Method and device for cooperatively and intelligently regulating and controlling multi-environment data in greenhouse

Abstract

The invention relates to the field of intelligent control, and provides a multi-environment data collaborative intelligent regulation and control method and device in a greenhouse. The method comprises the following steps: acquiring predicted values of various environmental data at a future moment according to various environmental data influencing crop growth in a greenhouse at the current moment and meteorological data in a period of time in the future; and generating a first target control strategy of each controllable device in the greenhouse at the current moment according to the optimal values and the predicted values of the various environmental data required by the growth of the crops in the greenhouse at the current moment, and controlling each controllable device according to the first target control strategy. According to the application, the current situation that each factor in current greenhouse control operates independently and is lack of collaboration can be changed, collaborative, accurate and intelligent regulation and control of various environmental data (such as light, temperature, water, gas, fertilizer and the like) in the greenhouse can be realized, an unprecedented stable and balanced growth environment is provided for crops, and the operation energy consumption is remarkably reduced.

Description

Technical Field

[0001] This application relates to the field of intelligent control technology, specifically to a method and device for collaborative intelligent regulation of multiple environmental data in a greenhouse. Background Technology

[0002] Currently, greenhouses, as core facilities of modern agriculture, are valuable because they create a stable environment suitable for crop growth through artificial intervention, thereby achieving high-yield, high-quality, and off-season production. At present, greenhouse environmental control methods mainly remain at two levels: one is rudimentary management relying on human experience, which is labor-intensive, subjective, and cannot guarantee precise control around the clock; the other is the more widely used single-factor threshold automatic control, which monitors a specific environmental indicator (such as temperature) through independent sensors, and triggers corresponding equipment (such as fans) to switch on and off when the temperature exceeds a preset threshold. However, with the deepening development of precision agriculture, the inherent defects of these existing technologies are becoming increasingly prominent, becoming a key bottleneck restricting further improvement in greenhouse production efficiency.

[0003] The most fundamental flaw in existing technologies lies in the isolation and lack of synergy in their control logic. Environmental data within a greenhouse, such as light, temperature, humidity, carbon dioxide concentration, and soil water and fertilizer levels, form an interconnected and tightly coupled organic whole. For example, opening ventilation to cool the greenhouse inevitably leads to a simultaneous loss of indoor humidity and carbon dioxide concentration; increasing light to promote photosynthesis also requires a corresponding increase in carbon dioxide supply and irrigation. Existing threshold control systems precisely ignore this inherent coupling, controlling each factor independently and in isolation. This easily leads to conflicting actions between different devices (such as heating while simultaneously opening windows for ventilation), resulting not only in significant active energy waste but also in drastic fluctuations in environmental parameters, creating continuous stress on crop growth.

[0004] Furthermore, another major drawback of existing technologies lies in the passivity and lag of their control methods. They can only respond in a reactive, "better late than never" manner after environmental parameters have deviated from the crop's optimal range, failing to incorporate forward-looking predictions and adjustments based on external information such as weather forecasts. Simultaneously, their control strategies are fixed and unchanging, with a single set of threshold parameters applied throughout the crop's growth process. This fails to dynamically match the differentiated environmental needs of crops at different life stages, such as seedling, flowering, and fruiting, thus limiting the full realization of the crop's growth potential. This passive, rigid, and uncoordinated control model ultimately leads to persistently high energy consumption in greenhouse operations, severe waste of water and fertilizer resources, and a significant increase in production costs. Summary of the Invention

[0005] This application provides a method and apparatus for intelligent control of multiple environmental data in a greenhouse, which solves the technical problem of the current situation in greenhouse control where each environmental data operates independently and lacks coordination.

[0006] In a first aspect, embodiments of this application provide a method for collaborative intelligent control of multiple environmental data within a greenhouse, comprising: Based on current environmental data affecting crop growth in the greenhouse and meteorological data for a period of time in the future, predictive values ​​of various environmental data for future times are obtained. Based on the optimal and predicted values ​​of various environmental data required for crop growth in the greenhouse at the current moment, the first target control strategy for each controllable device in the greenhouse at the current moment is generated. The controllable device is the device used to control the environmental data. Control each controllable device according to the first objective control strategy.

[0007] In one embodiment, based on the optimal and predicted values ​​of various environmental data required for crop growth in the greenhouse at the current moment, a first target control strategy for each controllable device in the greenhouse at the current moment is generated, including: The objective function is determined based on the weighted sum of squared errors between the predicted value and the optimal value, the first dynamic coefficient, the total power required to execute the optional control strategy, the second dynamic coefficient, the penalty term for the degree of change between the optional control strategy at the current time and the optional control strategies of each controllable device at the previous time, and the third dynamic coefficient. A multi-objective optimization algorithm is adopted to select the first objective control strategy that minimizes the objective function from multiple optional control strategies of each controllable device at the current time.

[0008] In one embodiment, the method further includes: Based on the second target control strategy of each controllable device at the previous moment, determine the first dynamic coefficient, the second dynamic coefficient, and the third dynamic coefficient.

[0009] In one embodiment, determining the first dynamic coefficient, the second dynamic coefficient, and the third dynamic coefficient based on the second target control strategy of each controllable device at the previous moment includes: According to the second objective control strategy, control each controllable device and acquire various environmental data and the operating status of each controllable device from the previous moment; Based on the optimal values ​​of various environmental data from the previous moment and the various environmental data required for crop growth in the greenhouse from the previous moment, determine the environmental accuracy bonus value. Based on the operating status of each controllable device in the previous moment, determine the total power consumed by executing the second target control strategy; Based on the operating status of each controllable device at the previous moment and the operating status of each controllable device at the moment before that, determine the energy consumption cost caused by the change in action. Based on the environmental accuracy reward value, the total power consumed in executing the second objective control strategy, and the energy consumption cost, determine the reward value corresponding to the execution of the second objective control strategy at the previous moment; Based on the reward value, determine the first dynamic coefficient, the second dynamic coefficient, and the third dynamic coefficient.

[0010] In one embodiment, based on various environmental data affecting crop growth in the greenhouse at the current moment and meteorological data for a future period, predicted values ​​of various environmental data for future moments are obtained, including: Feature data is obtained by splicing and feature engineering various environmental data affecting crop growth in the greenhouse at the current moment and meteorological data for a period of time in the future; The feature data is input into a long short-term memory network to obtain predicted values ​​of various environmental data at future moments.

[0011] In one embodiment, the optimal values ​​for various environmental data required for crop growth in the greenhouse at the current moment are determined based on the current growth stage of the crops in the greenhouse.

[0012] Secondly, embodiments of this application provide a multi-environmental data collaborative intelligent control device for greenhouses, comprising: The first acquisition module is used to acquire predicted values ​​of various environmental data for future times based on the current environmental data affecting crop growth in the greenhouse and meteorological data for a period of time in the future. The second acquisition module is used to generate the first target control strategy for each controllable device in the greenhouse at the current moment based on the optimal and predicted values ​​of various environmental data required for crop growth in the greenhouse at the current moment. The controllable devices are devices used to control environmental data. The control module is used to control each controllable device according to the first target control strategy.

[0013] Thirdly, embodiments of this application provide an electronic device, including a processor and a memory storing a computer program, wherein the processor executes the program to implement the steps of the first aspect of the greenhouse multi-environment data collaborative intelligent control method.

[0014] Fourthly, embodiments of this application provide a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the multi-environmental data collaborative intelligent control method for greenhouses as described in the first aspect.

[0015] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the steps of the first aspect of the greenhouse multi-environment data collaborative intelligent control method.

[0016] The method and apparatus for collaborative intelligent control of multiple environmental data in greenhouses provided in this application calculate a set of collaborative control strategies involving multiple controllable devices by comprehensively considering real-time environmental data affecting crop growth within the greenhouse, predicted values ​​of each environmental data point for future times, and optimal values ​​of various environmental data required by the crop at the current moment. This approach moves beyond simply judging the "on / off" state of individual devices; instead, it treats the greenhouse as a dynamically coupled whole, achieving collaborative, precise, and intelligent control of multiple environmental data points (such as light, temperature, water, air, and fertilizer) within the greenhouse. This fundamentally solves the current situation where various factors operate independently and lack coordination in greenhouse control, providing crops with an unprecedentedly stable and balanced growth environment and significantly reducing operating energy consumption. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the intelligent control method for multi-environmental data collaboration in a greenhouse provided in an embodiment of this application. Figure 2 This is a schematic diagram of the structure of the intelligent environmental control software system provided in the embodiments of this application; Figure 3 This is a schematic diagram of the environmental change trend prediction process provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of the intelligent control device for multi-environmental data collaboration in a greenhouse provided in this application embodiment; Figure 5 This is a schematic diagram of the structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0020] To address the aforementioned challenges in existing technologies, this application presents a method and device for collaborative intelligent control of multiple environmental data within greenhouses. The aim is to fundamentally change the current situation where various factors operate independently and lack coordination in greenhouse control. By constructing an intelligent decision-making core capable of comprehensively analyzing multi-dimensional environmental data, possessing predictive capabilities, and performing closed-loop self-learning, it achieves collaborative, precise, and intelligent control of multiple environmental factors in the greenhouse, including light, temperature, water, air, and fertilizer. The ultimate goal is to minimize system energy and resource consumption while ensuring high-quality and high-yield crops, thereby comprehensively improving the level of intelligence and overall economic benefits of greenhouse production.

[0021] Figure 1 This is a flowchart illustrating the intelligent control method for multi-environmental data collaboration in a greenhouse provided in this application embodiment. The executing entity can be an industrial control computer, a local computer, a cloud server, or other mobile electronic terminals with data processing functions, such as iPads or mobile phones. This invention does not specifically limit the specific implementation of these devices.

[0022] Reference Figure 1 This application provides a method for collaborative intelligent control of multiple environmental data in a greenhouse, which may include steps 110, 120 and 130.

[0023] Step 110: Based on the current environmental data affecting crop growth in the greenhouse and the meteorological data for a period of time to come, obtain the predicted values ​​of various environmental data for future times.

[0024] Acquire various environmental data within the greenhouse, such as temperature, humidity, light radiation, and carbon dioxide concentration, to provide data support for the coordinated control of various controllable devices within the greenhouse.

[0025] To improve the authenticity and reliability of the data, this step utilizes different types of sensors installed inside the greenhouse to comprehensively measure and process the environmental data within the greenhouse. This eliminates errors and noise from individual sensors, improves the accuracy and stability of the data, effectively addresses the limitations and shortcomings of single sensors, and enhances the integrity and credibility of the environmental data within the greenhouse.

[0026] For example, an environmental intelligent control software system deployed on a cloud platform or edge computing node, such as Figure 2 As shown, it achieves a complete closed loop from data perception to intelligent decision-making and adaptive optimization through a series of closely connected software modules and algorithms.

[0027] This intelligent environmental control software system adopts a layered and decoupled design, resulting in a clear overall architecture that facilitates functional expansion and maintenance. The architecture mainly includes a data interface and adaptation layer, a core service layer, and an application and presentation layer.

[0028] Through a data interface and adaptation layer, data from various sensor nodes inside the greenhouse is subscribed to and received in real time. This data is typically encapsulated in JSON format, containing fields such as controllable device ID, timestamp, data type, and numerical value. The real-time environmental data collected by various sensors inside the greenhouse is first sent to the "data preprocessing module" for preprocessing. Specific implementations here include: data cleaning: using a 3-sigma principle or moving window mean filtering algorithm to automatically identify and remove obvious outliers caused by sensor malfunctions or signal interference; data completion: for missing data at certain times, the mean of adjacent times or linear interpolation is used to complete the data, ensuring the continuity of the data stream; data standardization: to facilitate subsequent processing by neural networks and other models, all numerical data (such as temperature, humidity, and light intensity) are scaled to a uniform range of [0, 1] using a max-min normalization method. The processed clean data is then written to a "time-series database" for persistent storage.

[0029] In addition, the system is equipped with a "video monitoring" module for real-time monitoring of the entire processing.

[0030] In this step, the "preprocessing module" preprocesses various environmental data collected in real time within the greenhouse at the current moment, extracting this data along with meteorological data (e.g., weather forecast data) for a future period (e.g., 3 hours, 6 hours, etc.). Figure 2 The feature vector obtained from the third-party weather forecast API (as shown) is used to predict the natural changing trends of various environmental data over a future period. Based on this changing trend, the predicted values ​​of various environmental data for future timeframes can be obtained.

[0031] For example, this portion of the energy supply can be provided by Figure 2 The "Predictive Analytics Engine" shown is now complete.

[0032] Step 120: Based on the optimal and predicted values ​​of various environmental data required for crop growth in the greenhouse at the current moment, generate the first target control strategy for each controllable device in the greenhouse at the current moment. The controllable device is the device used to control the environmental data.

[0033] In this step, the optimal values ​​of various environmental data required for crop growth in the greenhouse at the current moment can be obtained by querying the pre-established "crop growth model library".

[0034] For example, some contents of the "Crop Growth Model Library" are shown in Table 1: Table 1: Crop Growth Model Library

[0035] In one embodiment, the optimal values ​​for various environmental data required for crop growth in the greenhouse at the current moment are determined based on the current growth stage of the crops in the greenhouse.

[0036] For example, based on the crop type and its fixed date, the current growth stage of the crop can be determined. Combined with Table 1, the optimal values ​​of various environmental data required for crop growth at the current moment can be obtained.

[0037] In practice, the system will automatically query this table based on the crop type set by the user and the current number of days planted, match the current "Growth_Stage", and extract all the optimal environmental parameters (i.e. the best values ​​of various environmental parameters) corresponding to this stage to form a multi-dimensional "environmental target vector" for subsequent decision engine use.

[0038] Unlike traditional fixed threshold control, the environmental target vector in this application is dynamic and determined by the current growth stage of the crop.

[0039] A pre-set decision-making model or algorithm can be used to analyze the optimal and predicted values ​​of various environmental data required for crop growth in the greenhouse at the current moment, thereby generating the target control strategy for various controllable devices in the greenhouse at the current moment, i.e., the first target control strategy. Here, controllable devices refer to those used to regulate environmental data, such as fans, skylights, and supplemental lighting. The target control strategy refers to the combination of control actions of all controllable devices used to regulate environmental data, such as increasing or decreasing fan speed, increasing or decreasing skylight opening, and increasing or decreasing supplemental lighting power.

[0040] Step 130: Control each controllable device according to the first target control strategy.

[0041] Control of each controllable device is achieved based on the control actions corresponding to each controllable device in the first target control strategy.

[0042] The intelligent multi-environmental data collaborative control method for greenhouses provided in this application calculates a collaborative control strategy involving multiple controllable devices by comprehensively considering real-time environmental data affecting crop growth, predicted values ​​of various environmental data at future times, and optimal values ​​of various environmental data required by the crop at the current moment. This approach moves beyond simply judging the "on / off" of individual device parameters; instead, it treats the greenhouse as a dynamically coupled whole, achieving collaborative, precise, and intelligent control of various environmental data (such as light, temperature, water, air, and fertilizer) within the greenhouse. This fundamentally solves the current situation where various factors operate independently and lack collaboration in greenhouse control, providing crops with an unprecedentedly stable and balanced growth environment and significantly reducing operating energy consumption.

[0043] In one embodiment, step 120, generating a first target control strategy for each controllable device in the greenhouse at the current moment based on the optimal and predicted values ​​of various environmental data required for crop growth in the greenhouse at the current moment, may include: The objective function is determined based on the weighted sum of squared errors between the predicted value and the optimal value, the first dynamic coefficient, the total power required to execute the optional control strategy, the second dynamic coefficient, the penalty term for the degree of change between the optional control strategy at the current time and the optional control strategies of each controllable device at the previous time, and the third dynamic coefficient. A multi-objective optimization algorithm is adopted to select the first objective control strategy that minimizes the objective function from multiple optional control strategies of each controllable device at the current time.

[0044] In practical implementation, it can be done through methods such as... Figure 2 The "collaborative decision-making engine" shown uses the optimal values ​​and predicted values ​​of various environmental data required for crop growth in the greenhouse at the current moment to construct an objective function. And by solving the objective function The system selects the control strategy that minimizes the objective function from multiple available control strategies for each controllable device at the current moment; this control strategy is the first objective control strategy. The core of this "collaborative decision engine" is a multi-objective optimization solver based on a multi-objective optimization algorithm (such as a genetic algorithm).

[0045] Wherein, objective function The expression is as follows: In the formula, The first dynamic coefficient, This is the second dynamic coefficient. The third dynamic coefficient, This represents all available control strategies for each control device at the current moment. This represents all available control strategies for each control device at the previous moment. , and These represent different emphases on "environmental precision", "energy saving" and "stable equipment operation".

[0046] This is the weighted sum of squared errors between the predicted values ​​of various environmental data at the current moment and the "dynamic target" (i.e., the optimal value). The smaller the value, the closer the predicted future environmental data is to the optimal environmental target required for the crop's current growth stage. The predicted environmental data mentioned here does not refer to the current actual readings, but rather the predicted values ​​of various environmental data for future moments obtained by the "predictive analysis engine" based on the current environmental data in the greenhouse and meteorological data for a period of time in the future. The "dynamic target" it compares can be understood as a "list of ideal environmental parameters" tailored to the crop and updated in real time according to its growth stage, which includes specific values ​​such as the most suitable temperature, humidity, and carbon dioxide concentration for that stage.

[0047] Is it to execute optional control strategies? The smaller the total power required (an estimated value of total power can be used here), the more likely an optional control strategy will be implemented. The less total energy required, the better. First, the "optional control strategy" must be clearly defined. "Power consumption" refers to a set of instructions that includes specific operations for all "controllable devices". Therefore, the process of calculating the total power consumption is that the system estimates the power consumption of each device in the operating state under the instruction based on this set of instructions, and then adds them all together to obtain the total energy cost of performing this complete operation.

[0048] It measures the optional control strategy to be executed at the current moment. The optional control strategy executed in the previous moment. The penalty term represents the degree of drastic change between adjacent time steps. A smaller value indicates a smaller and smoother change in the current optional control strategy compared to the previous time step. This function quantifies the stability of the optional control strategy by comparing the differences between the optional control strategies at adjacent time steps. A drastic change will receive a high penalty score, thus guiding the system to make smoother and more gradual adjustments, thereby protecting the equipment and stabilizing the environment.

[0049] The selectable control strategies for each controllable device at each time point can be randomly generated by multi-objective optimization algorithms, such as genetic algorithms.

[0050] In practice, hundreds of possible control strategies are randomly generated as the initial "population." For each individual in the population, its "fitness" (the smaller the value, the higher the fitness) is calculated using the objective function described above. Simulating biological evolution, superior individuals are selected using methods such as roulette wheel selection, and they are then subjected to "crossover" (exchanging some control parameters) and "mutation" (randomly fine-tuning a control parameter) to generate a new generation of the population. This process is repeated for dozens of generations, and the population gradually evolves until it converges to a control strategy that minimizes, reaches, or becomes globally optimal or suboptimal for the objective function. X_best (i.e., the first objective control strategy).

[0051] In other embodiments, the evolutionary process may be configured to stop and output the currently found optimal solution when any of the following preset conditions are met: 1. Reach the maximum number of iterations: Complete the preset number of evolution generations (e.g., 100 generations).

[0052] 2. Fitness convergence: The optimal solution does not improve significantly over multiple generations (i.e., the objective function reaches its minimum value).

[0053] 3. Time limit: The maximum calculation time (e.g., 2 minutes) is set to ensure the real-time performance of the system.

[0054] This is the final result. X_best This is the optimal collaborative control strategy for the system at present. It is no longer a simple on / off command, but a sophisticated combination of actions involving multiple devices.

[0055] The multi-environmental data collaborative intelligent control method for greenhouses provided in this application differs from existing technologies that commonly employ single-variable threshold control logic, treating various environmental factors within the greenhouse in isolation. This leads to conflicts between control actions, causing drastic environmental fluctuations and energy waste. The key innovation of this application lies in constructing a multi-objective collaborative optimization decision model. This model no longer simply makes "on / off" judgments on individual parameters, but treats the greenhouse as a dynamically coupled whole. It can comprehensively evaluate all real-time environmental data, future trends, and the optimal needs of the crop at its current stage, thereby calculating a set of collaborative control strategies involving multiple devices with different action amplitudes. This collaborative control mode fundamentally solves the resource consumption problem caused by action conflicts in traditional methods, providing crops with an unprecedentedly stable and balanced growth environment and significantly reducing operating energy consumption.

[0056] In one embodiment, the above method may further include: Based on the second target control strategy of each controllable device at the previous moment, determine the first dynamic coefficient, the second dynamic coefficient, and the third dynamic coefficient.

[0057] In this embodiment of the application, the environmental intelligent control software system will implement the target control strategy. X_best After the order is issued and implemented, it will be done through methods such as... Figure 2 The self-learning and optimization module shown evaluates and learns from this control.

[0058] Once a control strategy is completed, the system calculates a "reward value" based on actual environmental data. For example, if the environment stably reaches the target within a short period of time and energy consumption is low, a high positive reward is given; conversely, if the environment fluctuates greatly or energy consumption is high, a negative reward is given.

[0059] Specifically, after each control strategy is executed, the system quantifies the overall effect of the control through a precise reward function as the basis for the agent's learning, and adjusts the first, second, and third dynamic coefficients of the objective function based on the calculated reward value.

[0060] For example, for the current moment The first, second, and third dynamic coefficients in the corresponding objective function can be determined based on the previous time step. The target control strategy (i.e., the second target control strategy) for each controllable device is used to calculate the reward value. Based on the reward value, the update gradients of the first dynamic coefficient, the second dynamic coefficient, and the third dynamic coefficient are calculated, and the gradient descent method is used to adjust the control strategy. The first dynamic coefficient, the second dynamic coefficient, and the third dynamic coefficient in the objective function at time t.

[0061] Similarly, for The first, second, and third dynamic coefficients of the objective function at time t can be obtained by... The primary objective control strategy for each controllable device at any given time is determined, the corresponding reward value is calculated, and adjustments are made accordingly. The first dynamic coefficient, the second dynamic coefficient, and the third dynamic coefficient in the objective function at time t.

[0062] In one embodiment, determining the first dynamic coefficient, the second dynamic coefficient, and the third dynamic coefficient based on the second target control strategy of each controllable device at the previous moment may include: According to the second objective control strategy, control each controllable device and acquire various environmental data and the operating status of each controllable device from the previous moment; Based on the optimal values ​​of various environmental data from the previous moment and the various environmental data required for crop growth in the greenhouse from the previous moment, determine the environmental accuracy bonus value. Based on the operating status of each controllable device at the previous moment, determine the total power consumed by executing the second target control strategy; Based on the operating status of each controllable device at the previous moment and the operating status of each controllable device at the moment before that, determine the energy consumption cost caused by the change in action. Based on the environmental accuracy reward value, the total power consumed in executing the second objective control strategy, and the energy consumption cost, determine the reward value corresponding to the execution of the second objective control strategy at the previous moment; Based on the reward value, determine the first dynamic coefficient, the second dynamic coefficient, and the third dynamic coefficient.

[0063] In this embodiment of the application, the previous moment ( The reward function for time () is as follows: In the formula, , and These are the weighting coefficients. for Reward value at any moment This is the environmental accuracy bonus value, used to measure the degree to which the actual environment matches the target. Its calculation method is as follows: In the formula, yes The first moment The actual value of the environmental data represents the value actually measured by the sensor at time t-1 after controlling each controllable device through the second objective control strategy. Environmental data.

[0064] It is the first time at time t-1 The optimal value for each environmental data point represents the value of the i-th environmental data point. The ideal value that should be achieved at any given time.

[0065] It is the tolerance parameter, represented by the first... An acceptable error range is set for each environmental data point, if Very small; for example, if the temperature tolerance is set to 0.5, then a deviation of 0.5 degrees will cause a significant drop in the score, indicating that very high precision is required. If A relatively large tolerance, for example, setting the CO2 concentration tolerance to 50 ppm means that a relatively large fluctuation is allowed.

[0066] This is the energy consumption cost, representing the total power consumed in executing the second objective control strategy. Its calculation formula is: In the formula, It refers to the operating status of controllable equipment, representing the first... The operating status of a controllable device at time t-1. For example, the fan speed, the power percentage of the supplementary lighting, and the on / off status (0 or 1) of the heater.

[0067] This is a power consumption function for controllable devices. It's a function used to calculate the power consumption of a device. Given the operating state of a device, it outputs the real-time power consumption of that device in that state.

[0068] This is the energy consumption cost, used to penalize drastic changes in control actions, representing the energy consumption cost resulting from the change in action. Its calculation formula is: In the formula, It is the first The device at the previous time ( The state of action at any given moment It represents the square of the change in operating state. The difference between the operating state of the controllable equipment at the previous moment and the operating state at the moment before that is calculated, and then squared. Squaring ensures the result is always positive and also inflicts disproportionately high penalties on drastic changes. For example, the penalty for a 2-unit change is 4, while the penalty for a 4-unit change is 16, much more than twice the former. This effectively promotes smooth and gradual adjustments in the system.

[0069] The system records each complete "(state + prediction) -> (action) -> (reward)" process as training data. By analyzing this data, the agent learns which actions to take under what states and predictions to obtain the greatest long-term cumulative reward.

[0070] The reward value of the previous moment is calculated based on the above formula, and the first dynamic coefficient, second dynamic coefficient, and third dynamic coefficient of the objective function at the current moment are adjusted accordingly.

[0071] For example, if it detects that recent high energy consumption has led to a low reward value, it will automatically increase the weight of the energy consumption item (i.e., the second dynamic coefficient). This allows the collaborative decision-making engine to prioritize energy conservation in subsequent decisions. Through this continuous self-feedback and optimization, the system can adapt to dynamic factors such as seasonal changes and equipment aging, achieving true self-adaptation and intelligence.

[0072] In practical applications, after initial model training and system deployment, the actual application of this application in the greenhouse will follow a continuous and automated closed loop of "perception-prediction-decision-execution-learning". A typical control cycle flow is as follows: Initialization and configuration: When the system is first started, users only need to configure the basic information of the greenhouse, the types of crops currently being planted and their planting dates.

[0073] Start the automated control closed loop: The system enters fully automatic operation mode, and the following steps will be executed repeatedly.

[0074] Sensing and Calibration: The system collects real-time environmental data from all sensors and preprocesses it through a data preprocessing module. Simultaneously, it calculates the crop's growth stage based on the current date and queries the "Crop Growth Model Library" to determine the optimal values ​​for various environmental data at the current moment.

[0075] Predicting future trends: By calling a pre-trained long short-term memory model and combining it with various environmental data and external weather forecasts at the current moment, the system accurately predicts the natural change trend of the greenhouse environment in the next 1-3 hours, and obtains the predicted values ​​of various environmental data in the greenhouse at future moments.

[0076] Intelligent decision-making: The "collaborative decision-making engine" is activated. It takes the "predicted value" and the "optimal value" as inputs, and runs a genetic algorithm with the latest optimized weights of the agent to quickly solve for the collaborative control strategy with the best energy efficiency and the most stable control, i.e., the target control strategy.

[0077] Execution and Feedback Learning: The system parses the target control strategy into specific instructions and issues them to the controllable equipment for execution. After a control cycle ends, the system calculates a reward value—a newly generated valuable experience data—based on the actual effect and energy consumption of this control. This reward value is then fed back to the agent as training data for its next round of optimization learning.

[0078] The intelligent control method for multi-environmental data collaboration in greenhouses provided in this application, compared to the static and unchanging control strategies of existing technologies, adopts a dynamic target adaptation based on crop growth models. The system can automatically adjust the optimal environmental target according to the different growth stages of the crop, achieving precise and personalized management throughout the crop's entire life cycle. Based on closed-loop feedback self-optimization through reinforcement learning, the system continuously learns the correlation between the "control strategy" and the "environmental feedback results," constantly self-correcting and optimizing its internal decision-making model. This self-learning mechanism gives this application the characteristic of "becoming smarter with use," gradually adapting to the physical characteristics of a specific greenhouse and the growth habits of specific crops, ultimately achieving long-term optimal operation without human intervention.

[0079] In one embodiment, based on various environmental data affecting crop growth in the greenhouse at the current moment and meteorological data for a future period, predicted values ​​of various environmental data for future moments are obtained, including: Feature data is obtained by splicing and feature engineering various environmental data affecting crop growth in the greenhouse at the current moment and meteorological data for a period of time in the future; The feature data is input into a long short-term memory network to obtain predicted values ​​of various environmental data at future moments.

[0080] For example, refer to Figure 3 The system utilizes a "predictive analytics engine" to stitch together and feature-engineer various environmental data affecting crop growth within the greenhouse at the current moment, along with meteorological data for a future period. The resulting feature data (typically represented as feature vectors) is then input into a pre-trained Long Short-Term Memory (LSTM) network. Due to its excellent memory capacity for time-series data, this network can learn the thermal inertia of a specific greenhouse, its response speed to external light, and other factors. This allows it to accurately predict the natural trends of various environmental data within the greenhouse over a future period (e.g., 1 to 3 hours) without any intervention (e.g., the temperature and humidity curves within the greenhouse over the next 3 hours), thereby obtaining predicted values ​​for each environmental data point in the future.

[0081] The intelligent control method for multi-environmental data collaboration in greenhouses provided in this application differs from traditional control methods, which are passive and reactive, only taking remedial action after the environment has deviated from its ideal state. This application introduces a future situational awareness capability based on time-series prediction. By fusing real-time environmental data within the greenhouse with external weather forecast data and utilizing deep learning algorithms such as Long Short-Term Memory (LSTM) networks, it can accurately predict the natural changing trends of the greenhouse environment over a future period. This "predictive capability" allows the control system to intervene and plan ahead, such as preheating before a cold wave or slowly closing the shading net before strong sunlight. This proactive control method completely eliminates control lag, avoids overshooting and frequent oscillations of environmental parameters, and makes the entire control process smoother and more efficient.

[0082] The following describes the intelligent control device for multi-environmental data collaboration in a greenhouse provided in the embodiments of this application. The intelligent control device for multi-environmental data collaboration in a greenhouse described below can be referred to in correspondence with the intelligent control method for multi-environmental data collaboration in a greenhouse described above.

[0083] Figure 4 This is a schematic diagram of the structure of the intelligent control device for multi-environmental data collaboration in a greenhouse provided in this application embodiment, with reference to... Figure 4 The intelligent control device for multi-environmental data collaboration in a greenhouse provided in this application may include: The first acquisition module 410 is used to acquire predicted values ​​of various environmental data for future times based on the current environmental data affecting crop growth in the greenhouse and meteorological data for a period of time in the future. The second acquisition module 420 is used to generate the first target control strategy for each controllable device in the greenhouse at the current moment based on the optimal and predicted values ​​of various environmental data required for crop growth in the greenhouse at the current moment. The controllable device is the device used to control environmental data. The control module 430 is used to control each controllable device according to the first target control strategy.

[0084] The intelligent control device for multi-environmental data collaboration in greenhouses provided in this application calculates a collaborative control strategy involving multiple controllable devices by comprehensively considering real-time environmental data affecting crop growth, predicted values ​​of various environmental data for future times, and optimal values ​​of various environmental data required by the crop at the current moment. It moves beyond simply judging the "on / off" of individual device parameters, treating the greenhouse as a dynamically coupled whole to achieve collaborative, precise, and intelligent control of various environmental data (such as light, temperature, water, air, and fertilizer). This fundamentally solves the current situation where various factors operate independently and lack collaboration in greenhouse control, providing crops with an unprecedentedly stable and balanced growth environment and significantly reducing operating energy consumption.

[0085] Figure 5 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 5 As shown, the electronic device may include: a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510, the communication interface 520, and the memory 530 communicate with each other via the communication bus 540. The processor 510 can call the computer program in the memory 530 to execute the steps of the multi-environment data collaborative intelligent control method in the greenhouse, such as including: Based on current environmental data affecting crop growth in the greenhouse and meteorological data for a period of time in the future, predictive values ​​of various environmental data for future times are obtained. Based on the optimal and predicted values ​​of various environmental data required for crop growth in the greenhouse at the current moment, the first target control strategy for each controllable device in the greenhouse at the current moment is generated. The controllable device is the device used to control the environmental data. Control each controllable device according to the first objective control strategy.

[0086] Furthermore, the logical instructions in the aforementioned memory 530 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0087] On the other hand, this application also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can perform the steps of the greenhouse multi-environment data collaborative intelligent control method provided in the above embodiments, such as including: Based on current environmental data affecting crop growth in the greenhouse and meteorological data for a period of time in the future, predictive values ​​of various environmental data for future times are obtained. Based on the optimal and predicted values ​​of various environmental data required for crop growth in the greenhouse at the current moment, the first target control strategy for each controllable device in the greenhouse at the current moment is generated. The controllable device is the device used to control the environmental data. Control each controllable device according to the first objective control strategy.

[0088] On the other hand, embodiments of this application also provide a processor-readable storage medium storing a computer program for causing a processor to perform the steps of the methods provided in the above embodiments, such as including: Based on current environmental data affecting crop growth in the greenhouse and meteorological data for a period of time in the future, predictive values ​​of various environmental data for future times are obtained. Based on the optimal and predicted values ​​of various environmental data required for crop growth in the greenhouse at the current moment, the first target control strategy for each controllable device in the greenhouse at the current moment is generated. The controllable device is the device used to control the environmental data. Control each controllable device according to the first objective control strategy.

[0089] The processor-readable storage medium can be any available medium or data storage device that the processor can access, including but not limited to magnetic memory (e.g., floppy disk, hard disk, magnetic tape, magneto-optical disk (MO)), optical memory (e.g., CD, DVD, BD, HVD), and semiconductor memory (e.g., ROM, EPROM, EEPROM, non-volatile memory (NAND FLASH), solid-state drive (SSD)).

[0090] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0091] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for collaborative intelligent control of multiple environmental data within a greenhouse, characterized in that, include: Based on current environmental data affecting crop growth in the greenhouse and meteorological data for a period of time in the future, predictive values ​​of various environmental data for future times are obtained. Based on the optimal values ​​of various environmental data required for crop growth in the greenhouse at the current moment and the predicted values, a first target control strategy is generated for each controllable device in the greenhouse at the current moment, wherein the controllable device is a device used to control environmental data. Control each controllable device according to the first target control strategy.

2. The method for collaborative intelligent control of multiple environmental data in a greenhouse according to claim 1, characterized in that, Based on the optimal values ​​of various environmental data required for crop growth in the greenhouse at the current moment and the predicted values, a first target control strategy is generated for each controllable device in the greenhouse at the current moment, including: The objective function is determined based on the weighted sum of squared errors between the predicted value and the optimal value, the first dynamic coefficient, the total power required to execute the optional control strategy, the second dynamic coefficient, the penalty term for the degree of change between the optional control strategy at the current time and the optional control strategies of each controllable device at the previous time, and the third dynamic coefficient. A multi-objective optimization algorithm is used to select the first objective control strategy that minimizes the objective function from multiple optional control strategies of each controllable device at the current time.

3. The method for collaborative intelligent control of multiple environmental data in a greenhouse according to claim 2, characterized in that, The method further includes: Based on the second target control strategy of each of the controllable devices at the previous moment, the first dynamic coefficient, the second dynamic coefficient, and the third dynamic coefficient are determined.

4. The method for collaborative intelligent control of multiple environmental data in a greenhouse according to claim 3, characterized in that, Based on the second target control strategy of each of the controllable devices at the previous moment, the first dynamic coefficient, the second dynamic coefficient, and the third dynamic coefficient are determined, including: According to the second target control strategy, control each of the controllable devices to acquire various environmental data and the operating status of each of the controllable devices at the previous moment; Based on the optimal values ​​of various environmental data from the previous moment and the various environmental data required for crop growth in the greenhouse from the previous moment, determine the environmental accuracy bonus value. Based on the operating status of each controllable device at the previous moment, determine the total power consumed by executing the second target control strategy; Based on the operating status of each controllable device at the previous moment and the operating status of each controllable device at the moment before that, determine the energy consumption cost caused by the change in action; Based on the environmental accuracy reward value, the total power consumed in executing the second target control strategy, and the energy consumption cost, determine the reward value corresponding to executing the second target control strategy at the previous moment; Based on the reward value, the first dynamic coefficient, the second dynamic coefficient, and the third dynamic coefficient are determined.

5. The method for collaborative intelligent control of multiple environmental data in a greenhouse according to any one of claims 1-4, characterized in that, The process of obtaining predicted values ​​for various environmental data for future times based on current environmental data affecting crop growth in the greenhouse and meteorological data for a future period includes: Feature data is obtained by splicing and feature engineering various environmental data affecting crop growth in the greenhouse at the current moment and meteorological data for a future period of time. The feature data is input into a long short-term memory network to obtain predicted values ​​of various environmental data at future times.

6. The method for collaborative intelligent control of multiple environmental data in a greenhouse according to any one of claims 1-4, characterized in that, The optimal values ​​for various environmental data required for crop growth in the greenhouse at the current moment are determined based on the current growth stage of the crops in the greenhouse.

7. A multi-environmental data collaborative intelligent control device for greenhouses, characterized in that, include: The first acquisition module is used to acquire predicted values ​​of various environmental data for future times based on the current environmental data affecting crop growth in the greenhouse and meteorological data for a period of time in the future. The second acquisition module is used to generate a first target control strategy for each controllable device in the greenhouse at the current moment based on the optimal values ​​of various environmental data required for crop growth in the greenhouse at the current moment and the predicted values. The controllable device is a device used to control environmental data. The control module is used to control each controllable device according to the first target control strategy.

8. An electronic device comprising a processor and a memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the multi-environmental data collaborative intelligent control method in a greenhouse as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the multi-environment data collaborative intelligent control method in the greenhouse as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the multi-environmental data collaborative intelligent control method in a greenhouse as described in any one of claims 1 to 6.

Images