Greenhouse air conditioning regulation method and system based on deep learning and electronic equipment

Abstract

The application discloses a greenhouse air conditioner regulation and control method and system based on deep learning and electronic equipment, and the method comprises the following steps: S1, collecting data in a predetermined time period, the data comprising greenhouse internal and external environmental factor data, environmental control equipment operation data and crop physiological and ecological data; S2, constructing a combined prediction model of the load of the air conditioning equipment in the greenhouse and the temperature of the crop leaves according to the data; S3, inputting the real-time collected data into the combined prediction model to obtain real-time prediction values of the load of the air conditioning equipment and the temperature of the crop leaves; S4, constructing a data set of the air conditioning equipment according to the data information of the air conditioning equipment and the load of the air conditioning equipment, selecting data samples of the data information of the air conditioning equipment that meet the suitable value range of crop growth and have the minimum load of the air conditioning equipment in the real-time prediction values, and forming an optimal energy-saving regulation and control strategy of the air conditioning equipment. The greenhouse air conditioner regulation and control method can realize intelligent regulation and control of the temperature in the greenhouse.

Description

Technical Field

[0001] This application relates to the field of smart agriculture technology, and mainly applies for a method for controlling air conditioning in a greenhouse, an intelligent control system for air conditioning in a greenhouse, a computer storage medium, and electronic equipment. Background Technology

[0002] With the rapid development of facility agriculture, greenhouses, as an important means of realizing facility agricultural production, have a typical microclimate environment inside. They are affected by the internal and external environments and have characteristics such as strong interference, multiple couplings, and large lag.

[0003] Temperature, as a crucial environmental factor in crop growth, directly determines the crop's growth and development; unsuitable temperatures can lead to significant yield reductions or even crop failure. However, different crops require different optimal temperatures. With the rapid changes in outdoor weather conditions, it is difficult to achieve intelligent temperature control in greenhouses using only ordinary equipment such as fans and evaporative cooling pads. Combining these with other precision equipment has become a new trend in intelligent greenhouse environmental control. In small and medium-sized greenhouses, using air conditioning equipment to regulate the temperature has become an effective means of intelligent greenhouse environmental control.

[0004] However, current greenhouse air conditioning equipment suffers from problems such as low precision, delayed effects, and severe resource consumption, limiting its widespread application in agricultural production. Therefore, achieving precise dynamic control of air conditioning equipment has become a key issue for refined greenhouse management and energy conservation in facility production. Summary of the Invention

[0005] In view of this, this application provides a method, system, computer storage medium, and electronic device for controlling air conditioning in a greenhouse, which can solve the technical problem of high energy consumption caused by the automatic control of air conditioning equipment in the prior art.

[0006] To address the aforementioned technical problems, this application provides a method for controlling air conditioning in a greenhouse, comprising the following steps: S1, collecting data within a predetermined time period, the data including environmental factor data inside and outside the greenhouse, operational data of environmental control equipment, and crop physiological and ecological data; S2, constructing a combined prediction model of the load of the air conditioning equipment and the temperature of crop leaves based on the data; S3, inputting the real-time collected data into the combined prediction model to obtain real-time predicted values ​​of the load of the air conditioning equipment and the temperature of the crop leaves; S4, constructing a dataset of the air conditioning equipment based on the data information and load of the air conditioning equipment, selecting data samples of the air conditioning equipment data information that meet the suitable range for crop growth and have the lowest load of the air conditioning equipment from the real-time predicted values ​​to form an optimal energy-saving control strategy for the air conditioning equipment.

[0007] According to one embodiment of this application, in step S2, the data is cleaned, and the cleaned data is classified and integrated into a heating dataset and a cooling dataset according to the temperature control type.

[0008] According to one embodiment of this application, step S2 includes: S21, dividing the data into a heating dataset and a cooling dataset according to the temperature control type and air conditioning usage scenario, and further dividing the heating dataset and the cooling dataset into a training set and a test set, respectively; S22, constructing a single prediction model based on the training set, and obtaining predicted values ​​of the load of the air conditioning equipment and the temperature of the crop leaves based on the single prediction model; S23, generating weight coefficient values ​​of the single prediction model at different times to obtain a time-varying weighted combined prediction model; S24, selecting the test set to test the time-varying weighted combined prediction model to obtain the combined prediction model.

[0009] According to one embodiment of this application, in step S22, the single prediction model is constructed based on the random forest algorithm and the gated recurrent unit neural network; in step S23, the weight coefficient values ​​of the single prediction model at different times are generated by the sparrow search algorithm.

[0010] According to one embodiment of this application, the fitness function of the sparrow search algorithm is set as the mean square error of the combined prediction model at time i: (1), In equation (1), T represents the total number of moments. For a moment, For the first Actual value at any time For the first The output value of the random forest model at each time step. For the first The output value of the time-gated recurrent unit neural network model. For the first The weight coefficients of the RF model at time step [time]. For the first Weight coefficients of the time-gated recurrent unit neural network model; Step S23 includes: S231. Update the discoverer's location. (2); + ( ) (3); In equation (2), t represents the current iteration number; The position of the sparrow; A uniformly random number between (0, 1); Represents a range vector; This is the warning value; This is a safe value; In formula (3) , This represents the final value of the weight. This represents the maximum number of iterations. S232, Update follower positions. (4); In equation (4) express Worst position; This represents the position of the sparrow with the best fitness in the (t+1)th iteration; Represents a matrix; S233, Update the location of the vigilant. (5); In equation (4) This represents the optimal position of the sparrow in the t-th round; This represents the position of the sparrow with the best fitness in the (t+1)th iteration; This represents the worst position of the sparrow in the t-th round; K is a random number, taking a value between -1 and 1. These are the step size control parameters; To obtain sufficiently small numbers, avoid denominators of 0; and These represent the sparrow's fitness, worst fitness, and best fitness values, respectively.

[0011] According to one embodiment of this application, the data information of the air conditioning equipment includes the number of air conditioning units that are started and stopped and their placement location.

[0012] Secondly, embodiments of this application provide an intelligent air conditioning control system for greenhouses, used to implement the method described in this application, comprising: a data monitoring module, which collects data including environmental factor data inside and outside the greenhouse, operation data of environmental control equipment, and crop physiological and ecological data; a cloud platform service module, which analyzes the data collected by the data monitoring module and generates an optimal air conditioning energy-saving control strategy; and an air conditioning equipment control execution module, which performs automatic control operations on the air conditioning equipment according to the optimal air conditioning energy-saving control strategy.

[0013] According to one embodiment of this application, the cloud platform service module includes: a data storage and cleaning module, which can store and clean the data; and a data analysis and decision-making module, which constructs a combined prediction model of the load of the air conditioning equipment and the temperature of the crop leaves using a random forest algorithm and a gated recurrent unit neural network, and generates an optimal energy-saving control strategy for the air conditioning equipment based on the combined prediction model and the data information of the air conditioning equipment.

[0014] Thirdly, embodiments of this application provide a computer storage medium including one or more computer instructions, which, when executed, implement any of the methods described above.

[0015] Fourthly, embodiments of this application provide an electronic device, including a memory and a processor. The memory is used to store one or more computer instructions, and the processor is used to call and execute the one or more computer instructions, thereby realizing the deep learning-based greenhouse air conditioning control method as described above.

[0016] Additional aspects and advantages of this application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of this application. Attached Figure Description

[0017] To more clearly illustrate the implementation of this application, the graphics involved in the embodiments will be briefly described below.

[0018] Figure 1 This is a schematic diagram of the intelligent control system for air conditioning equipment in a greenhouse, according to an embodiment of this application. Figure 2 This is a schematic diagram of the air conditioning control strategy generation process according to an embodiment of this application; Figure 3 This is a flowchart of the improved algorithm according to an embodiment of this application; Figure 4 This is a schematic diagram of an electronic device according to an embodiment of this application.

[0019] Figure label: Electronic equipment 300; Memory 310; Operating system 311; Application program 312; Processor 320; Network interface 330; Input device 340; Hard disk 350; Display device 360. Detailed Implementation

[0020] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0021] In the description of this application, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential," etc., indicating orientation or positional relationships based on the orientation or positional relationships shown in the accompanying drawings, are used only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application. Furthermore, features defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, unless otherwise stated, "a plurality of" means two or more.

[0022] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0023] The following describes, with reference to the accompanying drawings, an intelligent control method for air conditioning in a greenhouse according to an embodiment of this application.

[0024] like Figures 1 to 3 As shown, the greenhouse air conditioning control method according to this application includes the following steps: S1. Collect data within a predetermined time period, including environmental factor data inside and outside the greenhouse, operation data of environmental control equipment, and crop physiological and ecological data.

[0025] The environmental factor data for the greenhouse in step S1 includes both indoor and outdoor environmental factor data. Indoor environmental factor data may include air temperature and humidity, light intensity, photosynthetically active radiation (RADR), and carbon dioxide concentration. Outdoor environmental factor data may include outdoor air temperature and humidity, RDR, and wind speed.

[0026] In addition, equipment operation data may include equipment operation methods and operation time. Crop physiological and ecological data may include crop growth data such as leaf temperature, stem diameter, and plant height.

[0027] S2. Based on the data, construct a combined prediction model of the load of air conditioning equipment and crop leaf temperature in the greenhouse, which can predict the future dynamic changes of the load of air conditioning equipment and crop leaf temperature in the greenhouse under different scenarios.

[0028] S3. Input the real-time collected data into the combined prediction model to obtain the real-time predicted values ​​of the load of the air conditioning equipment and the temperature of the crop leaves.

[0029] S4. Based on the data information and load of the air conditioning equipment, construct a dataset for the air conditioning equipment. Select data samples from the real-time predicted values ​​that meet the suitable range for crop growth and where the load of the air conditioning equipment is at its minimum to form an optimal energy-saving control strategy for the air conditioning equipment. For example, ... Figure 2 As shown, a dataset of all possible combinations of the number and location of air conditioning equipment on / off can be constructed. This dataset, combined with the current greenhouse environment, the operating status of environmental control equipment (fans, wet curtains, internal and external shading, etc.), and crop physiological and ecological data, serves as the model input parameters. The model then predicts the corresponding air conditioning load and crop leaf temperature. The predicted crop leaf temperature is compared to determine if it meets the crop's growth requirements. If it does, the corresponding data sample of the air conditioning equipment on / off quantity, location, and predicted load becomes a strategy for air conditioning control. This method forms a solution set for air conditioning control strategies. Based on this solution set, the minimum predicted air conditioning load is found, and its corresponding combination of the number and location of on / off air conditioning equipment is used as the optimal air conditioning energy-saving control strategy. This optimal strategy is then used to precisely control the operation of the air conditioning system.

[0030] As can be seen, this application provides a method for generating the optimal energy-saving control strategy for air conditioning equipment in a greenhouse and intelligently controlling the air conditioning, which can not only control the temperature inside the greenhouse, but also effectively save energy.

[0031] According to one embodiment of this application, in step S2, the data is cleaned, and the cleaned data is then classified and integrated into a heating dataset and a cooling dataset based on the temperature control type. In other words, the data can be classified and integrated into a heating dataset and a cooling dataset based on the temperature control type.

[0032] In some specific embodiments of this application, step S2 includes: S21. Based on the temperature control type and air conditioning usage scenario, divide the data into a heating dataset and a cooling dataset, and further divide the heating dataset and cooling dataset into a training set and a test set, respectively. S22. Construct a single prediction model based on the training set, and obtain predicted values ​​for the load of the air conditioning equipment and the temperature of the crop leaves based on the single prediction model. S23. Generate the weight coefficient values ​​of the single prediction model at different times to obtain a time-varying weighted combined prediction model. S24. Select the test set to test the time-varying weighted combined prediction model to obtain the combined prediction model.

[0033] According to one embodiment of this application, such as Figure 3 As shown, in step S22, the single prediction model is constructed based on the Random Forest (RF) algorithm and the Gated Recurrent Unit (GRU) neural network; in step S23, the weight coefficient values ​​of the single prediction model at different times are generated by the Sparrow Search algorithm.

[0034] According to one embodiment of this application, the fitness function of the sparrow search algorithm is set as the first... Mean squared error of the time-combination prediction model: (1), In equation (1), T represents the total number of moments. For a moment, For the first Actual value at any time For the first The output value of the random forest model at each time step. For the first The output value of the time-gated recurrent unit neural network model. For the first The weight coefficients of the RF model at time step [time]. For the first Weight coefficients of the time-gated recurrent unit neural network model; Step S23 includes: S231. Update the discoverer's location. (2); + ( ) (3); In equation (2), t represents the current iteration number; The position of the sparrow; A uniformly random number between (0, 1); Represents a range vector; This is the warning value; This is a safe value; In formula (3) , This represents the final value of the weight. This represents the maximum number of iterations.

[0035] It should be noted that the original calculation method used in the existing technology to update the discoverer's location is... The original SSA algorithm suffers from weak global search capabilities and is prone to getting trapped in local optima during the optimization process. This application proposes a dynamic adaptive weighting method, which allows the discoverer sufficient time for a global search in the early stages of the algorithm's search, and enables the discoverer to conduct a depth search on a specific local region in the later stages, thus accelerating the algorithm's convergence. That is, this application employs... ; + ( ) .in and These are the initial and final values ​​of the weights, respectively. This represents the maximum number of iterations.

[0036] S232, Update follower positions. (4); In equation (4) express Worst position; This represents the position of the sparrow with the best fitness in the (t+1)th iteration; Represents a matrix; S233, Update the location of the vigilant. (5); In equation (4) This represents the optimal position of the sparrow in the t-th round; This represents the position of the sparrow with the best fitness in the (t+1)th iteration; This represents the worst position of the sparrow in the t-th round; K is a random number, taking a value between -1 and 1. These are the step size control parameters; To obtain sufficiently small numbers, avoid denominators of 0; and These represent the sparrow's fitness, worst fitness, and best fitness values, respectively.

[0037] In other words, the prediction method for the greenhouse air conditioning load and crop leaf temperature in step S2 adopts a time-varying weight combination model based on random forest and gated recurrent unit neural network. The algorithm flowchart is as follows. Figure 3 As shown, the main steps include: S1. Based on the temperature control type and air conditioning usage scenario, the sample dataset is divided into two types: cooling dataset and heating dataset, which are then used as training and testing sets respectively.

[0038] S2. Construct RF and GRU neural network prediction models based on the training set to obtain the corresponding future values ​​of air conditioning equipment load and crop leaf temperature.

[0039] S3. An improved Sparrow Search Algorithm (SSA) is used to obtain time-varying weight coefficients, meaning that a single model has different weight coefficients at different times, achieving a minimum error for the combined prediction model. The fitness function of SSA is set as the... Mean square error of time-combination prediction model ,in For the first Actual value at any time For the first The output value of the RF model at time 10:00. For the first The output value of the GRU model at time 10:00. For the first The weight coefficients of the RF model at time step [time]. For the first The weight coefficients of the GRU model at time step.

[0040] S4. Based on the improved SSA, obtain the optimal time-varying weight coefficients to generate the combined model prediction values. ,in .

[0041] S5. Real-time collection of environmental factor data inside and outside the greenhouse, crop data, and equipment operation status data, and input of real-time data into the optimal time-varying weighted combined model to predict the future values ​​of air conditioning equipment load and crop leaf temperature.

[0042] In step S2, the gated recurrent unit neural network (GRU) is an improved long short-term memory network (LSTM). By optimizing the number of gates, it improves the training speed of the model while maintaining the time-series prediction accuracy of LSTM by simplifying the model structure. Its structural design contains only two "gates": an "update gate" and a "reset gate." The update gate controls the information input and the retention of historical state information; its calculation formula is as follows: The reset gate is primarily used to determine the degree to which previous information has been forgotten; the calculation formula is as follows: The output candidate value after the reset gate is The current output information is .

[0043] Step S3 describes the improved SSA that optimizes the weight coefficients of the two single models, RF and GRU, and includes the following steps: S31. Update the discoverer's location: Due to the use of existing technologies ;in This represents the current iteration number; The position of the sparrow; A uniformly random number between (0, 1); Represents a range vector; This is the warning value; This is a safe value. The original SSA (Search and Optimize) method suffers from weaknesses in global search capabilities and is prone to getting trapped in local optima during the optimization process.

[0044] The dynamic adaptive weight method proposed in this application allows the discoverer in the early stage of the algorithm search to have sufficient time to conduct a global search, and the discoverer in the later stage of the algorithm search can conduct a deep search on a certain local area, thereby accelerating the convergence of the algorithm.

[0045]

[0046] + ( )

[0047] in and These are the initial and final values ​​of the weights, respectively. This represents the maximum number of iterations.

[0048] S32. Update follower positions:

[0049] in express Worst position; This represents the position of the sparrow with the best fitness in the (t+1)th iteration; Represents a matrix.

[0050] S33. Update the location of the vigilant.

[0051] in It is a random number, with a value between -1 and 1; These are the step size control parameters; To obtain sufficiently small numbers, avoid denominators of 0; and These represent the sparrow's fitness, worst fitness, and best fitness values, respectively.

[0052] In some specific embodiments of this application, the data information of the air conditioning equipment includes the number of air conditioning units that are started and stopped and their placement position. By controlling the above-mentioned data information of the air conditioning equipment, the greenhouse temperature can be controlled on the basis of energy saving, so that the temperature meets the needs of crop growth.

[0053] As can be seen, this application can use a combined model based on the fusion of RF and GRU neural networks to predict the load of air conditioning equipment and the temperature of crop leaves. It uses an improved SSA algorithm to obtain the weight parameters of the single model at different times, effectively combining the advantages of RF and GRU neural network models and improving the prediction accuracy of the combined model.

[0054] like Figure 1 As shown, this application also provides an intelligent air conditioning control system for greenhouses, used to implement the methods described in this application, including: a data monitoring module, a cloud platform service module, and an air conditioning equipment control execution module.

[0055] Specifically, the data monitoring module collects data, including environmental factor data inside and outside the greenhouse, operation data of environmental control equipment, and crop physiological and ecological data. The cloud platform service module analyzes the data collected by the data monitoring module to generate an optimal air conditioning energy-saving control strategy. The air conditioning equipment control execution module performs automatic control operations on the air conditioning equipment according to the optimal air conditioning energy-saving control strategy.

[0056] In other words, this application also provides an intelligent air conditioning control system for greenhouses, the system structure of which is shown in the schematic diagram below. Figure 1 As shown, the system may include a data monitoring module, a cloud platform service module, and an air conditioning equipment control execution module. The data monitoring module can transmit the collected data to the cloud platform server for data cleaning, such as erroneous data removal, data supplementation, and data normalization.

[0057] For example, the data monitoring module collects data on environmental factors inside and outside the greenhouse, equipment operation data, and crop physiological and ecological data. Greenhouse environmental factor data includes air temperature and humidity, light intensity, photosynthetically active radiation (PAR), and carbon dioxide concentration; greenhouse external environmental factor data includes outdoor air temperature and humidity, PAR, and wind speed; equipment operation data includes equipment operation methods and operation times; and crop physiological and ecological data includes crop growth data such as leaf temperature, stem diameter, and plant height.

[0058] The data monitoring module primarily collects data on environmental factors inside and outside the greenhouse, equipment operation data, and crop physiological and ecological data. The environmental factor data includes air temperature and humidity, photosynthetically active radiation, and wind speed; the equipment operation data includes the operating status and duration of environmental control equipment; and the crop physiological and ecological data includes crop growth information such as leaf temperature and plant height. The cloud platform service module utilizes big data technology to accurately store and clean the massive amounts of data transmitted by the data monitoring module. It then uses artificial intelligence algorithms to effectively analyze the data, generating optimal air conditioning energy-saving control strategies and sending them to the control execution module. The control execution module then performs automatic control operations on the air conditioning equipment based on these optimal energy-saving control strategies.

[0059] In some specific embodiments of this application, the cloud platform service module includes a data storage and cleaning module and a data analysis and decision-making module.

[0060] Specifically, the data storage and cleaning module can store and clean the data, such as removing erroneous data, supplementing data, and normalizing data. The data analysis and decision-making module constructs a combined prediction model of the air conditioning equipment load and crop leaf temperature using a random forest algorithm and a gated recurrent unit neural network, and generates an optimal energy-saving control strategy for the air conditioning equipment based on the combined prediction model and the data information of the air conditioning equipment.

[0061] For example, when transmitting the collected data to the cloud platform server, the data storage and cleaning module transmits the collected data to the cloud platform server to perform data cleaning such as erroneous data removal, data supplementation, and data normalization.

[0062] Furthermore, after the cleaned data is further classified and integrated into heating and cooling datasets according to the temperature control type, the data analysis and decision-making module can predict the future dynamic changes in the load of air conditioning equipment and crop leaf temperature in the greenhouse under different scenarios based on the classified datasets.

[0063] The data analysis and decision-making module constructs a dataset containing all possible combinations of the number and location of air conditioning equipment in operation. This dataset, combined with the current greenhouse environment, the operating status of environmental control equipment (fans, wet curtains, internal and external shading, etc.), and crop physiological and ecological data, serves as input parameters for the model. The model then predicts the corresponding air conditioning load and crop leaf temperature. The predicted crop leaf temperature is compared to the crop's growth requirements. If the requirements are met, the corresponding data sample of the air conditioning equipment start / stop combination and the predicted air conditioning load becomes a control strategy. This method forms a solution set for air conditioning control strategies. Based on this solution set, the module finds the minimum predicted air conditioning load and uses its corresponding combination of start / stop and location data as the optimal energy-saving control strategy. This optimal energy-saving control strategy is then used to precisely control the air conditioning operation.

[0064] The data analysis and decision-making module mainly involves the generation of air conditioning equipment load and crop leaf temperature prediction models, as well as the optimal energy-saving control strategy for air conditioning equipment. The air conditioning equipment load and crop leaf temperature prediction models are constructed using a combination of random forest and gated recurrent unit neural network methods. The optimal energy-saving control strategy for air conditioning equipment is generated based on the predicted values ​​from the combined models and the air conditioning control template.

[0065] The method and system of this application will be described below with reference to specific embodiments. The main steps of the method are as follows: 1) Collect data on environmental factors inside and outside the greenhouse, equipment operation data, and crop physiological and ecological data within a predetermined time period.

[0066] 2) Clean all the above data and divide the processed sample set into training set and test set.

[0067] 3) Based on data cleaning, single prediction models are constructed using the RF algorithm and the GRU neural network, respectively, and the analysis and prediction results of each single model are obtained.

[0068] 4) The improved sparrow search algorithm (SSA) is used to generate the weight coefficient values ​​of the single prediction model at different times (i.e., the variable weight coefficients), and then the combined model is derived to obtain the combined prediction values ​​of the air conditioning equipment load and crop leaf temperature in the greenhouse.

[0069] 5) Select a test sample set to test the above time-varying weighted prediction model based on RF and GRU, and obtain a combined prediction model with high prediction accuracy.

[0070] 6) Collect environmental factor data, equipment operation data, and crop physiological and ecological data inside and outside the greenhouse in real time, and input the real-time data into the combined prediction model to obtain the future values ​​of the air conditioning equipment load and crop leaf temperature in the greenhouse.

[0071] 7) Based on the number and placement of air conditioners in the greenhouse, construct a dataset template with different start / stop counts and locations for air conditioning equipment. Use the aforementioned prediction model to obtain real-time predicted values ​​of air conditioning equipment load and crop leaf temperature. Select data samples from the predicted crop leaf temperature values ​​that meet the suitable range for crop growth and have the lowest air conditioning equipment load, and formulate the optimal energy-saving control strategy for air conditioning equipment.

[0072] In summary, this application provides a deep learning method for accurately and dynamically predicting the load of air conditioning and crop leaf temperature in greenhouses. Based on the predicted values, an energy-saving control strategy for air conditioning is generated, thereby achieving intelligent control of air conditioning equipment in greenhouses. The system described in this application comprehensively acquires information on environmental factors inside and outside the greenhouse, crop physiological and ecological information, and operational information of control equipment (fans, wet curtains, skylights, and air conditioners, etc.) based on the Internet of Things. A combined model of random forest and gated recurrent unit neural network is used to predict the future dynamic changes of air conditioning load and crop leaf temperature. The dataset is classified according to the temperature control type to construct corresponding prediction models, accurately and dynamically predicting the air conditioning load and crop leaf temperature. Based on the future predicted values ​​of air conditioning load and crop leaf temperature, the number and location of air conditioner start-ups and shutdowns are adjusted in real time to maintain the air temperature in the greenhouse at a suitable value for crop growth.

[0073] In addition, embodiments of the present invention also provide a computer storage medium, the computer storage medium including one or more computer instructions, which, when executed, implement any of the methods described above.

[0074] In other words, the computer storage medium stores a computer program, which, when executed by a processor, causes the processor to perform any of the methods described above.

[0075] like Figure 4 As shown, an embodiment of the present invention provides an electronic device 300, including a memory 310 and a processor 320. The memory 310 is used to store one or more computer instructions, and the processor 320 is used to call and execute the one or more computer instructions, thereby implementing any of the methods described above.

[0076] In other words, the electronic device 300 includes a processor 320 and a memory 310, wherein computer program instructions are stored in the memory 310, and wherein, when the computer program instructions are executed by the processor, the processor 320 performs any of the methods described above.

[0077] Furthermore, such as Figure 4As shown, the electronic device 300 also includes a network interface 330, an input device 340, a hard disk 350, and a display device 360.

[0078] The various interfaces and devices described above can be interconnected via a bus architecture. The bus architecture can include any number of interconnecting buses and bridges. Specifically, various circuits representing one or more central processing units (CPUs) (represented by processor 320) and one or more memories (represented by memory 310) are connected together. The bus architecture can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits. It is understood that the bus architecture is used to implement communication between these components. In addition to the data bus, the bus architecture also includes a power bus, a control bus, and a status signal bus, all of which are well known in the art and will not be described in detail herein.

[0079] The network interface 330 can be connected to a network (such as the Internet, a local area network, etc.), obtain relevant data from the network, and store it in the hard disk 350.

[0080] The input device 340 can receive various instructions input by the operator and send them to the processor 320 for execution. The input device 340 may include a keyboard or a clicking device (e.g., a mouse, trackball, touchpad, or touchscreen).

[0081] The display device 360 ​​can display the results obtained by the processor 320 executing instructions.

[0082] The memory 310 is used to store programs and data necessary for the operation of the operating system, as well as intermediate results and other data during the calculation process of the processor 320.

[0083] It is understood that the memory 310 in the embodiments of the present invention may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory may be random access memory (RAM), which is used as an external cache. The memory 310 of the apparatus and methods described herein is intended to include, but is not limited to, these and any other suitable types of memory.

[0084] In some implementations, memory 310 stores elements such as executable modules or data structures, or subsets thereof, or extended sets thereof: operating system 311 and application programs 312.

[0085] The operating system 311 includes various system programs, such as the framework layer, core library layer, and driver layer, used to implement various basic business functions and handle hardware-based tasks. The application program 312 includes various applications, such as a browser, used to implement various application functions. Programs implementing the methods of this embodiment of the invention can be included in the application program 312.

[0086] When the processor 320 calls and executes the application program and data stored in the memory 310, specifically the program or instructions stored in the application program 312, it distributes one of the first set and the second set to the nodes where the other of the first set and the second set are distributed, wherein the other set is distributed and stored on at least two nodes; and performs intersection processing on a node-by-node basis according to the node distribution of the first set and the node distribution of the second set.

[0087] The methods disclosed in the above embodiments of the present invention can be applied to or implemented by processor 320. Processor 320 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above methods can be completed by integrated logic circuits in the hardware of processor 320 or by instructions in software form. The processor 320 may be a general-purpose processor, digital signal processor (DSP), application-specific integrated circuit (ASIC), off-the-shelf programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components, and can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of the present invention can be directly embodied in the execution of a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software modules may reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory 310, and processor 320 reads information from memory 310 and, in conjunction with its hardware, completes the steps of the above methods.

[0088] It is understood that the embodiments described herein can be implemented in hardware, software, firmware, middleware, microcode, or a combination thereof. For hardware implementation, the processing unit can be implemented in one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers, microprocessors, other electronic units for performing the functions described herein, or combinations thereof.

[0089] For software implementation, the techniques described herein can be achieved through modules (e.g., procedures, functions, etc.) that perform the functions described herein. The software code can be stored in memory and executed by a processor. The memory can be implemented within the processor or externally.

[0090] Specifically, the processor 320 is also used to read the computer program and execute any of the methods described above.

[0091] In the several embodiments provided in this application, it should be understood that the disclosed methods and apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0092] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can be physically comprised separately, or two or more units can be integrated into one unit. The integrated unit described above can be implemented in hardware or in the form of hardware plus software functional units.

[0093] The integrated units implemented as software functional units described above can be stored in a computer-readable storage medium. These software functional units, stored in a storage medium, include several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute some steps of the transmission and reception methods described in 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.

[0094] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A deep learning-based method for controlling air conditioning in a greenhouse, characterized in that, Includes the following steps: S1. Collect data within a predetermined time period, including data on environmental factors inside and outside the greenhouse, operational data of environmental control equipment, and crop physiological and ecological data. S2. Construct a combined prediction model of the load of air conditioning equipment in the greenhouse and the temperature of crop leaves based on the data; S3. Input the real-time collected data into the combined prediction model to obtain the real-time predicted values ​​of the load of the air conditioning equipment and the temperature of the crop leaves; S4. Construct a dataset for the air conditioning equipment based on the data information and load of the air conditioning equipment, select data samples of the air conditioning equipment data information that meet the suitable range for crop growth and have the lowest load of the air conditioning equipment from the real-time predicted values, and form the optimal energy-saving control strategy for the air conditioning equipment.

2. The deep learning-based greenhouse air conditioning control method according to claim 1, characterized in that, In step S2, the data is cleaned, and the cleaned data is then classified and integrated into a heating dataset and a cooling dataset according to the temperature control type.

3. The deep learning-based greenhouse air conditioning control method according to claim 2, characterized in that, Step S2 includes: S21. Based on the temperature control type and air conditioning usage scenario, the data is divided into the heating dataset and the cooling dataset, and the heating dataset and the cooling dataset are respectively divided into training set and test set; S22. Construct a single prediction model based on the training set, and obtain the predicted values ​​of the load of the air conditioning equipment and the temperature of the crop leaves based on the single prediction model; S23. Generate the weight coefficient values ​​of the single prediction model at different times to obtain the time-varying weighted combination prediction model; S24. Select the test set and test the time-varying weight combination prediction model to obtain the combination prediction model.

4. The deep learning-based greenhouse air conditioning control method according to claim 3, characterized in that, In step S22, the single prediction model is constructed based on the random forest algorithm and the gated recurrent unit neural network; in step S23, the weight coefficient values ​​of the single prediction model at different times are generated by the sparrow search algorithm.

5. The deep learning-based greenhouse air conditioning control method according to claim 4, characterized in that, The fitness function of the sparrow search algorithm is set to the first... Mean squared error of the time-combination prediction model: （1）， In equation (1), T represents the total number of moments. For a moment, For the first Actual value at any time For the first The output value of the random forest model at each time step. For the first The output value of the time-gated recurrent unit neural network model. For the first The weight coefficients of the RF model at time step [time]. For the first Weight coefficients of the time-gated recurrent unit neural network model; Step S23 includes: S231. Update the discoverer's location. （2）； + ( ) （3）； In equation (2), t represents the current iteration number; The position of the sparrow; A uniformly random number between (0, 1); Represents a range vector; This is the warning value; This is a safe value; In formula (3) The initial values ​​for the weights are... This represents the final value of the weight. This represents the maximum number of iterations. S232, Update follower positions. （4）； In equation (4) express Worst position; This represents the position of the sparrow with the best fitness in the (t+1)th iteration; Represents a matrix; S233, Update the location of the vigilant. （5）； In equation (5) This represents the optimal position of the sparrow in the t-th round; This represents the position of the sparrow with the best fitness in the (t+1)th iteration; This represents the worst position of the sparrow in the t-th round; K is a random number, taking a value between -1 and 1. These are the step size control parameters; To obtain sufficiently small numbers, avoid denominators of 0; and These represent the sparrow's fitness, worst fitness, and best fitness values, respectively.

6. The deep learning-based greenhouse air conditioning control method according to claim 1, characterized in that, The data information of the air conditioning equipment includes the number of air conditioning units that are started and stopped, and their placement location.

7. A greenhouse air conditioning intelligent control system, used to implement the method according to any one of claims 1-6, characterized in that, include: The data monitoring module collects data, including data on environmental factors inside and outside the greenhouse, operational data of environmental control equipment, and crop physiological and ecological data. The cloud platform service module analyzes the data collected by the data monitoring module and generates the optimal air conditioning energy-saving control strategy. An air conditioning equipment control execution module, which performs automatic control operations on the air conditioning equipment according to the optimal air conditioning energy-saving control strategy.

8. The intelligent air conditioning control system for greenhouses according to claim 7, characterized in that, The cloud platform service modules include: A data storage and cleaning module, which is capable of storing and cleaning the data; The data analysis and decision-making module constructs a combined prediction model of the load of the air conditioning equipment and the temperature of the crop leaves using a random forest algorithm and a gated recurrent unit neural network. Based on the combined prediction model and the data information of the air conditioning equipment, the module generates the optimal energy-saving control strategy for the air conditioning equipment.

9. A computer storage medium, characterized in that, It includes one or more computer instructions, which, when executed, implement the deep learning-based greenhouse air conditioning control method as described in any one of claims 1-6.

10. An electronic device comprising a memory and a processor, characterized in that, The memory is used to store one or more computer instructions; The processor is used to invoke and execute one or more computer instructions to implement the method as described in any one of claims 1-6.

Images