirmation method is subjective, and the duration threshold and trend judgment standard are set randomly. The risk index calculation dimension is single, which cannot comprehensively evaluate the spatio-temporal distribution characteristics of the abnormality. The regulation strategy generation is linear, which lacks multi-objective optimization mechanism. The existing technology needs to establish a whole-process intelligent regulation scheme from data acquisition to strategy generation. SUMMARY

[0004] The purpose of the present application is to provide a greenhouse environment regulation method and system for plant cultivation to solve the problems raised in the background.

[0005] To achieve the above purpose, the present application provides a greenhouse environment regulation method for plant cultivation, which comprises:

[0006] initializing the environmental parameter benchmark based on the plant growth model, the plant growth model being predefined according to the plant type and growth cycle;

[0007] real-time acquisition of multi-source environmental sensor data inside the greenhouse, the multi-source environmental sensor data including temperature, humidity, carbon dioxide concentration and light intensity;

[0008] time series smoothing processing of the multi-source environmental sensor data to obtain a smoothed environmental data sequence;

[0009] calculating a deviation degree of the smoothed environment data sequence from the environment parameter benchmark, the deviation degree being a weighted sum of deviations of respective parameters;

[0010] identifying a potential environment anomaly when the deviation degree exceeds an adaptive threshold;

[0011] analyzing a duration and a change trend of the potential environment anomaly to confirm a real environment anomaly event;

[0012] counting a frequency of occurrence and an influence range of the real environment anomaly event to calculate an environment risk index;

[0013] dynamically adjusting an environment monitoring frequency and a control strategy according to the environment risk index.

[0014] Preferably, when initializing the environment parameter benchmark based on the plant growth model, the method comprises:

[0015] obtaining plant growth stage information, the plant growth stage information being extracted from a planting plan;

[0016] querying a standard environment parameter library according to the plant growth stage information to obtain target parameter values of respective stages;

[0017] correcting the target parameter values in combination with greenhouse structure parameters and historical climate data to generate the environment parameter benchmark;

[0018] wherein the greenhouse structure parameters include a greenhouse area and a height, and the historical climate data includes seasonal average temperature and humidity.

[0019] Preferably, when performing time series smoothing processing on the multi-source environment sensor data, the method comprises:

[0020] filtering original sensor data by using a sliding window algorithm, a size of the sliding window being dynamically adjusted according to the environment risk index, and a moving average value of data in the window being calculated as a smoothed environment data sequence;

[0021] when the environment risk index increases, reducing the size of the sliding window to improve response speed; and when the environment risk index decreases, increasing the size of the sliding window to improve stability.

[0022] Preferably, when calculating the deviation degree of the smoothed environment data sequence from the environment parameter benchmark, the method comprises:

[0023] assigning a weight coefficient to each environment parameter, the weight coefficient being determined based on an influence degree of the parameter on plant growth;

[0024] calculating an absolute deviation of a current value of each parameter from a corresponding value in the environment parameter benchmark, summing the absolute deviations multiplied by the weight coefficients to obtain the deviation degree;

[0025] The adaptive threshold is dynamically calculated according to historical deviation data and plant growth stages.

[0026] Preferably, the analysis of the duration and change trend of the potential environmental anomaly comprises:

[0027] Extracting data points in the time period of the potential environmental anomaly, calculating the duration of the anomaly, fitting the change trend of the anomaly data by using a linear regression method to obtain a trend slope;

[0028] When the duration of the anomaly exceeds a preset duration threshold and the absolute value of the trend slope is greater than a trend threshold, the real environmental anomaly event is confirmed.

[0029] Preferably, the statistics of the occurrence frequency and the impact range of the real environmental anomaly event comprises:

[0030] Recording the type and occurrence time of each real environmental anomaly event, calculating the number of anomaly events in a unit time as the occurrence frequency, and evaluating the parameter deviation range caused by the anomaly event as the impact range;

[0031] The occurrence frequency and the impact range are normalized and then combined with a weight to obtain an environmental risk index.

[0032] Preferably, the environmental monitoring frequency is dynamically adjusted according to the environmental risk index, comprising:

[0033] Setting a basic monitoring frequency, the basic monitoring frequency being a fixed value;

[0034] Calculating a frequency adjustment factor according to the environmental risk index, the frequency adjustment factor being positively correlated with the environmental risk index, multiplying the basic monitoring frequency by the frequency adjustment factor to obtain an adjusted environmental monitoring frequency;

[0035] Controlling the sensor data collection interval according to the adjusted environmental monitoring frequency.

[0036] Preferably, the control strategy is dynamically adjusted according to the environmental risk index, comprising:

[0037] When the environmental risk index is lower than a low risk threshold, a conservative control strategy is adopted, and only major deviations are adjusted;

[0038] When the environmental risk index is between the low risk threshold and a high risk threshold, an active control strategy is adopted, and the environmental parameters are real-time fine-tuned;

[0039] When the environmental risk index is higher than the high risk threshold, an aggressive control strategy is adopted, and all standby environmental devices are immediately started.

[0040] Preferably, the method further comprises the following steps:

[0041] According to the environmental risk index, future environmental trends are predicted, preventive regulation instructions are generated, the preventive regulation instructions are fused with real-time control strategies, comprehensive control signals are output, and all data in the regulation process are recorded for updating the plant growth model and the environmental parameter benchmark.

[0042] Preferably, the present application further comprises a greenhouse environment regulation system for plant cultivation, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor implements the steps of the above-mentioned greenhouse environment regulation method for plant cultivation when executing the computer program.

[0043] Compared with the prior art, the present application has the following beneficial effects:

[0044] The deviation degree of the smoothed environmental data sequence and the environmental parameter benchmark is calculated, and the deviation degree is a weighted sum of parameter deviations. The smoothed environmental data sequence is generated by moving average or exponential smoothing method to eliminate random fluctuations and measurement noise. The environmental parameter benchmark is dynamically set according to the plant growth model, reflecting the optimal environmental conditions at different growth stages. The parameter deviation calculation uses the absolute value or square difference method to quantify the difference between the actual value and the benchmark value. The weighted sum calculation considers the relative importance of temperature, humidity, carbon dioxide concentration and light intensity on plant growth. The weight coefficient is obtained by expert knowledge or historical data training, reflecting the influence degree of different environmental factors. The deviation degree value is normalized for cross-parameter comparison and comprehensive evaluation. When the current deviation degree exceeds the adaptive threshold, potential environmental abnormalities are identified. The adaptive threshold is dynamically adjusted according to the environmental stability requirement and plant tolerance. The threshold updating mechanism considers factors such as seasonal changes, weather conditions and plant growth status. The potential abnormality identification adopts a soft decision method, providing risk warning rather than absolute alarm. The identification process is carried out in real time to ensure the timeliness of abnormality discovery. Through deviation degree calculation and adaptive threshold judgment, multi-parameter comprehensive evaluation of environmental abnormalities is realized. The weighted sum integrates multi-dimensional information to avoid single parameter misjudgment; the adaptive threshold improves the judgment flexibility and reduces the false alarm probability.

[0045] The occurrence frequency and influence range of statistical real environment abnormal events are calculated to calculate the environment risk index. The occurrence frequency statistics adopts a sliding time window method to record the number of abnormal events in a unit time. The influence range evaluation determines the geographical distribution area of abnormal influence through spatial interpolation analysis. The risk index calculation integrates the frequency and range dimensions, using weighted geometric mean or fuzzy reasoning method. The index value quantifies the potential threat degree of environmental abnormalities to plant growth. The risk assessment considers the dynamic characteristics such as abnormal duration, intensity change and recovery speed. The index is graded into multiple risk levels, corresponding to different response strategies. According to the environmental risk index, the environmental monitoring frequency and control strategy are dynamically adjusted. The monitoring frequency adjustment adopts an adaptive sampling algorithm, which increases the sampling rate in high-risk situations and reduces the sampling rate in low-risk situations to save energy. The control strategy optimization includes the coordinated configuration of actuator response speed, adjustment amplitude and action time. High-risk situations adopt proactive preventive control, and low-risk situations adopt responsive adjustment. The strategy adjustment process is smooth transition to avoid system oscillation. The dynamic adjustment mechanism realizes resource optimization configuration and improves system energy efficiency. Through the coordinated action of abnormality statistics, risk quantification and dynamic adjustment, intelligent control of greenhouse environment is realized. Frequency statistics identify abnormal rules, influence range evaluation determines severity, risk index quantifies comprehensive threat, and dynamic adjustment realizes precise response. This closed-loop control method significantly improves the fine and intelligent level of greenhouse environment management. BRIEF DESCRIPTION OF DRAWINGS

[0046] Figure 1 A working principle diagram of the greenhouse environment regulation method for plant cultivation described in the present application;

[0047] Figure 2 A flowchart for initializing environment parameter benchmarks based on plant growth models;

[0048] Figure 3 A flowchart for calculating the deviation of smooth environment data sequence from environment parameter benchmarks;

[0049] Figure 4 An analysis diagram of the occurrence frequency and influence range of greenhouse environment abnormal events;

[0050] Figure 5 A comparison analysis diagram of predicted and actual values of greenhouse environment risk index. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0052] Referring to Figure 1 The present application provides a greenhouse environment regulation method and system for plant cultivation, the method comprising: initializing environment parameter benchmarks based on a plant growth model, which is predefined according to plant types and growth periods, ensuring that the benchmark parameters meet the needs of each stage of the plant. Real-time collection of multi-source environmental sensor data inside the greenhouse, including key parameters such as temperature, humidity, carbon dioxide concentration, and light intensity, is transmitted to the processing unit at fixed intervals through a sensor network. Time series smoothing processing is performed on the collected multi-source environmental sensor data, and a sliding window algorithm is used to filter out noise, generating a smooth environmental data sequence to improve data reliability and stability. Calculate the deviation of the smooth environmental data sequence from the environment parameter benchmarks, and the deviation is defined as the weighted sum of the parameter deviations, where the weight coefficients are dynamically assigned based on the impact of the parameters on plant growth. When the deviation exceeds the adaptive threshold, the system identifies potential environmental abnormalities, and the threshold is adaptively calculated based on historical data and plant growth stages to avoid false positives. Analyze the duration and trend of potential environmental abnormalities, and confirm real environmental abnormal events through statistical methods to ensure that only persistent and trend obvious abnormalities are processed. Calculate the environmental risk index by counting the frequency and impact range of real environmental abnormal events, which reflects the environmental stability comprehensively. Dynamically adjust the environmental monitoring frequency and control strategy according to the environmental risk index, increase the monitoring frequency and adopt active control when the risk is high, and vice versa to save resources.

[0053] Embodiment 1: Referring to Figure 2 In specific implementation, obtaining plant growth stage information is the initial step, and the plant growth stage information is extracted from the pre-prepared planting plan, which is structured data stored in the system database in the form of digital files, including plant variety identifier, sowing date, expected germination period, vegetative growth period, flowering period and maturity period. The system determines the specific growth stage of the plant by analyzing the planting plan file and combining real-time clock data, such as determining whether the plant is in the seedling stage or the rapid growth stage, which is achieved through a date comparison algorithm to ensure the accuracy of the growth stage division. Query the standard environment parameter library according to the plant growth stage information, which is a pre-set knowledge base storing the ideal environmental parameter ranges of various typical plants in different growth stages verified by agricultural experts. The query operation is performed through database query language, and the input parameters are plant species code and growth stage code, and the output result is the target parameter value set of the corresponding stage, including but not limited to temperature target value, humidity target value, carbon dioxide concentration target value and light intensity target value. Each parameter is usually represented in the form of target range or optimal value, such as daytime suitable temperature range and nighttime minimum humidity requirement. The standard environment parameter library supports version management, allowing updates and maintenance based on the latest research findings.

[0054] It can be understood that the target parameter values obtained by querying need to be corrected locally to adapt to the unique microclimate conditions of the specific greenhouse. Correcting the target parameter values in combination with the greenhouse structure parameters and historical climate data is a key link. The greenhouse structure parameters are static data describing the physical characteristics of the greenhouse, including the greenhouse area, the internal vertical height, the building orientation, the covering material type and the light transmittance, the ventilation system configuration, etc., which directly affect the heat accumulation, air flow and light distribution inside the greenhouse. The historical climate data are long-term statistical data obtained from local meteorological stations or greenhouse historical records, including seasonal average temperature, average humidity, typical sunshine hours, frequency and intensity of extreme weather events, etc. The correction process adopts a weighted adjustment algorithm, which calculates a correction coefficient for each target parameter value according to the greenhouse structure parameters and historical climate data. For example, for the temperature target value, it will be adjusted up or down according to the heat preservation performance of the greenhouse and the local seasonal average temperature, to generate environment parameter benchmarks that are more consistent with the actual application scenario.

[0055] In specific implementation, the process of generating environment parameter benchmarks is systematic, and the corrected target parameter values are organized into a time series data structure, with each data point in the sequence corresponding to a specific day or key growth milestone in the plant growth cycle. The environment parameter benchmark file is saved in the system configuration file format, containing parameter name, benchmark value, valid time range and data source identification, etc. The system automatically triggers the environment parameter benchmark initialization process when starting or detecting a planting plan change event, and the initialization process includes an integrity check step to check whether all necessary parameters have been correctly assigned, to avoid empty values or out-of-range values. After the initialization of the environment parameter benchmark is completed, the system will record the initialization log, including the benchmark version number, the generation timestamp and the planting plan identifier it is based on, for auditing and tracing. Optionally, during the initialization of the environment parameter benchmark, the system can integrate real-time sensor calibration data to further improve the accuracy of the benchmark. The multi-source environmental sensors deployed in the greenhouse environment will perform a self-check and calibration process after the system starts, and the calibration data includes the zero point offset of the sensor, the sensitivity coefficient, etc. The system will use the calibration data as an input factor to participate in the correction calculation of the target parameter value, for example, if there is a known measurement positive deviation of the light intensity sensor, the light intensity benchmark will be set with corresponding compensation, thereby reducing the system error, so that the environment parameter benchmark has higher consistency with the real physical measurement value. This integrated calibration mechanism makes the environment parameter benchmark not only rely on theoretical models and historical data, but also closely combined with the characteristics of the actual measurement system.

[0056] In some embodiments, the initialization process of the environmental parameter baseline is designed to be repeatable and dynamically fine-tuned. When the system has been running for a period of time, if updated planting plan instructions are received or significant changes in external climate patterns are detected, the system can restart the initialization process to generate a new version of the environmental parameter baseline. The dynamic fine-tuning function allows the system to make small adjustments to individual parameters in the environmental parameter baseline based on feedback on the effects of short-term environmental regulation without stopping. The fine-tuning logic is based on pre-set rules, for example, if the actual growth rate of the plant during a certain growth period deviates from the expected model for multiple consecutive times, the light or temperature baseline value for that stage is automatically fine-tuned. It can be understood that the storage and access method of the environmental parameter baseline has a direct impact on the performance of the system. The generated environmental parameter baseline is usually loaded into the system's high-speed memory cache to reduce data access delay during subsequent real-time comparison calculations. The baseline data is organized in key-value pairs or array structures in the cache, with the key being the time index or growth stage code and the value being the corresponding parameter set. The system sets an access interface for the environmental parameter baseline, and other processing modules obtain the environmental parameter baseline value at the current time or for a specified growth stage by calling the interface. The interface design ensures data consistency and thread safety. At the same time, a complete copy of the baseline is stored persistently in non-volatile memory to prevent data loss due to system power failure or restart.

[0057] In specific implementations, there is a clear data dependency relationship between the environmental parameter baseline initialization module and other modules of the system. The initialization module relies on the planting plan management module to provide plant growth stage information, relies on the basic information library module to provide greenhouse structure parameters and historical climate data, and interacts with the sensor management module to obtain calibration data. Successful execution of the initialization process is a prerequisite for subsequent environmental monitoring, deviation calculation, and control strategy application. The system status monitor continuously checks the status of the environmental parameter baseline, and if it finds that the baseline data is invalid or expired, it triggers an alarm and attempts to reinitialize, thereby ensuring the reliability of the entire environmental regulation chain. The entire initialization process emphasizes automation and data-driven, minimizing the need for human intervention and improving the efficiency and accuracy of greenhouse environmental intelligent regulation.

[0058] Optionally, for large-scale or zoned management of greenhouse groups, the environment parameter baseline initialization supports zoned differentiated configuration. The system can independently initialize environment parameter baselines for different areas of a greenhouse, and each area can be associated with different planting plans or have unique structure parameters. Zoned initialization allows fine and differentiated environmental management for different crops or different growth stages of the same crop in the same greenhouse. The system maintains a zoned mapping table to associate physical sensor nodes with logical zoned identifiers, and obtains corresponding planting plan information and structure parameters according to the zoned identifiers during initialization to generate a set of parallel and independent environment parameter baseline sets. This zoned management capability significantly improves the flexibility and pertinence of environmental regulation in complex greenhouse scenarios.

[0059] Embodiment 2: refer to Figure 3 In specific implementation, when performing time series smoothing processing on multi-source environmental sensor data, the system uses a sliding window algorithm to filter the original sensor data, and the size of the sliding window is dynamically adjusted according to the real-time calculation of the environmental risk index, which is a numerical indicator reflecting the stability of the greenhouse environment. When the environmental risk index rises, it indicates that the environmental fluctuation intensifies, and the system will reduce the size of the sliding window to improve the responsiveness of the data, for example, reducing the window size from the default of 15 data points to 8 data points. Conversely, when the environmental risk index decreases, it indicates that the environment tends to be stable, and the system will increase the size of the sliding window to 25 data points or more to enhance the smoothing effect of the data. The sliding window algorithm manages the data stream in the form of a queue, and the oldest data point is removed when a new data point enters the window, maintaining the freshness of the data in the window. The moving average of the data in the window is calculated as the smoothed environmental data sequence, and the moving average is calculated by the arithmetic mean method, that is, the sum of the values of all data points in the window is divided by the window size. For some parameters such as light intensity, if the data distribution is skewed, a weighted moving average method is used to give higher weights to recent data points. The smoothed environmental data sequence is stored in a circular buffer, and the buffer size matches the maximum possible sliding window size to ensure data continuity is not interrupted. During the smoothing process, the system integrates an outlier detection mechanism, which uses the Z-score method based on standard deviation to identify and eliminate obvious outliers. The Z-score threshold is set to 3, that is, data points more than three times the standard deviation of the mean are considered noise and are filtered out to prevent outliers from distorting the smoothing results.

[0060] In some embodiments, the dynamic adjustment logic of the sliding window size is implemented by a mapping function, which takes the environmental risk index as input and outputs the recommended window size value. The mapping function uses piecewise linear interpolation method, and predefines several key points of environmental risk index and their corresponding window sizes, for example, when the environmental risk index is 0, the window size is 30, when the environmental risk index is 0.5, the window size is 15, and when the environmental risk index is 1, the window size is 5. The system calculates the actual window size by linear interpolation according to the current environmental risk index value, ensuring that the adjustment process is smooth and has no jump. The window size adjustment event triggers data reprocessing, the system clears the current window and accumulates data points based on the new window size for smooth calculation, and records the adjustment log for audit. It can be understood that the quality of the smoothed environmental data sequence directly affects the accuracy of the subsequent deviation calculation, therefore the system monitors the stability indicators of the smoothing process, such as the variance of the moving average, and triggers calibration check if the variance is continuously too high.

[0061] When calculating the deviation of the smoothed environmental data sequence from the environmental parameter benchmark, the system assigns a weight coefficient to each environmental parameter, which is determined based on the degree of influence of the parameter on plant growth. The influence degree data comes from plant physiology models, for example, during the active period of photosynthesis, the weight coefficient of light intensity is set to 0.4, the weight coefficient of temperature is set to 0.3, the weight coefficient of humidity is set to 0.2, and the weight coefficient of carbon dioxide concentration is set to 0.1. The weight coefficients are stored in a configurable weight table and are dynamically updated according to the plant growth stage, and the update event is activated by the growth stage switch trigger. In order to ensure dimensional consistency, the relative deviation of the current smoothed value of each parameter from the corresponding value in the environmental parameter benchmark is calculated, which is the difference between the current smoothed value and the benchmark value divided by the benchmark value, and the absolute value is taken, thereby obtaining a dimensionless percentage value. The relative deviation of each parameter is multiplied by its weight coefficient and summed to obtain the comprehensive deviation, and the deviation calculation formula is:

[0062]

[0063] wherein: represents the deviation, represents the number of environmental parameters, represents the weight coefficient of the th parameter, represents the current smoothed value of the th parameter, represents the environmental parameter benchmark value of the th parameter. The deviation calculation process is executed periodically, and the calculation frequency is synchronized with the data acquisition frequency.

[0064] In some embodiments, the calculation of adaptive threshold relies on historical deviation data and current plant growth stage information, the historical deviation data refers to a sequence of deviation values calculated in the past, the length of time window is generally 24 hours or a complete growth cycle, the system uses moving statistical methods to analyze the historical sequence, for example, calculates the average and standard deviation of historical deviation, the adaptive threshold is set to the average value plus K times the standard deviation, the value of K is adjusted according to the plant growth stage, the value of K is set to 1.5 in the sensitive stage such as seedling stage to reduce the threshold and improve the sensitivity, the value of K is set to 2.5 in the mature stage to increase the threshold and reduce false positives. The plant growth stage information is obtained from the planting plan module, and the selection logic of K value is used to modulate the K value. The adaptive threshold update period is set to once an hour to ensure that the threshold can keep up with the rhythm of environmental changes. Optionally, the system also introduces a minimum threshold guarantee mechanism to prevent the threshold from being too low and causing excessive sensitivity, the minimum threshold is set based on a fixed percentage of deviation, for example, set the minimum adaptive threshold to 0.05 (i.e. 5% relative deviation benchmark).

[0065] It can be understood that the deviation calculation and adaptive threshold judgment are continuous pipeline processes, the calculated deviation is immediately compared with the current adaptive threshold, if the deviation exceeds the threshold, it is marked as a potential environmental anomaly, at the same time, the deviation value is appended to the historical deviation database for subsequent threshold update. The system maintains an independent relative deviation calculation pipeline for each environmental parameter, but the comprehensive deviation is the result of aggregating the relative deviations of all parameters, which enables the system to focus on the overall environmental state and trace individual parameter contributions. In specific implementation, the allocation of weight coefficients considers the coupling effect between parameters, for example, when temperature and high humidity occur at the same time, it may cause superimposed stress to plants, therefore, the system defines an interaction weight item in the weight table, but for the sake of simplifying the initial implementation, the interaction item is temporarily set to zero, which can be optimized by machine learning model in the future. The smoothed environmental data sequence and deviation value are both time stamped and recorded in the time series database, which supports subsequent trend analysis and backtracking query. Optionally, for resource-constrained embedded systems, fixed-point arithmetic operations can be used for smoothing and deviation calculation to improve efficiency while ensuring calculation accuracy to meet the needs of agricultural applications.

[0066] In a specific implementation, when analyzing the duration and trend of a potential environmental anomaly, the system starts from the time instance marked as a potential environmental anomaly, extracts all data points within the time period of the potential environmental anomaly, the exact start point of the time period is the sampling timestamp when the deviation first exceeds the adaptive threshold, and the end point of the time period is the first sampling timestamp after the deviation continues to fall below the adaptive threshold and remains stable. The system extracts this sequence of data points in chronological order from a ring buffer dedicated to storing the latest smooth environmental data sequence and the corresponding deviation calculation results. The duration of the anomaly is directly obtained by comparing the difference between the end timestamp and the start timestamp of the time period. The unit of the time difference can be seconds, minutes, or hours according to system configuration. For example, a potential temperature anomaly that starts at 10:05:00 and ends at 10:15:30 has an anomaly duration of 10 minutes and 30 seconds. The system records the anomaly duration as a key attribute in the temporary event object.

[0067] The linear regression method is used to fit the trend of the abnormal data. The system takes the extracted data point sequence as input, with the X-axis representing the elapsed time relative to the start of the anomaly and the Y-axis representing the corresponding deviation value or specific value of the anomaly parameter. The linear regression algorithm calculates the best fitting straight line for these data points, using the least squares method. The final output is the trend slope, which is the slope value of the fitted straight line. The trend slope is a signed number, with a positive trend slope indicating an upward trend in deviation or parameter value during the anomaly duration, and a negative trend slope indicating a downward trend. The system also calculates the goodness-of-fit index of linear regression, such as the R-square value, to assess the reliability of the trend analysis, but the R-square value is not directly used in the decision logic. When the anomaly duration exceeds the pre-set duration threshold and the absolute value of the trend slope is greater than the trend threshold, the system confirms the potential environmental anomaly event as a real environmental anomaly event. The pre-set duration threshold is a fixed value set for different environmental parameter types, for example, the pre-set duration threshold for temperature parameters may be set to 5 minutes, while the pre-set duration threshold for humidity parameters may be set to 10 minutes. The trend threshold is also parameter-dependent and is an empirical constant used to filter out fluctuations that are long-lasting but weak in change.

[0068] In a specific implementation, after confirming a real environment abnormal event, the system creates a structured real environment abnormal event record, which contains an event unique identifier, an abnormal parameter type, an event start timestamp, an event end timestamp, an abnormal duration, a trend slope value, a maximum deviation value, and a specific value sequence of the parameter that triggered the event. The record is stored persistently in a special event log database, which is indexed to facilitate subsequent quick queries and analysis by time and type. The event confirmation logic is managed by a state machine, which monitors the entire state transition process from potential abnormality to confirmed abnormality, ensuring that the life cycle of each event is completely tracked. When counting the occurrence frequency and impact range of real environment abnormal events, the system periodically scans the event log database, counts the number of abnormal events in a unit of time as the occurrence frequency, and the length of the unit of time window is configurable, for example, it can be set to 1 hour, 4 hours, or 24 hours. The system slides the time window to count the number of real environment abnormal events that end within the window, for example, it counts how many real temperature abnormal events occurred in the past hour, and the occurrence frequency is an integer value. The parameter deviation range caused by the abnormal event is evaluated as the impact range, and the impact range is calculated for each real environment abnormal event by finding the maximum absolute deviation percentage of the actual value of the abnormal parameter from the environmental parameter baseline value during the duration of the event. For example, in a temperature abnormal event, the temperature reaches 30°C, while the baseline value is 25°C, the maximum absolute deviation is 5°C, and the deviation percentage is (5 / 25)*100%=20%, which is the impact range value of the event. The system records the impact range for each event.

[0069] The occurrence frequency and impact range are normalized and combined to obtain the environmental risk index. Normalization is a process of mapping the original values of occurrence frequency and impact range to a scale of 0 to 1. The normalization of occurrence frequency uses a linear normalization method, which is divided by the theoretical maximum number of events in a statistical period. The theoretical maximum value is based on historical extreme cases. The normalization of impact range is also linearly mapping the actual percentage to the 0-1 interval, where 0% is mapped to 0 and 100% or a set maximum reasonable deviation percentage is mapped to 1. The normalized occurrence frequency is denoted as F_norm, and the normalized impact range is denoted as I_norm. The environmental risk index ERI is calculated by the following formula:

[0070]

[0071] Wherein: ERI represents the environmental risk index, which is a value between 0 and 1; F_norm is the normalized occurrence frequency; I_norm is the normalized impact range; and are weight coefficients, satisfying The values of weight coefficients a and b can be adjusted according to plant species and growth stages, for example, during sensitive growth periods, a higher weight b can be assigned to the impact range I norm to pay more attention to the severity of abnormal events. The environmental risk index is recalculated regularly, and the update cycle is aligned with the statistical unit time window.

[0072] In some embodiments, the statistics of occurrence frequency can be further refined to distinguish different types of abnormal events, for example, temperature anomalies and humidity anomalies are counted separately, and then a weighted sum is calculated as the comprehensive occurrence frequency, and the weights of different types of anomalies can be different. It can be understood that the calculation of the environmental risk index is to quantify the overall instability of the greenhouse environment, and the higher the index value represents the greater the risk faced by the environment control, which requires more active intervention. The system records the calculated environmental risk index together with the current timestamp and publishes it to the message bus of the system for subscription by the environment monitoring frequency adjustment module and the control strategy adjustment module. Optionally, for the calculation of the impact range, in addition to using the maximum absolute deviation percentage, the system can also use the average deviation percentage or the integral value of the deviation area as an alternative indicator to more comprehensively reflect the overall impact strength of abnormal events. The design of the event log database supports advanced queries, for example, the frequency distribution of a specific type of anomaly in a specific growth stage can be queried, and these aggregated data can be used for long-term trend analysis and model optimization. The historical data of the environmental risk index form a time series, and the system can perform simple trend analysis on this time series, such as calculating the moving average, to determine whether the environmental risk is increasing or decreasing, providing input for predictive regulation.

[0073] In some embodiments, the statistical process needs to consider the uniqueness of the event to avoid repeated counting, for example, if a long-duration abnormal event spans two statistical time windows, the system is configured to attribute it entirely to the time window at the end of the event, ensuring statistical accuracy. The calculation module of the environmental risk index has a fault-tolerant mechanism, if there is no real environmental abnormal event occurring within the statistical time window, the occurrence frequency is 0, if the impact range of all events is small, may be close to 0, at this time the environmental risk index ERI will be low, reflecting a good environment state.

[0074] Referring to Figure 4, which is a key data visualization result of the greenhouse environment regulation method and system for plant cultivation, focusing on the statistical analysis of the occurrence frequency and impact range of real environment abnormal events. In the figure, temperature abnormality, humidity abnormality, CO2 concentration abnormality, light abnormality, and comprehensive abnormality are used as classification dimensions. The occurrence frequency of each type of abnormality is presented through blue column charts, reflecting the frequency of different environmental factor abnormalities in a unit of time. The impact range is presented through red line charts and node annotations, quantifying the degree to which each type of abnormality causes the environment parameters to deviate from the benchmark. This figure provides direct data support for the calculation of the environmental risk index. By intuitively comparing the occurrence patterns and impact strengths of different types of abnormalities, the system can accurately judge the environmental risk level, thereby providing a basis for dynamically adjusting the environmental monitoring frequency and control strategy, solving the technical pain points of single abnormality statistical dimension and lack of intuitive evaluation in traditional greenhouse environment regulation, and promoting the intelligent transformation of greenhouse environment regulation from experience to data-driven. This lays a data visualization foundation for realizing multi-parameter coordinated greenhouse environment intelligent management.

[0075] In a specific implementation, when dynamically adjusting the environmental monitoring frequency according to the environmental risk index, the system sets a basic monitoring frequency, which is a fixed value, for example, collecting sensor data every 300 seconds. This value is a conservative value pre-set based on hardware capability, energy consumption considerations, and general monitoring needs, and is stored in the system configuration file. A frequency adjustment factor is calculated based on the environmental risk index. The frequency adjustment factor is positively related to the environmental risk index and is calculated using a linear function with the function expression:

[0076]

[0077] wherein: represents the frequency adjustment factor, is a proportional constant greater than zero, and its value determines the influence of the environmental risk index on the adjustment factor, represents the environmental risk index, and its value is between 0 and 1. When the environmental risk index E is 0, the frequency adjustment factor is 1, indicating no adjustment is needed; when the environmental risk index E is 1, the frequency adjustment factor reaches the maximum value of 1+k. Multiplying the basic monitoring frequency by the frequency adjustment factor gives the adjusted environmental monitoring frequency. For example, the basic monitoring frequency is 300 seconds, the current environmental risk index is 0.6, and the proportional constant k is set to 1. The frequency adjustment factor is 1.6, and the adjusted environmental monitoring frequency is seconds, meaning the sensor data collection interval is shortened to approximately 188 seconds. The sensor data collection interval is controlled according to the adjusted environmental monitoring frequency, which is implemented by a programmable timer whose period is dynamically reloaded according to the new frequency value, triggering the data collection interrupt, thus precisely controlling the time interval of sensor readings.

[0078] In some embodiments, the value of the proportionality constant k can be configured according to the greenhouse operation strategy, for example, in the fine management mode, k can be set to 1.5 to enhance responsiveness, and in the energy saving priority mode, k can be set to 0.8 to reduce energy consumption. The calculation period of the frequency adjustment factor is consistent with the update period of the environmental risk index, ensuring that the monitoring frequency can respond to changes in environmental risk in a timely manner. The system records the history log of monitoring frequency adjustment, including adjustment time, frequency before adjustment, frequency after adjustment, and environmental risk index value triggering adjustment, for subsequent analysis and audit. In order to avoid the instability of the system caused by frequent switching of monitoring frequency, the system sets a minimum adjustment interval, for example, at least 5 minutes between two frequency adjustments, unless the environmental risk index jumps dramatically.

[0079] When the control strategy is dynamically adjusted according to the environmental risk index, the system presets two key thresholds: a low risk threshold and a high risk threshold, which divide the environmental risk index range into three intervals, corresponding to different levels of control strategy. The low risk threshold is usually set to a lower value, for example 0.3, and the high risk threshold is set to a higher value, for example 0.7. The system maintains a control strategy lookup table that maps the environmental risk index interval to a specific set of control strategy instructions. When the environmental risk index is below the low risk threshold, the system adopts a conservative control strategy. The characteristics of the conservative control strategy are high control action trigger threshold, long response delay, and moderate execution, only adjusting significant deviations beyond a large tolerance range, for example, only when the temperature deviation continues to exceed ±3°C and the duration exceeds 10 minutes, the heating or cooling equipment is started, and the equipment runs at a lower power, and the generation period of control instructions is also longer, for example, evaluating every 15 minutes whether action is needed.

[0080] When the environmental risk index is between the low risk threshold and the high risk threshold, the system adopts an active control strategy, the core of which is to fine-tune the environmental parameters in real time. The system checks the deviation of the environmental parameters from the reference values at a high frequency, and uses a proportional-integral-derivative controller to calculate accurate control amounts, such as fine-tuning the opening of the ventilation window or the power output percentage of the heater according to the current temperature deviation and its trend. The fine-tuning amplitude is proportional to the deviation size, aiming to smoothly maintain the environmental parameters around the target value. When the environmental risk index is higher than the high risk threshold, the system adopts an aggressive control strategy, which gives priority to quickly stabilizing the environment. All available backup environmental control devices are immediately started in parallel, such as high-power fans, auxiliary humidifiers, supplementary carbon dioxide systems, and sunshade nets. The control instructions have the highest priority, bypassing complex control algorithm calculations, and directly outputting maximum or preset emergency control signals, ignoring energy consumption and equipment wear and tear factors, until the environmental risk index falls below the high risk threshold.

[0081] It can be understood that the switching of the control strategy is managed by a finite state machine, and the states of the state machine include a "conservative control state", an "active control state", and an "aggressive control state". State migration is triggered by the comparison result of the environmental risk index and the threshold. In order to avoid frequent switching (chattering) of the strategy caused by slight fluctuations of the environmental risk index near the threshold, the system introduces a hysteresis mechanism, for example, the environmental risk index needs to exceed 0.3 to switch from the conservative state to the active state, but the environmental risk index needs to be lower than 0.25 to switch back from the active state to the conservative state. Referring to Table 1, the main features of the control strategy are shown:

[0082] Table 1: Control strategy feature comparison table

[0083]

[0084] Optionally, under the aggressive control strategy, the system can send a high-level alarm notification to the user, prompting manual attention or intervention. The control strategy module outputs arbitrated comprehensive control signals, which are digital or analog signals sent to specific actuators such as frequency converters, solenoid valves, and relays through industrial buses or IO modules, to drive physical devices such as fans, water pumps, and heaters to act. The system continuously monitors the environmental improvement effect after the execution of the control strategy, and uses the effect feedback to evaluate the effectiveness of the control strategy, but this evaluation result is not directly used for the calculation of the current environmental risk index. In some embodiments, different thresholds can be set for different environmental parameters, so as to realize fine-tuned control strategy adjustment by parameter classification, for example, temperature control may enter the active control state earlier than humidity control.

[0085] Example 5: In a specific implementation, when predicting future environmental trends based on the environmental risk index, the system adopts a time series analysis-based method, with the input data being a historical sequence of environmental risk index values, which are collected at fixed time intervals and stored in a ring buffer. For example, the environmental risk index values calculated every 15 minutes in the past 24 hours are stored, forming a time series containing 96 data points. The prediction model uses an exponential smoothing method, whose core formula is:

[0086]

[0087] wherein: represents the predicted value of the environmental risk index at time point t+1 based on the value at time point t; represents the actual calculated environmental risk index at time point t; represents the predicted value at time point t based on the value at time point t-1; is the smoothing coefficient, whose value is between 0 and 1, determining the weight of the latest observation in the prediction. The value of the smoothing coefficient λ can be adaptively adjusted based on historical prediction errors, with the initial value usually set to 0.2. After each calculation period, the system updates the prediction value using the latest obtained environmental risk index , which is considered as an estimate of the environmental risk trend in the next period.

[0088] The generation of preventive control instructions is a logical judgment process based on the predicted value . The system presets one or more preventive control thresholds, and when the predicted value exceeds a certain threshold, the corresponding preventive control instruction is generated. For example, if the preventive control threshold is set to 0.6, when the predicted value is greater than 0.6, the system concludes that there is a risk of environmental deterioration in the future, and generates an instruction to "prepare to start enhanced ventilation". The preventive control instruction contains information such as instruction type, target device, recommended action intensity, and expected execution time, and is placed in a preliminary instruction queue with a lower priority, waiting for arbitration.

[0089] ​The preventive regulation instruction is fused with the real-time control strategy through a priority arbitration module. The real-time control strategy is an immediate control instruction directly generated by the dynamic control strategy adjustment module described in embodiment 4 according to the current environmental risk index, and has a higher real-time priority. The principle followed by the arbitration module is that the real-time control instruction usually has a higher execution priority than the preventive regulation instruction. The fusion process is to check whether the preventive regulation instruction in the standby instruction queue conflicts with the real-time control instruction currently being executed or about to be executed in terms of the target or effect; if there is no conflict and the system control resources are redundant, the arbitration module will combine the preventive regulation instruction with the real-time control instruction and output a comprehensive control signal; if there is a conflict, the execution of the real-time control instruction is prioritized, and the preventive regulation instruction will be suspended or discarded. For example, the current real-time control strategy is executing an instruction to lower the temperature, and the preventive regulation instruction is to prepare to lower the humidity. These two instructions do not conflict and can be fused and executed; but if the preventive instruction is to prepare to raise the temperature, it conflicts with the real-time instruction and will be suppressed by the arbitration module.

[0090] The output comprehensive control signal is all control commands determined to be executed after arbitration, converted into a standard signal form that the actuator can recognize. The signal types include on-off signals, analog signals, or pulse width modulation signals. The comprehensive control signal is sent to the corresponding environmental control device driver through a specified communication protocol, thereby driving the actuator such as a variable frequency fan, a light supplement lamp, and an electromagnetic valve to act. The system records all data in the regulation process. The recorded data range includes raw sensor readings, smoothed environmental data sequences, calculated environmental risk indexes, generated prediction values , all generated preventive regulation instructions and real-time control strategy instructions, arbitration decision logs, and the history of the final output comprehensive control signal. These data are indexed by time stamp and persistently stored in the non-volatile memory of the system to form a complete regulation process data chain.

[0091] In some embodiments, the recorded data is used to update the plant growth model and the environmental parameter benchmark. The updating process is not performed in real time, but a batch task is started in a longer period. This task analyzes the recorded environmental data, control instructions, and growth state observation data of the plants during the past period of time, evaluates the accuracy of the parameter settings in the current plant growth model through regression analysis or machine learning algorithms, and fine-tunes the unreasonable parts. For example, if the data shows that during a certain growth stage, the plant growth indicators are still good despite frequent deviations of the temperature from the benchmark, the environmental parameter benchmark range for the temperature at that stage can be appropriately relaxed. Similarly, the target value in the environmental parameter benchmark can also be recalibrated according to the seasonal changes in historical climate data.

[0092] It can be understood that the addition of the predictive regulation function enables the system to have the ability of feedforward control, and no longer completely relies on feedback response to environmental deviation, which helps to reduce system overshoot and improve environmental stability. The whole process constitutes a complete intelligent control closed loop including perception, prediction, decision-making, execution and learning. Optionally, the smoothing coefficient λ can be designed not as a fixed value, but dynamically adjusted according to the recent rate of change of the environmental risk index. When the environmental risk index changes dramatically, increase the value of λ to make the prediction model more sensitive to recent changes; when the environmental risk index is stable, reduce the value of λ to exert the smoothing effect of historical data, so that the prediction is more robust. In some embodiments, the generation logic of the preventive regulation instruction can be more complex, not only based on the predicted value of the environmental risk index, but also combined with the independent prediction trend of the specific environmental parameter to make a comprehensive judgment, so as to generate more targeted preparatory actions.

[0093] Referring to Figure 5 The figure is the core data visualization result of the greenhouse environment regulation method and system for plant cultivation, focusing on the comparison analysis of the predicted and actual values of the environmental risk index. Based on the historical time series of the environmental risk index, the exponential smoothing method is used to generate the predicted risk index, and the dynamic comparison with the actual risk index is used to quantify the prediction accuracy by RMSE. The figure is a direct embodiment of the technology link of generating preventive regulation instructions based on the prediction of future environmental trends of the environmental risk index, solving the technical pain points of traditional greenhouse environment regulation lacking of foresight and relying only on real-time feedback. By intuitively presenting the deviation between the predicted value and the actual value, the preventive regulation instruction generation of the system is supported, and the greenhouse environment regulation is promoted from responsive to predictive, and from single feedback control to predictive-feedback integrated intelligent control. It is the core data carrier of the prediction link in the whole process of intelligent regulation closed loop from data acquisition to strategy generation, helping the system to predict the environmental trend in advance, generate accurate preventive regulation instructions, and thus improve the stability and regulation efficiency of the greenhouse environment, and provide more forward-looking environmental protection for plant cultivation.

[0094] It should be noted that, in this document, the terms such as first and second are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "contain" or any other variant thereof are intended to cover non-exclusive inclusion, so that the process, method, article or equipment including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or equipment.

[0095] While embodiments of the application have been shown and described, it is to be understood that the embodiments described are merely exemplary of the principles and application of the present application. Numerous modifications and adaptions can be effected without departing from the spirit and scope of the present application, which is not limited to the exact construction and arrangement described. It is intended, therefore, to cover all modifications and adaptions that fall within the scope of the claims and their equivalents.

Claims

1. A method for controlling the greenhouse environment for plant cultivation, characterized in that, Includes the following steps: The environmental parameter baseline is initialized based on a plant growth model, which is predefined according to plant type and growth cycle. Real-time acquisition of multi-source environmental sensor data inside the greenhouse, including temperature, humidity, carbon dioxide concentration, and light intensity; The multi-source environmental sensor data is subjected to time series smoothing to obtain a smoothed environmental data sequence; Calculate the deviation of the smoothed environmental data sequence from the environmental parameter benchmark, wherein the deviation is the weighted sum of the deviations of each parameter; When the deviation exceeds the adaptive threshold, a potential environmental anomaly is identified. Analyze the duration and trends of potential environmental anomalies to confirm actual environmental anomaly events; Statistically analyze the frequency and impact of real-world environmental anomalies, and calculate the environmental risk index; The frequency of environmental monitoring and control strategies are dynamically adjusted based on the environmental risk index. The analysis of the duration and trends of potential environmental anomalies includes: Data points within the time period of potential environmental anomalies are extracted, the duration of the anomaly is calculated, and the trend of the anomaly data is fitted using a linear regression method to obtain the trend slope. When the duration of the anomaly exceeds a preset duration threshold and the absolute value of the trend slope is greater than the trend threshold, it is confirmed as a real environmental anomaly event. When calculating the frequency and scope of impact of abnormal events in the real environment, the following are included: Record the type and occurrence time of each real-world environmental anomaly, calculate the number of anomalies per unit time as the occurrence frequency, assess the range of parameter deviations caused by the anomalies as the impact range, and then normalize and weight the occurrence frequency and impact range to obtain the environmental risk index.

2. The greenhouse environment control method for plant cultivation according to claim 1, characterized in that: The initialization of environmental parameter baselines based on plant growth models includes: Obtain plant growth stage information, which is extracted from the planting plan; By querying the standard environmental parameter database based on plant growth stage information, the target parameter values ​​for each stage can be obtained; By combining greenhouse structural parameters and historical climate data, the target parameter values ​​are corrected to generate environmental parameter benchmarks; Among them, greenhouse structural parameters include greenhouse area and height, and historical climate data include seasonal average temperature and humidity.

3. The greenhouse environment control method for plant cultivation according to claim 2, characterized in that: The time-series smoothing process for the multi-source environmental sensor data includes: The original sensor data is filtered using a sliding window algorithm. The size of the sliding window is dynamically adjusted according to the environmental risk index. The moving average of the data within the window is calculated as a smoothed environmental data sequence. When the environmental risk index increases, the sliding window size is reduced to improve response speed; when the environmental risk index decreases, the sliding window size is increased to improve stability.

4. The greenhouse environment control method for plant cultivation according to claim 3, characterized in that: The calculation of the deviation between the smoothed environmental data sequence and the environmental parameter benchmark includes: Each environmental parameter is assigned a weighting coefficient, which is determined based on the degree of influence of the parameter on plant growth. Calculate the absolute deviation between the current value of each parameter and the corresponding value in the environmental parameter baseline, multiply the absolute deviation by the weighting coefficient, and sum them to obtain the deviation degree; The adaptive threshold is dynamically calculated based on historical deviation data and plant growth stage.

5. The greenhouse environment control method for plant cultivation according to claim 4, characterized in that: When dynamically adjusting the environmental monitoring frequency based on the environmental risk index, the following is included: Set a basic monitoring frequency, which is a fixed value; The frequency adjustment factor is calculated based on the environmental risk index. The frequency adjustment factor is positively correlated with the environmental risk index. The adjusted environmental monitoring frequency is obtained by multiplying the basic monitoring frequency by the frequency adjustment factor. The sensor data acquisition interval is controlled according to the adjusted environmental monitoring frequency.

6. The greenhouse environment control method for plant cultivation according to claim 5, characterized in that: When dynamically adjusting the control strategy based on the environmental risk index, the following are included: When the environmental risk index is below the low-risk threshold, a conservative control strategy is adopted, and adjustments are made only for major deviations. When the environmental risk index is between the low-risk threshold and the high-risk threshold, an active control strategy is adopted to fine-tune environmental parameters in real time. When the environmental risk index exceeds the high-risk threshold, an aggressive control strategy is adopted, and all backup environmental equipment is immediately activated.

7. The greenhouse environment control method for plant cultivation according to claim 6, characterized in that: It also includes the following steps: Based on the environmental risk index, future environmental trends are predicted, and preventive control instructions are generated. These preventive control instructions are then integrated with real-time control strategies to output a comprehensive control signal. All data during the control process are recorded and used to update plant growth models and environmental parameter benchmarks.

8. A greenhouse environment control system for plant cultivation, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the greenhouse environment control method for plant cultivation as described in any one of claims 1 to 7.

Images