A Smart Greenhouse Temperature Control Method Based on Multi-Source Sensors

Abstract

This invention discloses an intelligent greenhouse temperature control method based on multi-source sensors, relating to the field of seed and seedling cultivation technology. The method includes the following steps: data acquisition; preliminary temperature prediction based on the acquired data; designing a temperature control strategy based on the temperature prediction results; adjusting the greenhouse environment according to the temperature control strategy; predicting crop growth based on temperature and humidity data; intelligent irrigation management based on the crop growth prediction results; optimizing greenhouse energy efficiency based on the irrigation adjustment results; and providing comprehensive greenhouse management decision support based on the energy efficiency optimization results. This invention improves the method for converting growth prediction values ​​output by deep learning models and designs the reward function of reinforcement learning algorithms, employing an innovative reinforcement learning reward function design method. This innovative approach not only ensures precise alignment between crop growth status and greenhouse environmental control but also ensures healthy crop growth while improving energy efficiency.

Description

Technical Field

[0001] This invention relates to the field of seed and seedling cultivation technology, specifically to an intelligent greenhouse temperature control method based on multi-source sensors. Background Technology

[0002] In recent years, advanced artificial intelligence technologies such as deep learning and reinforcement learning have been increasingly applied in greenhouse management for seed and seedling cultivation, especially in crop growth prediction and environmental control strategy optimization, demonstrating strong potential. With the rapid development of agricultural intelligence, the regulation of greenhouse environments (such as temperature, humidity, and light intensity) increasingly relies on the synergistic application of sensor data and intelligent algorithms.

[0003] To ensure that crops receive the best growing environment in greenhouses, researchers have proposed a method for fusing multi-source sensor data. This method uses information such as temperature, humidity, air velocity, light intensity, and soil moisture to monitor and adjust in real time, and uses a reward function based on a reinforcement learning algorithm to achieve intelligent control of the greenhouse environment.

[0004] In the existing technology, the reward function design of reinforcement learning algorithms is insufficient to fully consider the complexity of multi-source sensor data. Existing reinforcement learning algorithms often rely on simple reward function design, which fails to fully consider the complex nonlinear relationship between crop growth and greenhouse environment. This makes it difficult for optimization strategies to cope with changing environmental conditions and cannot achieve long-term, stable energy optimization and crop growth improvement in greenhouses. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides an intelligent greenhouse temperature control method based on multi-source sensors to solve the problems mentioned in the background section.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] In a first aspect, embodiments of the present invention provide an intelligent greenhouse temperature control method based on multi-source sensors, comprising the following steps:

[0008] S1, Data Acquisition;

[0009] Environmental data inside the greenhouse is collected by multi-source sensors and cleaned using data preprocessing techniques to obtain the collected data.

[0010] S2. Perform preliminary temperature prediction based on the collected data;

[0011] Preliminary temperature predictions are made from the collected data using prediction software to obtain temperature prediction results.

[0012] S3. Obtain a temperature control strategy based on the temperature prediction results;

[0013] Based on fuzzy control, a temperature control strategy is designed according to the temperature prediction results, and a temperature control strategy is obtained.

[0014] S4. Adjust the greenhouse environment according to the temperature control strategy;

[0015] The temperature and humidity data are adjusted by comparing the real-time data collected by the multi-source sensors with the temperature control strategy to obtain the temperature and humidity data.

[0016] S5. Predict crop growth based on temperature and humidity data;

[0017] By combining temperature and humidity data, a deep learning model is used to predict crop growth, and crop growth prediction results are obtained.

[0018] S6. Intelligent irrigation management based on crop growth prediction results;

[0019] Based on crop growth forecasts, irrigation volume and timing are dynamically adjusted to ensure that the crops receive the required water, thus yielding irrigation adjustment results.

[0020] S7. Optimize greenhouse energy efficiency based on irrigation adjustment results;

[0021] Based on the irrigation adjustment results, the greenhouse energy efficiency is optimized using a reinforcement learning algorithm to obtain the energy efficiency optimization results;

[0022] S8. Decision support for integrated greenhouse management based on energy efficiency optimization results;

[0023] Based on the energy efficiency optimization results, decision support is provided for the comprehensive management of greenhouses, and a comprehensive report is generated.

[0024] To further optimize this technical solution, the environmental data inside the greenhouse in step S1 includes:

[0025] Temperature, humidity, air velocity, light intensity, and soil moisture. These data were collected using multi-source sensors.

[0026] To further optimize this technical solution, the deep learning model in step S5 is based on:

[0027] The multi-input neural network uses a weighted fusion mechanism to process and fuse five environmental parameters: temperature, humidity, air velocity, light intensity, and soil moisture.

[0028] To further optimize this technical solution, the construction steps of the multi-input neural network are as follows:

[0029] First, obtain data on five environmental factors from environmental data collected by multi-source sensors, as well as temperature and humidity data:

[0030] Temperature, in degrees Celsius;

[0031] Humidity, in % %

[0032] Air velocity, measured in m / s;

[0033] Light intensity, measured in lux;

[0034] Soil moisture, in percentages (%)

[0035] Then, a multi-input neural network is constructed, including an input layer, a hidden layer, and an output layer;

[0036] Input layer:

[0037] Dynamic weights of each input environmental factor The system dynamically adjusts based on the current environmental conditions, and then constructs the input layer formula:

[0038] , ;

[0039] in,

[0040] Represents the treatment function for each environmental factor;

[0041] It is the dynamic weight of each environmental factor;

[0042] These are the raw input data for environmental factors;

[0043] It is a weighted input for each environmental factor;

[0044] Hidden layer:

[0045] The input features obtained from the input layer are fused using multiple hidden layers, as shown in the formula:

[0046] ;

[0047] in,

[0048] It represents the weight of each environmental factor at each node in the hidden layer;

[0049] It is a preset bias term, which is the amount of translation of the model on the x-axis;

[0050] It is the output of the hidden layer;

[0051] Indicates the number of nodes in the hidden layer;

[0052] Output layer:

[0053] Map the output of the hidden layer to crop growth predictions. In the formula, it is:

[0054] ;

[0055] in,

[0056] This is the model's final output, i.e., the growth prediction value;

[0057] It is the sigmoid function, the activation function of the output layer, and its mathematical expression is: ;

[0058] It represents the weight of each node in the hidden layer;

[0059] It is the bias term of the output layer, the amount of translation of the model on the x-axis, which is set manually.

[0060] To further optimize this technical solution, the reinforcement learning algorithm in step S7 includes:

[0061] State space and action space;

[0062] Reward function;

[0063] Policy function;

[0064] Update strategy.

[0065] To further optimize this technical solution, the state space is:

[0066] Multiple environmental factors in the greenhouse and the growth status of the crops at each time step The following is represented as a vector:

[0067] ;

[0068] in,

[0069] For time The temperature at that time, in degrees Celsius;

[0070] For time Humidity at that time, in % %

[0071] For time The air velocity at that time is expressed in m / s.

[0072] For time The intensity of light at that time, measured in lux;

[0073] For time Soil moisture at that time, in % %.

[0074] For time Current crop growth status;

[0075] is a state space vector.

[0076] To further optimize this technical solution, the action space represents:

[0077] All possible operations, including temperature regulation, humidity regulation, and irrigation volume adjustment;

[0078] At each time step Construct a combined control strategy, represented as a vector:

[0079] ;

[0080] in,

[0081] For temperature control actions, increasing or decreasing the temperature;

[0082] This is a humidity control action, increasing or decreasing humidity;

[0083] This is an irrigation control action, used to start or stop irrigation;

[0084] This is the action space vector.

[0085] To further optimize this technical solution, the reward function includes:

[0086] The impact of state space and action space on greenhouse energy efficiency and crop growth;

[0087] reward function At each time step The formula is:

[0088] ;

[0089] in,

[0090] For the current time step The energy efficiency gain is the amount of energy saved;

[0091] For the current time step of the crop The amount of change in growth status;

[0092] The preset target temperature and humidity;

[0093] These are the preset weighting coefficients.

[0094] To further optimize this technical solution, the strategy function is:

[0095] Determine the optimal action space to be taken in the state space; the policy function is equal to , The value update formula is:

[0096] ;

[0097] in,

[0098] Current state and of The value represents the worth of that state-action pair;

[0099] The update step size is controlled by the preset learning rate;

[0100] A preset discount factor is used to balance the importance of current and future rewards, ranging from 0 to 1;

[0101] Indicates the next state Below, the optimal action value.

[0102] To further optimize this technical solution, the update strategy includes:

[0103] As training progresses, The values ​​are continuously updated, and the policy function tends towards the optimal policy. In each interaction, the reinforcement learning model selects the optimal action space based on the current state and the learned policy. To adjust the operating status of the greenhouse equipment and based on the instant reward function. Update decision-making strategies.

[0104] In a second aspect, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, they implement the steps of the intelligent greenhouse temperature control method based on multi-source sensors as described in the first aspect of the present invention.

[0105] Thirdly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, they implement the steps of the intelligent greenhouse temperature control method based on multi-source sensors as described in the first aspect of the present invention.

[0106] Compared with existing technologies, this invention provides an intelligent greenhouse temperature control method based on multi-source sensors, which has the following beneficial effects:

[0107] This intelligent greenhouse temperature control method based on multi-source sensors effectively overcomes the lack of physically meaningful conversion mechanisms in existing technologies by setting an improved method for converting growth prediction values ​​output by deep learning models and designing a reward function for reinforcement learning algorithms.

[0108] By normalizing, targeting, or weighting the growth predictions output by deep learning models, the accuracy of linking crop growth status with greenhouse environmental control is improved compared to existing technologies.

[0109] By employing an innovative reinforcement learning reward function design that fully considers the nonlinear relationship between multi-source sensor data and crop growth, the greenhouse control strategy is optimized compared to existing technologies, thereby improving energy efficiency while ensuring healthy crop growth. Attached Figure Description

[0110] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0111] Figure 1 This is a schematic flowchart of an intelligent greenhouse temperature control method based on multi-source sensors proposed in this invention.

[0112] Figure 2 This is a schematic diagram of the reinforcement learning algorithm for an intelligent greenhouse temperature control method based on multi-source sensors proposed in this invention.

[0113] Figure 3 This is a schematic diagram of the weighted fusion mechanism of an intelligent greenhouse temperature control method based on multi-source sensors proposed in this invention. Detailed Implementation

[0114] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0115] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0116] Secondly, the term "an embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places throughout this specification does not necessarily refer to the same embodiment, nor is it a single embodiment or an embodiment selectively excluded from other embodiments.

[0117] Example 1:

[0118] Reference Figures 1-3 This is the first embodiment of the present invention, which provides a smart greenhouse temperature control method based on multi-source sensors, including the following steps:

[0119] S1, Data Acquisition;

[0120] Environmental data inside the greenhouse is collected by multi-source sensors and cleaned using data preprocessing techniques to obtain the collected data.

[0121] In this embodiment, environmental data inside the greenhouse, including temperature, humidity, air velocity, light intensity, and soil moisture, is collected using multi-source sensors. Data acquisition is performed simultaneously by multiple sensors to ensure data diversity and accuracy. The data collected by the sensors is input into a preprocessing module for preliminary data cleaning, missing value imputation, and noise removal. The raw sensor data undergoes time-series synchronous processing to ensure that subsequent analysis accurately reflects the environmental state at each moment. This step relies on data preprocessing techniques, and during data cleaning, methods such as interpolation or Kalman filtering are used to remove sensor errors and noise.

[0122] The key to data acquisition and preprocessing lies in ensuring the seamless integration of data collected from multiple sensors, enabling accurate modeling and analysis in subsequent steps. In practical applications, different types of sensors may have different sampling frequencies and data accuracies; therefore, time-series synchronization ensures that all data are aligned along the same timeline. Furthermore, preprocessing eliminates invalid data or outliers, which is crucial for model training and accurate prediction. Ultimately, the preprocessed data serves as the foundation for further analysis and modeling in subsequent steps.

[0123] S2. Perform preliminary temperature prediction based on the collected data;

[0124] Preliminary temperature predictions were made based on the collected data using time series analysis techniques, and the temperature prediction results were obtained.

[0125] In this embodiment, time series analysis is used for preliminary temperature prediction. Time series data is indispensable in greenhouse environment monitoring, especially since temperature fluctuations often exhibit periodicity and regularity. By analyzing the trends in historical temperature data, the model can predict future temperature changes and provide a reference for subsequent temperature regulation. Using the ARIMA (Autoregressive Integral Moving Average) model for temperature prediction has the advantage of capturing the time dependence and seasonality of temperature.

[0126] ARIMA models are a mature and commonly used technique in time series forecasting. They can analyze the trends and seasonal components in temperature data through methods such as autoregression, differencing, and moving averages. In practical applications, greenhouse temperatures are influenced by factors such as external weather changes and airflow within the greenhouse, exhibiting strong periodic characteristics. ARIMA models can provide relatively accurate short-term temperature forecasts, thus offering a reference for temperature regulation and avoiding the adverse effects of temperature fluctuations on crop growth.

[0127] S3. Obtain a temperature control strategy based on the temperature prediction results;

[0128] Based on fuzzy control, a temperature control strategy is designed according to the temperature prediction results, and the temperature control strategy is obtained.

[0129] Fuzzy control technology is a control method based on fuzzy logic, widely used to handle complex, nonlinear, and uncertain systems. Unlike traditional precise control methods (such as PID control), fuzzy control achieves system regulation and control by processing fuzzy and imprecise information, making it particularly suitable for systems that cannot be simply described by mathematical models. Its basic idea is to simulate expert judgment using fuzzy sets and fuzzy rules, thereby achieving intelligent control of the system.

[0130] Temperature control strategy: In a greenhouse environment, the task of the strategy is to maintain a suitable temperature for crop growth. Since temperature is affected by a variety of factors (such as external weather, greenhouse equipment, and the crop's own heat release), and the changes in these factors are often difficult to model precisely, fuzzy control becomes a very effective temperature control strategy.

[0131] The temperature control strategy adjusts heating, ventilation, and shading equipment in the greenhouse based on the deviation between the real-time temperature and the set target temperature, as well as the rate of temperature change. The specific steps of the temperature control strategy are as follows:

[0132] Temperature deviation: The temperature control system first calculates the deviation between the current temperature and the target temperature. ,in For the target temperature, This is the current temperature.

[0133] Rate of temperature change: Then calculate the rate of temperature change. This refers to the change in temperature between the previous time step and the current time step.

[0134] in, This represents the temperature at the previous time step.

[0135] Fuzzy rules: Based on temperature deviation and rate of change, the control system performs inference using a predefined fuzzy rule base. For example:

[0136] If temperature deviation Large and temperature change rate Soon, stronger heating or cooling will be required.

[0137] If temperature deviation Very small, and the rate of change If the temperature is also relatively low, then maintain the current temperature.

[0138] These rules dynamically adjust the equipment's operating strategy based on specific temperature changes.

[0139] Control output: Based on the results of fuzzy inference, the defuzzification process converts the output fuzzy control quantity into specific equipment control commands (such as heater power, fan speed, and the opening and closing degree of the shade net).

[0140] Fuzzy control technology transforms temperature deviations and rates of change into control commands through fuzzification, inference, and defuzzification processes, which are then used to regulate environmental equipment within greenhouses. This method is well-suited for the complex temperature regulation problems in greenhouse environments, handling uncertainties and hysteresis to achieve highly efficient automated control.

[0141] In this embodiment, considering the inherent lag and uncertainty in temperature changes inside the greenhouse, fuzzy control allows for flexible adjustment of the control strategy. Based on the predicted temperature change trend, precise temperature control is achieved by adjusting heating equipment, fans, shading nets, and other facilities.

[0142] Fuzzy control is a widely used control method in complex, nonlinear systems. Greenhouse temperature control systems typically need to respond to changes in multiple environmental parameters, which often do not perfectly conform to linear laws. Therefore, fuzzy control provides an effective imprecise control method. By setting fuzzy linguistic variables such as "high," "medium," and "low" for temperature, and fuzzy rules for temperature deviation and rate of change, the system can dynamically adjust its operating strategy to achieve stable temperature control.

[0143] S4. Adjust the greenhouse environment according to the temperature control strategy;

[0144] The temperature and humidity data are adjusted by comparing the real-time data collected by the multi-source sensors with the temperature control strategy to obtain the temperature and humidity data.

[0145] In this embodiment, the temperature and humidity inside the greenhouse are precisely controlled by driving devices (such as fans, heaters, humidity regulators, etc.). The system compares real-time sensor data (such as feedback data from temperature and humidity sensors) with the set temperature control strategy, thereby adjusting the operating status of the equipment to ensure that the environment inside the greenhouse is maintained within the optimal range.

[0146] The core task is to implement temperature control strategies onto specific hardware devices to ensure that the temperature and humidity inside the greenhouse are maintained within an ideal range. Through a real-time feedback control mechanism, the system can dynamically adjust the operating mode of the equipment based on environmental parameters detected by sensors. For example, when the temperature inside the greenhouse is too high, the fans will automatically accelerate to help dissipate heat; when the humidity is too low, the humidity control equipment will activate to maintain suitable air humidity. The entire adjustment process is based on closed-loop control using real-time data, thereby ensuring the stability of the internal environment of the greenhouse.

[0147] S5. Predict crop growth based on temperature and humidity data;

[0148] By combining temperature and humidity data, a deep learning model is used to predict crop growth, resulting in crop growth predictions.

[0149] In this embodiment, the obtained temperature and humidity data are combined to use a deep learning model to predict crop growth. Crop growth is affected by various environmental factors such as temperature, humidity, and light, and the environmental control within a greenhouse can optimize these factors. By training a multi-input deep neural network (DNN), the model can predict the growth rate, health status, and yield of crops under different environmental conditions. Deep learning not only predicts single variables (such as temperature) but also combines multiple environmental factors for comprehensive prediction.

[0150] Deep learning models learn the complex relationship between environmental conditions and crop growth by inputting data from multiple sensor sources. In this process, the network can automatically extract nonlinear features from the data and perform efficient feature learning and prediction through the combination of multiple neural network layers. Temperature and humidity are two important factors affecting crop growth, but predictions based on a single variable may not be sufficient to fully reflect the crop's growth status. Therefore, deep learning, by integrating multiple factors, can more accurately assess the crop's growth status, providing growers with scientific decision support.

[0151] The deep learning model is based on a multivariate deep neural network (MDNN), which can effectively process and fuse multiple environmental parameters (temperature, humidity, air velocity, light intensity, and soil moisture), and extract the complex relationship between these factors and crop growth through feature fusion and weighting mechanisms. The core of this model is the weighted fusion mechanism, which can dynamically adjust the weights of each input factor to better adapt to crop growth patterns under different environmental conditions.

[0152] In conventional deep learning models, all input features are typically treated equally, and the model automatically learns the relationships between these features. However, in real-world applications, different environmental factors have varying degrees of impact on crop growth. For example, under certain seasonal or climatic conditions, light intensity may be more important than temperature, and vice versa. Therefore, we introduce a dynamic weighting mechanism based on reinforcement learning. The model can automatically adjust the weights of different input features according to the current environmental state (e.g., the current temperature and humidity combination), thereby improving the accuracy of model predictions. This is the environmental factor weight adjustment mechanism.

[0153] This model also employs some concepts from Convolutional Neural Networks (CNNs), embedding them into the deep learning network to extract local features from the input data. By using convolutional operations in the input layer, we are able to capture local correlations between environmental variables (such as the relationship between temperature and humidity), and these local features are further learned and fused in subsequent layers, thereby optimizing the overall prediction of crop growth. This is the multi-layer convolutional structure.

[0154] The construction steps for a multi-input neural network are as follows:

[0155] First, data on five environmental factors were obtained from the environmental data inside the greenhouse, including:

[0156] Temperature, in degrees Celsius;

[0157] Humidity, in % %

[0158] Air velocity, measured in m / s;

[0159] Light intensity, measured in lux;

[0160] Soil moisture, in percentages (%)

[0161] Then, a multi-input neural network is constructed, including an input layer, a hidden layer, and an output layer;

[0162] Input layer:

[0163] Define a weighted fusion mechanism to dynamically assign weights to each input environmental factor. The system dynamically adjusts based on the current environmental conditions. Then, the input layer formula is constructed:

[0164] , ;

[0165] in,

[0166] The REUL function represents the treatment function for each environmental factor; when When the value is greater than 0, the output value equals the input value; when... When the value is less than or equal to 0, the output value is 0.

[0167] It is the dynamic weight of each environmental factor;

[0168] These are the raw input data for environmental factors;

[0169] It is a weighted input for each environmental factor;

[0170] Hidden layer:

[0171] Each input feature obtained from the input layer is fused using multiple hidden layers, as shown in the formula:

[0172] ;

[0173] in,

[0174] It represents the weight of each environmental factor at each node in the hidden layer;

[0175] It is the bias term, the amount of translation of the model on the x-axis, which is set manually;

[0176] It is the output of the hidden layer;

[0177] Indicates the number of nodes in the hidden layer;

[0178] Output layer:

[0179] Map the output of the hidden layer to crop growth predictions. In the formula, it is:

[0180] ;

[0181] in,

[0182] This is the model's final output, i.e., the growth prediction value;

[0183] It is the sigmoid function, the activation function of the output layer, and its mathematical expression is: ;

[0184] It represents the weight of each node in the hidden layer;

[0185] It is the bias term of the output layer, which allows the model to be translated arbitrarily on the x-axis to better fit the data.

[0186] The application of this model can be explained as follows:

[0187] Input data: The model's input includes environmental factors such as temperature, humidity, air velocity, light intensity, and soil moisture. This data is collected in real-time by sensors and serves as input to the deep neural network. Each input environmental factor is processed using a dynamic weighting mechanism based on real-time environmental conditions and historical data, assigning it different weights. In this way, the model can automatically adjust the importance of input features according to the actual situation.

[0188] Weighting mechanism: At each time step, the reinforcement learning model automatically adjusts according to the current environmental state. The value of this value allows the system to optimize the model's predictive ability based on changing environmental conditions. In this way, the model can dynamically adapt to the impact of seasonal changes, weather changes, and other factors on crop growth.

[0189] Model Training: The training process of deep learning models uses labeled crop growth data for supervised learning. By optimizing the objective function (such as minimizing the mean squared error), the model learns the optimal parameter and weight values. During training, the reinforcement learning module automatically adjusts the weights of the input features based on the prediction error.

[0190] Predictive Output: The trained model can predict crop growth status (such as growth rate, health status, or yield) based on real-time sensor data. This prediction can provide decision support for greenhouse managers, helping them adjust greenhouse environmental conditions according to crop needs.

[0191] S6. Intelligent irrigation management based on crop growth prediction results;

[0192] Based on crop growth prediction results, fuzzy logic control (FLC) technology is used to further optimize the intelligent irrigation management strategy and obtain irrigation adjustment results.

[0193] In this embodiment, irrigation volume and timing are dynamically adjusted by predicting crop growth needs and environmental change trends to ensure that the crop's water requirements are met at different growth stages. FLC technology is employed to adjust the irrigation system's operating status based on real-time feedback from crop growth and soil moisture.

[0194] Irrigation is a crucial task in greenhouse management; too much or too little water can negatively impact crop growth. Traditional irrigation management often relies on fixed times or preset water volumes, while intelligent irrigation systems can automatically adjust based on environmental conditions and crop needs. In this process, based on crop growth predictions, the system uses fuzzy logic algorithms to adjust the irrigation amount, ensuring that the crop's water supply is always at an ideal level. Through real-time feedback adjustments, water resource utilization efficiency can be significantly improved, and waste reduced.

[0195] S7. Optimize greenhouse energy efficiency based on irrigation adjustment results;

[0196] Based on the irrigation adjustment results, the greenhouse energy efficiency is optimized using a reinforcement learning algorithm to obtain the energy efficiency optimization results.

[0197] In this embodiment, a reinforcement learning algorithm is introduced to optimize the greenhouse's energy efficiency based on irrigation adjustment results. Greenhouse energy efficiency optimization involves temperature control and humidity regulation. Reinforcement learning allows for continuous adjustment of control strategies to minimize energy consumption while maintaining the environmental conditions necessary for crop growth.

[0198] Reinforcement learning algorithms optimize control decisions through interaction with the environment. In greenhouse energy efficiency optimization, strategies are adjusted based on crop needs. The system must not only consider the current environmental conditions but also make forward-looking adjustments based on future environmental predictions to achieve minimum energy consumption and optimal crop growth. The advantage of reinforcement learning lies in its ability to autonomously explore optimal control strategies through continuous trial and feedback, thereby significantly improving the energy efficiency of greenhouse management while ensuring crop growth.

[0199] Reinforcement learning algorithms include:

[0200] State space. Multiple environmental factors in the greenhouse and the crop's growth status at each time step. The following is represented as a vector:

[0201] ;

[0202] in,

[0203] For time The temperature at that time, in degrees Celsius;

[0204] For time Humidity at that time, in % %

[0205] For time The air velocity at that time is expressed in m / s.

[0206] For time The intensity of light at that time, measured in lux;

[0207] For time Soil moisture at that time, in % %.

[0208] For time Current crop growth status;

[0209] The conversion formula is:

[0210] ;

[0211] in,

[0212] It is the output of the deep learning model, representing the crop at time step The predicted growth value.

[0213] and It is the current time step Crop growth forecast Minimum and maximum values ​​can be obtained through historical data or experiments.

[0214] Then, the crop growth state is transformed.

[0215] Normalized growth prediction values As This is used to represent the growth status of crops at a certain time step.

[0216] To facilitate the learning of reinforcement learning models, It can be used as a scalar state value for crop growth to calculate rewards. For example, the following two methods can be defined for transformation. Growth state value:

[0217] Targeting (discretization): Distributing the growth state The mapping is to high or low reward, depending on whether the ideal crop growth state is achieved. We set a threshold. ,when When the value exceeds the threshold, it indicates that the crop is in an ideal growth state and a high reward is given; otherwise, a low reward is given.

[0218] reward function Defined as:

[0219] ;

[0220] This approach helps reinforcement learning models understand whether crops are in an ideal growth state.

[0221] Weighted fusion processing (multi-dimensional fusion): If It is a vector composed of multiple growth dimensions (such as growth rate, health status, expected yield, etc.). A weighted fusion method can be used to synthesize the growth prediction values ​​of these dimensions into a comprehensive growth status value. The specific weighted fusion formula is as follows:

[0222] ;

[0223] in,

[0224] These represent the crop at time steps. The growth rate, health status, and expected yield of the plant.

[0225] The weighting coefficients represent the contributions of different dimensions to the crop growth status.

[0226] is a state space vector.

[0227] Action space. Represents all executable operations, including temperature regulation, humidity regulation, and irrigation volume adjustment.

[0228] At each time step Construct a combined control strategy, represented as a vector:

[0229] ;

[0230] in,

[0231] For temperature control actions, 1 represents increasing temperature and 0 represents decreasing temperature.

[0232] For humidity control actions, an increase is 1 and a decrease is 0.

[0233] For irrigation control actions, 1 is for starting and 0 is for stopping;

[0234] This is the action space vector.

[0235] Reward function. The impact of state space and action space on greenhouse energy efficiency and crop growth.

[0236] reward function At each time step The formula is:

[0237] ;

[0238] in,

[0239] For the current time step Internal energy efficiency gain refers to the amount of energy saved.

[0240] For the current time step Changes in the growth status of crops within the plant;

[0241] The preset target temperature and humidity;

[0242] These are the preset weighting coefficients.

[0243] Policy function. Determines the optimal action space to be taken within the state space; the policy function is equal to... , The value update formula is:

[0244] ;

[0245] in,

[0246] Current state and of The value represents the worth of that state-action pair;

[0247] The preset learning rate controls the update step size, which ranges from 0 to 1;

[0248] A preset discount factor is used to balance the importance of current and future rewards, ranging from 0 to 1;

[0249] Indicates the next state Below, optimal action space value.

[0250] Update the policy. As training progresses, the Q-value is continuously updated, and the policy function... The reinforcement learning model tends towards the optimal policy. In each interaction, the model selects the optimal action space based on the current state and the learned policy. To adjust the operating status of the greenhouse equipment and based on the instant reward function. Update decision-making strategies.

[0251] Steps for using the model:

[0252] Initialization: First, the initial environmental state of the greenhouse. Given data collected by sensors, and based on the initial state, the reinforcement learning algorithm randomly selects an action from the action space. .

[0253] Interaction process: At each time step, the model adjusts its parameters according to the current state. Choose an action (Based on the current strategy) Then, based on the actions performed, the environment state transitions. And calculate the reward .

[0254] Q-value update: based on instant rewards And the maximum Q-value of the next state. As training progresses, the Q-value tends to be optimal, thus gradually optimizing the policy.

[0255] Energy efficiency optimization: Through repeated interactions, the model learns the optimal equipment operation strategy, which maximizes greenhouse energy efficiency at each time step and ensures the growth status of crops.

[0256] In the model:

[0257] Calculate the change in crop growth status ,

[0258] Changes in crop growth status Represents the time step and The change in crop growth between [a certain period]. To calculate [the change in crop growth between different periods]. Using the current crop growth status values Crop growth status at the previous time step Differences between them:

[0259] ;

[0260] in,

[0261] It is the change in crop growth, representing the crop's growth at time step. and Growth changes between them.

[0262] and These are the crop growth status values ​​at the current time step and the previous time step, respectively.

[0263] This change It can be used as a feedback signal for reinforcement learning models to adjust the operating strategies of greenhouse equipment (such as heaters, irrigation systems, etc.), thereby achieving the goal of optimizing crop growth and greenhouse energy efficiency.

[0264] S8. Decision support for integrated greenhouse management based on energy efficiency optimization results;

[0265] Based on the energy efficiency optimization results, decision support is provided for the comprehensive management of greenhouses, and a comprehensive report is generated.

[0266] In this embodiment, by combining data on temperature, humidity, light intensity, soil moisture, crop growth, and energy efficiency optimization results, the system generates a comprehensive report to help managers make more scientific management decisions. The report includes energy efficiency indicators, crop growth forecasts, and temperature and humidity control recommendations, assisting greenhouse managers in achieving refined management.

[0267] The integrated management decision support system for greenhouses provides a one-stop decision support tool by integrating various data and optimization results. In practical applications, managers often face complex decision-making problems and need to adjust greenhouse management strategies based on multiple factors. This system summarizes the output results of each module, generates detailed reports, and provides specific operational suggestions. Ultimately, the system helps managers make more scientific and efficient management decisions under different growth cycles and climate change conditions.

[0268] Example 2:

[0269] This embodiment also provides a computer device applicable to a smart greenhouse temperature control method based on multi-source sensors, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the smart greenhouse temperature control method based on multi-source sensors as proposed in the above embodiment.

[0270] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements an intelligent greenhouse temperature control method based on multi-source sensors as proposed in the above embodiments.

[0271] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0272] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part 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 of the various embodiments of this invention. 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.

[0273] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0274] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0275] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0276] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for intelligent greenhouse temperature control based on multi-source sensors, characterized in that, Includes the following steps: S1, Data Acquisition; Environmental data inside the greenhouse is collected using multi-source sensors to obtain the collected data; S2. Perform preliminary temperature prediction based on the collected data; Preliminary temperature predictions are made based on the collected data to obtain temperature prediction results; S3. Obtain a temperature control strategy based on the temperature prediction results; Based on fuzzy control, a temperature control strategy is designed using the temperature prediction results, and the temperature control strategy is derived. S4. Adjust the greenhouse environment according to the temperature control strategy; The temperature and humidity data are adjusted by comparing the real-time data collected by the multi-source sensors with the temperature control strategy to obtain the temperature and humidity data. S5. Predict crop growth based on temperature and humidity data; By combining temperature and humidity data, a deep learning model is used to predict crop growth, and crop growth prediction results are obtained. S6. Intelligent irrigation management based on crop growth prediction results; Based on crop growth forecasts, irrigation volume and timing are dynamically adjusted to ensure that the crops receive the required water, thus yielding irrigation adjustment results. S7. Optimize greenhouse energy efficiency based on irrigation adjustment results; Based on the irrigation adjustment results, the greenhouse energy efficiency is optimized using a reinforcement learning algorithm to obtain the energy efficiency optimization results; The reinforcement learning algorithm includes: a state space and an action space, a reward function, a policy function, and a policy update function; wherein, the reward function includes: The impact of state space and action space on greenhouse energy efficiency and crop growth; reward function At each time step The formula is: ； in, For the current time step Internal energy efficiency gain refers to the amount of energy saved. For the current time step of the crop Changes in internal growth state; The preset target temperature and humidity; These are preset weighting coefficients; The strategy function is: Determine the optimal action space to be taken in the state space; the policy function is equal to , The formula for updating the value is: ； in, Current state and of Value, representing the worth of a state-action sequence; The preset learning rate controls the update step size, which ranges from 0 to 1; A preset discount factor is used to balance the importance of current and future rewards, ranging from 0 to 1; Indicates the next state Below, the optimal action value; It is a state space vector; For action space vectors; S8. Decision support for integrated greenhouse management based on energy efficiency optimization results; Based on the energy efficiency optimization results, decision support is provided for the comprehensive management of greenhouses, and a comprehensive report is generated.

2. The intelligent greenhouse temperature control method based on multi-source sensors according to claim 1, characterized in that, The environmental data inside the greenhouse in step S1 includes: Temperature, humidity, air velocity, light intensity, and soil moisture; data were collected through multi-source sensors.

3. The intelligent greenhouse temperature control method based on multi-source sensors according to claim 1, characterized in that, The deep learning model in step S5 is based on: The multi-input neural network uses a weighted fusion mechanism to process and fuse five environmental parameters: temperature, humidity, air velocity, light intensity, and soil moisture.

4. The intelligent greenhouse temperature control method based on multi-source sensors according to claim 3, characterized in that, The construction steps of the multi-input neural network are as follows: First, five environmental factor data are obtained from environmental data collected by multi-source sensors, as well as temperature and humidity data, including: temperature ,humidity air velocity Light intensity Soil moisture ; Then, a multi-input neural network is constructed, including an input layer, a hidden layer, and an output layer; Input layer: Dynamic weights of each input environmental factor The system dynamically adjusts based on the current environmental conditions, and then constructs the input layer formula: ， ； in, The REUL function represents the treatment function for each environmental factor; when When the value is greater than 0, the output value equals the input value; when... When the value is less than or equal to 0, the output value is 0; It is the dynamic weight of each environmental factor; These are the raw input data for environmental factors; It is a weighted input for each environmental factor; Hidden layer: Each input feature obtained from the input layer is fused using multiple hidden layers, as shown in the formula: ； in, It represents the weight of each environmental factor at each node in the hidden layer; It is a preset bias term, which is the amount of translation of the model on the x-axis; It is the output of the hidden layer; Indicates the number of nodes in the hidden layer; Output layer: Map the output of the hidden layer to crop growth predictions. In the formula, it is: ； in, This is the model's final output, i.e., the growth prediction value; It is the sigmoid function, the activation function of the output layer, and its mathematical expression is: ; It represents the weight of each node in the hidden layer; It is the bias term of the output layer, which allows the model to be translated arbitrarily on the x-axis to better fit the data.

5. The intelligent greenhouse temperature control method based on multi-source sensors according to claim 1, characterized in that, The state space includes: Multiple environmental factors in the greenhouse and the growth status of the crops at each time step The following is represented as a vector: ； in, , , , , , , respectively time Temperature, humidity, air velocity, light intensity, soil moisture, and crop growth status at that time.

6. The intelligent greenhouse temperature control method based on multi-source sensors according to claim 1, characterized in that, The action space includes: All possible operations, including temperature regulation, humidity regulation, and irrigation volume adjustment; At each time step Construct a combined control strategy, represented as a vector: ； in, For temperature control actions, 1 represents increasing temperature and 0 represents decreasing temperature. For humidity control actions, an increase is 1 and a decrease is 0. This is for irrigation control actions; 1 represents starting and 0 represents stopping.

7. The intelligent greenhouse temperature control method based on multi-source sensors according to claim 1, characterized in that, The update strategy includes: As training progresses, The values ​​are continuously updated, and the policy function tends towards the optimal policy. In each interaction, the reinforcement learning model selects the optimal action space based on the current state and the learned policy. To adjust the operating status of the greenhouse equipment and based on the instant reward function. Update decision-making strategies.

Images