Multi-target sunlight greenhouse ventilation decision-making method and device, electronic equipment and storage medium

Abstract

The invention provides a multi-target sunlight greenhouse ventilation decision-making method and device, electronic equipment and a storage medium. The multi-target sunlight greenhouse ventilation decision-making method comprises the steps that real-time environment information inside and outside a greenhouse is obtained, and greenhouse ventilation feature vectors are determined; generating a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector; performing multi-objective optimization on the preliminary ventilation decision strategy; and dynamically adjusting an optimal ventilation decision strategy to automatically adapt to meteorological changes inside and outside the greenhouse and growth stages of crops. According to the multi-target sunlight greenhouse ventilation decision-making method provided by the invention, technologies such as a multi-sensor fusion technology, a multi-mode variational auto-encoder algorithm, a soft behavior cloning algorithm, a meta-learning technology, a multi-target particle swarm optimization algorithm, a multi-target co-evolution algorithm, a multi-agent system and a transfer learning technology of reinforcement learning are comprehensively utilized; the greenhouse ventilation decision-making efficiency can be effectively improved, greenhouse ventilation energy consumption is reduced, and intelligent control over the temperature in the greenhouse and the growth quality of crops are improved.

Description

Technical Field

[0001] The present invention relates to the field of agricultural automation technology, and in particular to a multi-objective solar greenhouse ventilation decision-making method, device, electronic equipment and storage medium. Background Art

[0002] Solar greenhouses use solar energy and natural ventilation to adjust the environment in the greenhouse. They are energy-saving, low-carbon, and environmentally friendly, and are an important tool in agricultural production. However, solar greenhouses also face challenges such as uncertainty in the environment inside and outside the greenhouse, temperature fluctuations in the greenhouse, and crop diversity. These factors will affect the ventilation effect of the greenhouse and the growth quality of the crops.

[0003] Traditional greenhouse ventilation decision-making methods are usually based on experience or fixed thresholds, lack dynamic adaptation to the environment inside and outside the greenhouse, and do not take into account the temperature, humidity, energy consumption and other multi-objective requirements of crops in the greenhouse. The above methods often cannot achieve optimal control of greenhouse temperature, nor can they meet the growth needs of crops, resulting in low energy efficiency in the greenhouse and poor growth quality of crops.

[0004] Therefore, how to reasonably make greenhouse ventilation decisions based on real-time environmental data inside and outside the greenhouse is a key issue in improving greenhouse production efficiency and effectiveness. Summary of the invention

[0005] The present invention provides a multi-objective solar greenhouse ventilation decision-making method, device, electronic device and storage medium, which at least solves one of the technical problems existing in the prior art or related technology, such as low greenhouse ventilation decision-making efficiency, high energy consumption, poor crop growth quality, etc.

[0006] The present invention provides a multi-objective solar greenhouse ventilation decision-making method, comprising the following steps: Get real-time environmental information inside and outside the greenhouse; Determining a greenhouse ventilation characteristic vector according to the real-time environmental information; generating a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector; Performing multi-objective optimization on the preliminary ventilation decision strategy so that the optimized optimal ventilation decision strategy takes into account the temperature control in the greenhouse, the crop growth requirements, the operating cost and the effect; The optimal ventilation decision strategy is dynamically adjusted to automatically adapt to the meteorological changes inside and outside the greenhouse and the growth stage of the crops.

[0007] According to a multi-objective solar greenhouse ventilation decision-making method provided by the present invention, the real-time environmental information includes at least one of the following environmental data: indoor and outdoor temperature, indoor and outdoor humidity, indoor and outdoor light radiation value, wind speed, wind direction, indoor and outdoor images, greenhouse volume, greenhouse height-span ratio, greenhouse comprehensive thermal resistance, indoor and outdoor temperature difference, light transmittance, saturated water vapor pressure difference; The real-time environmental information inside and outside the greenhouse is obtained, including: The temperature measurements inside and outside the greenhouse are combined by weighted average method to obtain the temperature inside and outside the greenhouse; The temperature and humidity measurement values ​​inside and outside the greenhouse are integrated by weighted average method to obtain the temperature and humidity inside and outside the greenhouse; The light radiation measurement values ​​inside and outside the greenhouse are merged by weighted average method to obtain the light radiation inside and outside the greenhouse; The wind speed measurements inside and outside the greenhouse are combined by weighted average method to obtain the wind speed inside and outside the greenhouse; The wind direction measurements inside and outside the greenhouse are merged by vector averaging method to obtain the wind direction inside and outside the greenhouse; Fusing the real-time images of the inside and outside of the greenhouse at the pixel level or feature level to obtain the indoor and outdoor images; Calculate the volume of the greenhouse according to the length, width, height and other dimensions of the greenhouse; Calculating the height-to-span ratio of the greenhouse according to the height and span of the greenhouse; Calculating the comprehensive thermal resistance of the greenhouse based on the material thermal resistance and surface area of ​​each surface of the greenhouse; Calculating the indoor and outdoor temperature difference according to the indoor and outdoor temperatures of the greenhouse; Calculating the light transmittance according to the indoor and outdoor light radiation values ​​of the greenhouse; The saturated water vapor pressure difference is calculated according to the temperature in the greenhouse and the humidity in the greenhouse.

[0008] According to a multi-objective solar greenhouse ventilation decision-making method provided by the present invention, determining the greenhouse ventilation feature vector according to the real-time environmental information includes: Inputting the real-time environmental information into a multimodal variational autoencoder to obtain an initial feature vector related to the influencing factors of greenhouse ventilation and the law of temperature change; The initial feature vector is adaptively weighted and fused using a multimodal attention mechanism to obtain a greenhouse ventilation feature vector.

[0009] According to a multi-objective solar greenhouse ventilation decision-making method provided by the present invention, the initial feature vector is adaptively weighted and fused using a multimodal attention mechanism to obtain a greenhouse ventilation feature vector, including: Determining attention weights and gating mechanisms for different modalities of real-time environmental information constituting the initial feature vector; According to the attention weight and gating mechanism, the data of different modalities are adaptively weighted and fused to obtain the greenhouse ventilation feature vector.

[0010] According to a multi-objective solar greenhouse ventilation decision method provided by the present invention, generating a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector includes: Based on the soft behavior cloning algorithm, the objective function is constructed by maximizing the reward and minimizing the difference between the output action and the expert strategy; Determining a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector and a reward function determined by the objective function; According to the preliminary ventilation decision strategy, an output action for greenhouse ventilation is obtained; The objective function includes a policy network and a value network respectively constructed by using two neural networks; the policy network outputs Gaussian distribution parameters, and the Gaussian distribution parameters are used to characterize the probability distribution of the output action under the greenhouse ventilation feature vector; the value network outputs a scalar, and the scalar is used to characterize the expected reward value under the greenhouse ventilation feature vector.

[0011] According to a multi-objective solar greenhouse ventilation decision method provided by the present invention, after constructing the objective function, it also includes: The objective function is optimized based on the entropy regularization method, including adding an entropy term to the objective function, and the expression of the entropy term is: ; The expression of the optimized objective function is: ; in, Express expectations about variables; Greenhouse ventilation feature vector for greenhouse ventilation; Output action for greenhouse ventilation; Reward function for greenhouse ventilation; For strategic networks; and are model parameters; Demonstrate datasets for experts; and is a hyperparameter; is the entropy function; is the KL divergence; For expert strategies; is the entropy term of entropy regularization; is the objective function.

[0012] According to a multi-objective solar greenhouse ventilation decision method provided by the present invention, after determining a preliminary ventilation decision strategy, the method further includes: Multi-task learning and meta-gradient updating are used to adapt the greenhouse ventilation decision strategy according to the current greenhouse environment and crop types.

[0013] According to a multi-objective solar greenhouse ventilation decision method provided by the present invention, the preliminary ventilation decision strategy is optimized by multiple objectives, including: The multi-objective particle swarm optimization algorithm and the multi-objective co-evolution algorithm are used to perform multi-objective optimization on the preliminary decision-making actions, and the objectives of greenhouse temperature control, crop growth requirements, operating costs and effects are comprehensively considered to obtain the optimal ventilation decision strategy.

[0014] According to a multi-objective solar greenhouse ventilation decision method provided by the present invention, the optimal ventilation decision strategy is dynamically adjusted to automatically adapt to the meteorological changes inside and outside the greenhouse and the growth stage of the crops, including: By using a multi-agent system and transfer learning technology, the optimal ventilation decision strategy is optimized in real time by utilizing the similarities and differences between the source task and the target task; The target task is determined based on the meteorological changes inside and outside the greenhouse and the growth stage of the crops.

[0015] The present invention also provides a multi-objective solar greenhouse ventilation decision-making device, comprising the following modules: Information collection unit, used to obtain real-time environmental information inside and outside the greenhouse; An information processing unit, used for determining a greenhouse ventilation characteristic vector according to the real-time environmental information; A strategy generating unit, used for generating a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector; A strategy optimization unit, used for performing multi-objective optimization on the preliminary ventilation decision strategy, so that the optimized optimal ventilation decision strategy takes into account the temperature control in the greenhouse, the crop growth requirements, the operation cost and the effect; The strategy matching unit is used to dynamically adjust the optimal ventilation decision strategy to automatically adapt to the meteorological changes inside and outside the greenhouse and the growth stage of the crops.

[0016] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, any of the multi-objective solar greenhouse ventilation decision methods described above is implemented.

[0017] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the multi-objective solar greenhouse ventilation decision-making method as described above is implemented.

[0018] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the multi-objective solar greenhouse ventilation decision-making method as described above is implemented.

[0019] The multi-objective solar greenhouse ventilation decision-making method, device, electronic device and storage medium provided by the present invention comprehensively utilize multi-sensor fusion technology, multi-modal variational autoencoder algorithm, soft behavior cloning algorithm, meta-learning technology, multi-objective particle swarm optimization algorithm, multi-objective collaborative evolution algorithm, multi-agent system and reinforcement learning transfer learning technology, etc., which can effectively improve the efficiency of greenhouse ventilation decision-making, reduce greenhouse ventilation energy consumption, and improve the intelligent control of greenhouse temperature and crop growth quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 It is a flow chart of the multi-objective solar greenhouse ventilation decision-making method provided by the present invention.

[0022] Figure 2 It is a structural schematic diagram of the multi-objective solar greenhouse ventilation decision-making device provided by the present invention.

[0023] Figure 3 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] It should be noted that, in the description of the present invention, the terms "include", "comprises" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "includes a ..." do not exclude the existence of other identical elements in the process, method, article or device including the elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0026] Greenhouse ventilation decision-making is an important link in greenhouse production. The traditional technical methods are mainly the following: (1) Empirical method: Based on the experience and intuition of greenhouse producers, the ventilation opening of the greenhouse is manually or timed to control the temperature inside the greenhouse according to the temperature, humidity and other indicators inside and outside the greenhouse. The advantage is that it is simple and easy to implement, and does not require complex equipment and algorithms. However, the disadvantage is that it lacks scientific basis, cannot adapt to the dynamic changes of the environment inside and outside the greenhouse, and cannot take into account the multi-objective needs of crops in the greenhouse. It is easy to cause the temperature in the greenhouse to be too high or too low, affecting the growth quality of crops.

[0027] (2) Threshold-based method: According to the temperature, humidity and other indicators inside and outside the greenhouse, a fixed threshold is set. When the environmental data inside and outside the greenhouse exceeds or falls below the threshold, the ventilation opening of the greenhouse is automatically adjusted to achieve temperature control inside the greenhouse. The advantage is that it can realize the automation of greenhouse ventilation and reduce manual intervention. The disadvantage is that the setting of the threshold is often subjective and cannot reflect the real-time changes in the environment inside and outside the greenhouse. It also cannot take into account the multi-objective needs of the crops in the greenhouse, which can easily cause temperature fluctuations in the greenhouse and affect the growth quality of the crops.

[0028] (3) Model-based method: Based on the environmental data inside and outside the greenhouse, a mathematical model of the greenhouse is established, such as an energy balance model, a dynamic model, etc. Then, based on the greenhouse model, optimization algorithms such as linear programming, nonlinear programming, and fuzzy control are used to solve the optimal ventilation opening of the greenhouse to achieve temperature control in the greenhouse. The advantage is that it can optimize greenhouse ventilation and improve the energy efficiency of the greenhouse, but the disadvantage is that the establishment of a greenhouse model is often complex and requires a large number of parameters and data. In addition, the greenhouse model is often inaccurate and cannot fully reflect the complexity of the environment inside and outside the greenhouse, nor can it take into account the multi-objective needs of crops in the greenhouse, which can easily cause errors and instability in decision-making.

[0029] Existing greenhouse ventilation decision-making technology is mainly used in the field of facility agriculture, especially in the field of solar greenhouses, with the purpose of achieving reasonable control of greenhouse temperature and ensuring the growth and quality of crops. However, the existing technology still has the following shortcomings: (1) Insufficient data acquisition and processing capabilities: Existing technologies often only use single or limited sensor data and are unable to obtain multiple data types such as temperature, humidity, light radiation, etc. inside and outside the greenhouse. They are also unable to effectively fuse and analyze the data, resulting in insufficient acquisition and processing capabilities of greenhouse environmental information and an inability to reflect the real-time and comprehensive status of the greenhouse.

[0030] To this end, the present invention utilizes multi-sensor fusion technology to collect and analyze various data types inside and outside the greenhouse to obtain real-time and comprehensive information about the greenhouse environment, improve the quality and reliability of the data, and the ability to express and distinguish the data.

[0031] (2) Insufficient multi-objective optimization and coordination capabilities: Existing technologies often only consider a single or main goal, such as minimizing the temperature difference between the inside and outside of the greenhouse, maximizing the yield or quality of crops, etc., and ignore the mutual influence and constraints between other goals. This leads to insufficient multi-objective optimization and coordination capabilities for greenhouse ventilation, and is unable to balance the temperature control in the greenhouse and the needs of crop growth, as well as energy consumption and utilization.

[0032] To this end, the present invention utilizes a multi-objective optimization algorithm to comprehensively consider multiple objectives such as meteorological parameters inside and outside the greenhouse, crop growth requirements, and energy benefits of active heat storage and release, to achieve multi-objective optimization and coordination of greenhouse ventilation, and balance temperature control in the greenhouse with crop growth requirements, as well as energy consumption and utilization.

[0033] (3) Insufficient intelligence and adaptive capabilities: Existing technologies often rely on artificial or empirical set values, parameters or models, and are unable to automatically learn and optimize greenhouse ventilation decision strategies, nor can they dynamically adjust greenhouse ventilation decision strategies. This results in insufficient intelligence and adaptive capabilities for greenhouse ventilation, and is unable to adapt to meteorological changes inside and outside the greenhouse and the growth stage of crops.

[0034] To this end, the present invention utilizes reinforcement learning and meta-learning techniques to automatically learn and optimize greenhouse ventilation decision-making strategies, dynamically adjust greenhouse ventilation decision-making strategies, realize intelligent and adaptive greenhouse ventilation, and adapt to meteorological changes inside and outside the greenhouse and the growth stage of crops.

[0035] In summary, the multi-objective solar greenhouse ventilation decision-making method provided by the present invention has the following improvements: (1) Data-based method: Based on the environmental data inside and outside the greenhouse, data mining and machine learning techniques, such as clustering, classification, regression, neural networks, etc., are used to analyze and predict the data of greenhouse ventilation. Then, based on the data characteristics of greenhouse ventilation, decision trees, rules, expert systems, etc. are used to formulate decision rules for greenhouse ventilation to achieve temperature control in the greenhouse. This method can utilize a large amount of data to improve the accuracy and sensitivity of greenhouse ventilation decisions.

[0036] (2) Based on the deep learning method, deep learning techniques such as convolutional neural networks, recurrent neural networks, and autoencoders are used to extract and represent the deep features of greenhouse ventilation data according to the environmental data inside and outside the greenhouse. Then, based on the deep features of greenhouse ventilation, deep reinforcement learning techniques such as deep Q networks, policy gradients, and actor-critic are used to learn and optimize the decision-making strategy of greenhouse ventilation to achieve temperature control in the greenhouse. The powerful representation and generalization capabilities of deep learning can be used to improve the intelligence level and efficiency of greenhouse ventilation decision-making.

[0037] (3) Based on the multi-objective optimization method, according to the environmental data inside and outside the greenhouse and the multi-objective needs of the crops in the greenhouse, a multi-objective optimization problem of greenhouse ventilation temperature, humidity, light, etc. is established. Then, multi-objective optimization techniques such as genetic algorithm, particle swarm optimization, and collaborative evolution are used to solve the multi-objective optimization solution of greenhouse ventilation to achieve the control of the temperature in the greenhouse. It can comprehensively consider the multi-objective needs of greenhouse crops and improve the coordination and balance of decision-making.

[0038] Combine the following Figure 1-Figure 3 The invention describes a multi-objective solar greenhouse ventilation decision-making method, device, electronic device and storage medium, which aims to achieve a more accurate and detailed classification of the target area through the combination of panchromatic image and multispectral image, and can improve the classification efficiency to a certain extent.

[0039] Figure 1 This is one of the flow charts of the multi-objective solar greenhouse ventilation decision method provided by the present invention. The execution subject can be an industrial computer, a local computer, a cloud server, or other mobile electronic terminals with data processing functions, such as iPads, mobile phones, etc. The present invention does not make specific limitations on this. Figure 1 As shown, the method includes but is not limited to the following steps: Step 101, obtaining real-time environmental information inside and outside the greenhouse.

[0040] Acquiring various data types such as temperature, humidity, light radiation, wind speed, wind direction, images, etc. inside and outside the greenhouse as input states for greenhouse ventilation decisions. Its function is to obtain real-time information on the environment inside and outside the greenhouse and provide data support for greenhouse ventilation decisions.

[0041] In order to improve the quality and reliability of the data, this step adopts Multi-Sensor Fusion Technology (MSFT), which uses sensors of different types and locations to comprehensively measure and calibrate the environmental parameters inside and outside the greenhouse, so as to eliminate the errors and noise of a single sensor, improve the accuracy and stability of the data, effectively solve the limitations and shortcomings of a single sensor, and improve the integrity and credibility of the greenhouse environmental data.

[0042] Furthermore, the following data can be calculated and derived as collaborative modeling variables based on greenhouse structural parameter indicators and internal and external meteorological environment data: greenhouse volume, greenhouse height-span ratio, greenhouse comprehensive thermal resistance (according to the proportion of each surface area), indoor and outdoor temperature difference, light transmittance, saturated water vapor pressure difference (Vapor Pressure Deficit, VPD), etc.

[0043] Step 102: determining a greenhouse ventilation feature vector according to the real-time environmental information.

[0044] Based on the multi-source real-time environmental information obtained in the previous step (such as temperature, humidity, light radiation, wind speed, wind direction, images, etc.), and the derived data obtained by calculation (such as greenhouse volume, height-span ratio, comprehensive thermal resistance, indoor and outdoor temperature difference, light transmittance, saturated water vapor pressure difference, etc.), a feature vector reflecting the greenhouse ventilation characteristics is constructed, called the greenhouse ventilation feature vector. This greenhouse ventilation feature vector is the basis for subsequent greenhouse ventilation decision-making, which integrates multiple key factors that affect greenhouse ventilation decisions.

[0045] Step 103: generating a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector.

[0046] Preset decision models or algorithms (such as rule-based decision trees, fuzzy logic, neural networks, etc.) can be used to analyze the priority and necessity of greenhouse ventilation under current environmental conditions, thereby generating a preliminary ventilation decision strategy. This strategy may include the degree of opening and closing of ventilation windows, ventilation time, ventilation method (natural ventilation or mechanical ventilation), etc.

[0047] Step 104, multi-objective optimization is performed on the preliminary ventilation decision strategy, so that the optimized optimal ventilation decision strategy takes into account the temperature control in the greenhouse, the crop growth requirements, the operation cost and the effect.

[0048] After generating the preliminary ventilation decision strategy, it can be further optimized for multiple objectives. The optimization goal is to balance the temperature control in the greenhouse, the crop growth requirements, the operating costs and effects, etc., to ensure that the ventilation decision can not only meet the optimal environmental conditions for crop growth, but also effectively control energy consumption and operating costs.

[0049] During the optimization process, multi-objective optimization algorithms (such as NSGA-II, MOEA / D, etc.) may be used to comprehensively consider the conflicts and trade-offs among various objectives.

[0050] Step 105, dynamically adjusting the optimal ventilation decision strategy to automatically adapt to the weather changes inside and outside the greenhouse and the growth stage of the crops.

[0051] After completing the multi-objective optimization, the optimized ventilation decision strategy can be dynamically adjusted according to the real-time monitored weather changes inside and outside the greenhouse and the growth stage of the crops. This dynamic adjustment capability enables the system to flexibly respond to various unpredictable environmental changes and ensure that the greenhouse environment is always in the most suitable state for crop growth. At the same time, through continuous learning and adaptation, its decision logic can be gradually optimized to improve the accuracy and efficiency of ventilation decisions.

[0052] The multi-objective solar greenhouse ventilation decision-making method provided by the present invention can comprehensively utilize multi-sensor fusion technology, multi-modal variational autoencoder algorithm, soft behavior cloning algorithm, meta-learning technology, multi-objective particle swarm optimization algorithm, multi-objective collaborative evolution algorithm, multi-agent system and reinforcement learning transfer learning technology, etc., which can effectively improve the efficiency of greenhouse ventilation decision-making, reduce greenhouse ventilation energy consumption, and improve the intelligent control of greenhouse temperature and crop growth quality.

[0053] Based on the content of the above embodiment, as an optional embodiment, the real-time environmental information includes at least one of the following environmental data: indoor and outdoor temperature, indoor and outdoor humidity, indoor and outdoor light radiation value, wind speed, wind direction, indoor and outdoor images, greenhouse volume, greenhouse height-span ratio, greenhouse comprehensive thermal resistance, indoor and outdoor temperature difference, light transmittance, saturated water vapor pressure difference; The real-time environmental information inside and outside the greenhouse is obtained, including: The temperature measurements inside and outside the greenhouse are combined by weighted average method to obtain the temperature inside and outside the greenhouse; The temperature and humidity measurement values ​​inside and outside the greenhouse are integrated by weighted average method to obtain the temperature and humidity inside and outside the greenhouse; The light radiation measurement values ​​inside and outside the greenhouse are merged by weighted average method to obtain the light radiation inside and outside the greenhouse; The wind speed measurements inside and outside the greenhouse are combined by weighted average method to obtain the wind speed inside and outside the greenhouse; The wind direction measurements inside and outside the greenhouse are merged by vector averaging method to obtain the wind direction inside and outside the greenhouse; Fusing the real-time images of the inside and outside of the greenhouse at the pixel level or feature level to obtain the indoor and outdoor images; Calculate the volume of the greenhouse according to the length, width, height and other dimensions of the greenhouse; Calculating the height-to-span ratio of the greenhouse according to the height and span of the greenhouse; Calculating the comprehensive thermal resistance of the greenhouse based on the material thermal resistance and surface area of ​​each surface of the greenhouse; Calculating the indoor and outdoor temperature difference according to the indoor and outdoor temperatures of the greenhouse; Calculating the light transmittance according to the indoor and outdoor light radiation values ​​of the greenhouse; The saturated water vapor pressure difference is calculated according to the temperature in the greenhouse and the humidity in the greenhouse.

[0054] The present invention installs various types of sensors at multiple locations inside and outside the greenhouse, mainly including temperature sensors, humidity sensors, light radiation sensors, wind speed sensors, wind direction sensors, image sensors, etc., to measure the environmental parameters inside and outside the greenhouse, such as temperature, humidity, light radiation, wind speed, wind direction, images, etc., and sends the measured data to the data processing center as the execution body through a wireless transmission module as the input state of the greenhouse ventilation decision.

[0055] In order to improve the quality and reliability of data, the present invention adopts multi-sensor fusion technology, using sensors of different types and arranged at different positions to comprehensively measure and calibrate the environmental parameters inside and outside the greenhouse, eliminate sensor errors and noise, improve data accuracy and stability, effectively solve the limitations and shortcomings of a single sensor, and improve the integrity and credibility of greenhouse environmental data.

[0056] Specifically, this step adopts the following multi-sensor fusion technology: Temperature sensor fusion technology: Several temperature sensors (of different types) are installed inside and outside the greenhouse to measure the temperature values ​​inside and outside the greenhouse (called temperature measurement values). The temperature measurement values ​​are fused by weighted averaging to obtain the temperature fusion values ​​inside and outside the greenhouse.

[0057] The weighted average calculation formula is as follows: ; in, is the temperature fusion value, i.e. the final temperature, For the The temperature measurement value of each temperature sensor, For the The weight of each temperature sensor, is the total number of temperature sensors. Weight It is determined and dynamically adjusted according to factors such as the accuracy, position, and stability of the temperature sensor to ensure the accuracy of the temperature fusion value.

[0058] Humidity sensor fusion technology: install several humidity sensors inside and outside the greenhouse, measure the humidity values ​​inside and outside the greenhouse respectively, fuse the humidity values ​​through weighted average method, and get the humidity fusion value inside and outside the greenhouse. The calculation formula of weighted average is the same as that of temperature sensor fusion technology. Replaced with humidity measurement , temperature fusion value Replaced with humidity fusion value That's it.

[0059] Light radiation sensor fusion technology: install several light radiation sensors inside and outside the greenhouse to measure the light radiation values ​​inside and outside the greenhouse respectively, and fuse the light radiation measurement values ​​through the weighted average method to obtain the light radiation fusion value inside and outside the greenhouse. The formula of the weighted average method is the same as that of the temperature fusion technology. Replaced with light radiation measurements , temperature fusion value Replaced with light radiation fusion value That's it.

[0060] Wind speed sensor fusion technology: Install several wind speed sensors inside and outside the greenhouse (near the vents) to measure the wind speed inside and outside the greenhouse respectively. The measured values ​​are fused by weighted average method to obtain the fusion value of wind speed inside and outside the greenhouse. The formula of weighted average method is the same as that of temperature sensor fusion technology. Replace with wind speed value , temperature fusion value Replaced with wind speed fusion value That's it.

[0061] Wind direction sensor fusion technology: Install several wind direction sensors inside and outside the greenhouse to measure the wind direction values ​​inside and outside the greenhouse respectively. The measured values ​​are fused through the vector averaging method to obtain the fusion value of the wind direction inside and outside the greenhouse. The formula of the vector averaging method is as follows: ; in, is the wind direction fusion value, For the The measurement value of a wind direction sensor, is the total number of wind direction sensors.

[0062] Image sensor fusion technology: Several image sensors are installed inside and outside the greenhouse to collect images inside and outside the greenhouse respectively. The captured images are fused through image processing algorithms to obtain fused images of the inside and outside of the greenhouse.

[0063] Among them, the image processing algorithm includes steps such as image alignment, image fusion, and image enhancement, as follows: Image alignment geometrically transforms the images inside and outside the greenhouse to make the image's perspective, size, direction, etc. consistent, which is convenient for subsequent image fusion. Image fusion fuses the images inside and outside the greenhouse at the pixel level or feature level to make the image's brightness, contrast, color, etc. uniform, improving the image's clarity and quality. Image enhancement filters, sharpens, and denoises the images inside and outside the greenhouse to eliminate image defects such as blur, noise, and distortion, and improve the image's details and information content.

[0064] Furthermore, the present invention can calculate and derive the following data as collaborative modeling variables based on the greenhouse structural parameter indicators and the above-mentioned integrated greenhouse internal and external meteorological environment data.

[0065] Greenhouse volume: Calculate the volume of the greenhouse in cubic meters based on the length, width, and height of the greenhouse. The calculation formula for greenhouse volume is as follows: ; in, is the greenhouse volume, is the greenhouse length, is the width of the greenhouse, is the greenhouse height.

[0066] Greenhouse height-to-span ratio: Calculate the greenhouse height-to-span ratio based on the height and span of the greenhouse. The unit is dimensionless. The formula for the greenhouse height-to-span ratio is as follows: ; in, is the height-to-span ratio of the greenhouse, is the greenhouse height, is the greenhouse span, that is, the maximum horizontal width of the greenhouse.

[0067] Greenhouse comprehensive thermal resistance: Calculate the comprehensive thermal resistance of the greenhouse based on the material, thickness, area and other parameters of each surface of the greenhouse. The unit is square meter·Kelvin / Watt. The formula for the comprehensive thermal resistance of the greenhouse is as follows: ; in, is the comprehensive thermal resistance of the greenhouse, For greenhouse The thermal resistance of the material on each surface, For greenhouse The surface area, is the total number of greenhouse surfaces.

[0068] Among them, the thermal resistance of the material The calculation is based on Fourier's law, and the formula is as follows: ; in, For greenhouse The thickness of the surface, For greenhouse Thermal conductivity of a surface.

[0069] Indoor and outdoor temperature difference: Calculate the indoor and outdoor temperature difference of the greenhouse based on the temperature fusion value inside and outside the greenhouse, in degrees Celsius. The formula for the indoor and outdoor temperature difference is as follows: ; in, is the temperature difference between indoor and outdoor, is the temperature fusion value in the greenhouse, is the temperature fusion value outside the greenhouse.

[0070] Light transmittance: The light transmittance of the greenhouse is calculated based on the fusion value of light radiation inside and outside the greenhouse, and the unit is percentage. The formula for light transmittance is as follows: ; in, is the transmittance, is the light radiation fusion value in the greenhouse, is the light radiation fusion value outside the greenhouse.

[0071] Saturated water vapor pressure difference VPD: According to the temperature fusion value and humidity fusion value inside and outside the greenhouse, the vapor pressure deficit (Vapor Pressure Deficit) of the greenhouse is calculated in kilopascals.

[0072] The calculation formula of VPD is as follows: ; in, is the steam pressure deficit, is the saturated vapor pressure in the greenhouse, is the actual vapor pressure in the greenhouse, is the temperature fusion value in the greenhouse, is the humidity integration value in the greenhouse.

[0073] The calculation of saturated vapor pressure and actual vapor pressure is carried out according to the following empirical formula: ; .

[0074] Based on the content of the above embodiment, as an optional embodiment, determining the greenhouse ventilation feature vector according to the real-time environmental information includes: Inputting the real-time environmental information into a multimodal variational autoencoder to obtain an initial feature vector related to the influencing factors of greenhouse ventilation and the law of temperature change; The initial feature vector is adaptively weighted and fused using a multimodal attention mechanism to obtain a greenhouse ventilation feature vector.

[0075] The present invention can adopt a multimodal variational autoencoder MVAE to jointly model and generate real-time environmental information of various data types collected inside and outside the greenhouse by using variational reasoning and latent variable models, extract and represent the influencing factors and temperature change laws related to greenhouse ventilation, and obtain the initial feature vector characterizing greenhouse ventilation. The function is to perform deep learning and feature extraction on greenhouse environmental data and provide input feature representation for greenhouse ventilation decision-making.

[0076] In order to improve the expressive power and information content of feature vectors, a multimodal attention mechanism is adopted. The attention weight and gating mechanism are used to adaptively weight and fuse the real-time environmental information of different modalities, highlight important and related information, and suppress redundant and noisy information.

[0077] The multimodal attention mechanism effectively improves the fusion effect of multimodal data and the representation ability of feature vectors. Its specific implementation steps include: A multimodal variational autoencoder is used to jointly model and generate various data types inside and outside the greenhouse using variational inference and latent variable models, extract and represent the influencing factors of greenhouse ventilation and the temperature change law, and obtain the feature vector of greenhouse ventilation.

[0078] The multimodal variational autoencoder deep generative model can process data of multiple modes at the same time, and can convert and complement between different modes. In the structure of the multimodal variational autoencoder, for Real-time environmental information in different modalities; is a latent variable, i.e., the output feature vector of greenhouse ventilation; For encoder; For the decoder; and is the model parameter.

[0079] The goal of the multimodal variational autoencoder is to maximize the marginal log-likelihood of the data, namely: ; Since the integral of the above formula is unsolvable, we can use the variational inference method to introduce the approximate posterior distribution , and maximize the variational lower bound (ELBO), that is: ; in, express expectations; represents KL divergence; is the prior distribution of the latent variable, which is usually assumed to be a standard normal distribution; is the variational lower bound.

[0080] For ease of calculation, the approximate posterior distribution Constructed in the following way: ; in, is the normalization constant; express K The latent variable z The product of the posterior distribution of ; The encoder for unimodal is usually assumed to be Gaussian distributed, that is: ; in, and is the output of the encoder, representing the mean and variance of the latent variable; represents a Gaussian distribution; decoder The design is based on the real-time environmental information types of different modalities. For example, for numerical data, Gaussian distribution is used, for text data, Bernoulli distribution is used, for image data, multinomial distribution is used, and so on.

[0081] By inputting real-time environmental information into the multimodal variational autoencoder, the initial feature vector related to the influencing factors of greenhouse ventilation and the temperature change law can be obtained. ,The initial feature vector can contain information of various types inside and outside the greenhouse, reflecting the influencing factors and temperature variation laws of greenhouse ventilation, and providing feature representation for greenhouse ventilation decision-making.

[0082] As an optional embodiment, the present invention adopts a multimodal attention mechanism, and uses attention weights and gating mechanisms to adaptively weight and fuse real-time environmental information of different modalities, highlighting important information and related information, and suppressing redundant information and noise information. The multimodal attention mechanism effectively improves the fusion effect of multimodal data and the representation ability of feature vectors.

[0083] Specifically, the following multimodal attention mechanisms can be adopted, mainly including: First, consider the attention weight: for each modality of real-time environmental information, calculate its contribution to the initial feature vector, that is, the attention weight, expressed as ,in . No. k Attention weights for real-time environment information of each modality The calculation is performed by the following formula: ; in, is a learnable function used to evaluate the The importance of real-time environmental information of various modalities can be achieved using neural networks or other methods. The larger the value, the The greater the contribution of the real-time environmental information of a modality to the initial feature vector, the more it should be considered and integrated.

[0084] Then, consider the gating mechanism: for each modality of real-time environmental information, calculate its reliability for the initial feature vector, that is, the gating mechanism, expressed as ,in The calculation of the gating mechanism is performed in the following way: ; in, is the sigmoid function, is a learnable function used to evaluate the The reliability of real-time environmental information of this modality depends on whether it is implemented using neural networks or other methods. The closer the value is to 1, the The more reliable the real-time environmental information of a modality is, the more it should be retained and integrated.

[0085] Finally, based on the attention weights and gating mechanism determined above, the real-time environmental information of different modalities is adaptively weighted and fused to obtain the final greenhouse ventilation feature vector, which is expressed as: ; in, For the The latent variables obtained by passing the data of the modalities through the unimodal encoder are: ; in, is Gaussian noise, is element-wise multiplication; Indicates k Data of different modalities x ( k ) After passing through the unimodal encoder, the latent variable z ( k )’s mean (or expectation); Indicates k Data of different modalities x ( k ) After passing through the unimodal encoder, the latent variable z ( k ) is the covariance matrix (or variance, if the covariance matrix is ​​assumed to be diagonal or a scalar) of

[0086] Through multimodal fusion, a more integrated feature vector is obtained ,The feature vector can integrate the information of data of different modes, highlight the important and related information, suppress the redundant and noisy information, and provide a more effective feature representation for ventilation decision-making.

[0087] The multi-objective solar greenhouse ventilation decision method provided by the present invention uses a multimodal variational autoencoder (MVAE) algorithm to extract and represent features from data from multiple sensors and generate feature vectors for greenhouse ventilation. The innovative data processing method can effectively reduce the dimension and complexity of data, improve the expression and differentiation capabilities of data, and has higher accuracy and adaptability than traditional data layer fusion and feature layer fusion methods, and can adapt to different types of incomplete and uncertain data.

[0088] As an optional embodiment, in the multi-objective solar greenhouse ventilation decision method provided by the present invention, a preliminary ventilation decision strategy is generated according to the greenhouse ventilation feature vector, mainly including: Based on the soft behavior cloning algorithm, the objective function is constructed by maximizing the reward and minimizing the difference between the output action and the expert strategy; Determining a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector and a reward function determined by the objective function; According to the preliminary ventilation decision strategy, an output action for greenhouse ventilation is obtained.

[0089] Among them, the objective function includes a policy network and a value network constructed respectively by using two neural networks; the policy network outputs Gaussian distribution parameters, and the Gaussian distribution parameters are used to characterize the probability distribution of the output action under the greenhouse ventilation feature vector; the value network outputs a scalar, and the scalar is used to characterize the expected reward value under the greenhouse ventilation feature vector.

[0090] As an optional embodiment, the present invention adopts a soft actor-critic (SAC) algorithm and a dual neural network to automatically learn and optimize the greenhouse ventilation decision strategy according to the feature vector and reward function of greenhouse ventilation, and obtain the output action of greenhouse ventilation. The function is to perform decision learning and strategy optimization on the feature vector of greenhouse ventilation, and provide output actions for greenhouse ventilation decision.

[0091] In the structure of soft behavior cloning, is the greenhouse ventilation feature vector for greenhouse ventilation, Output action for greenhouse ventilation, is the reward function for greenhouse ventilation, For the strategy network, For the value network, and are model parameters. The goal of soft behavior cloning is to maximize the following objective function: ; in, Represents a variable expectations; Demonstrate datasets for experts; and is a hyperparameter; is the entropy function; is the KL divergence; For expert strategies.

[0092] The objective function is to minimize the difference from the expert policy while maximizing the reward and policy entropy. The role of policy entropy is to increase the exploratory and robustness of the policy and avoid overfitting the expert data. The role of KL divergence is to constrain the consistency of the policy with the expert data and avoid deviation from the expert data.

[0093] It should be noted that the present invention adopts a dual neural network to optimize the objective function, that is, two neural networks are used to represent the policy network and the value network respectively. The policy network outputs the parameters of the Gaussian distribution, which represents the probability distribution of the output action under a given feature vector; the value network outputs a scalar, which represents the expected reward value under a given feature vector and output action.

[0094] The parameters of the two networks are updated using the policy gradient method and the least squares method, respectively, as follows: ; ; in, and are the parameters of the strategy network and the value network respectively; or is the learning rate of the policy network; is the learning rate of the value network; is the objective function of the policy network, which is used to measure the policy network under given parameters i and ϕ Performance under is the objective function of the value network. Given the parameters ϕ ,state x and actions a The output below indicates that from the state x Start executing the action a Expected return after From the state x Execute an action a The expected reward value obtained after.

[0095] For the policy network, the parameters i Through policy gradient Update. The policy gradient is the objective function about i The gradient of i To maximize .

[0096] For the value network, the parameters ϕ By minimizing the predicted Q value With real rewards The squared error between is used to update. The gradient of this error Instructions on how to adjust ϕ To reduce this error. Note that this assumes is part of the true or target Q value, and a target network can also be used to stabilize the learning process.

[0097] The multi-objective solar greenhouse ventilation decision-making method provided by the present invention adopts a soft behavior cloning imitation learning method, which can simultaneously utilize expert demonstration data and environmental feedback data to learn a randomized strategy so that it can maximize the reward while being as close to the expert strategy as possible.

[0098] Based on the content of the above embodiment, as an optional example, after constructing the objective function, the method further includes: The objective function is optimized based on the entropy regularization method, including adding an entropy term to the objective function. The expression of the entire entropy term calculates the negative expectation of these logarithmic probabilities, that is, the entropy of the policy. The larger the entropy, the more average the probability of the policy network to choose different actions in a given state, that is, the higher the uncertainty of the policy, which is helpful for exploration. By adding the entropy term to the objective function and multiplying it by the hyperparameter α , we can control the trade-off between exploration and exploitation.

[0099] The expression of the entropy term is: ; The expression of the optimized objective function is: ; in, represents the expectation of the variable, Indicates action a expectations; Greenhouse ventilation feature vector for greenhouse ventilation; The output action for greenhouse ventilation is from the policy network In a given state x Sampling from the following distribution; Reward function for greenhouse ventilation; For strategic networks; and are model parameters; Demonstrate datasets for experts; and is a hyperparameter; is the entropy function; is the KL divergence; For expert strategies; is the entropy term of entropy regularization; is the objective function; is the policy network in a given state x Next select action a The logarithmic probability of .

[0100] The objective function also includes the reward function and KL divergence The reward function is used to evaluate how good a particular action is in a given state, while the KL divergence is used to measure the policy network With expert strategies This helps the policy network mimic expert behavior. The degree of imitation of expert behavior was controlled.

[0101] Based on the content of the above embodiment, as an optional embodiment, after determining the preliminary ventilation decision strategy, it also includes: Multi-task learning and meta-gradient updating are used to adapt the greenhouse ventilation decision strategy according to the current greenhouse environment and crop types.

[0102] Based on the above embodiments, the present invention adopts meta-learning to utilize multi-task learning and meta-gradient updating to quickly adapt and adjust the greenhouse ventilation decision-making strategy, adapt to the uncertainty and variability of the environment inside and outside the greenhouse, and achieve long-term optimal control of the temperature in the greenhouse.

[0103] Meta-learning can utilize data from multiple related tasks to learn a general initial model, which can be quickly adjusted and optimized on new tasks, thereby improving learning efficiency and adaptability.

[0104] Different scenarios or conditions related to greenhouse ventilation, such as different seasons, climates, crops, etc., are regarded as different tasks. Each task has a corresponding feature vector (such as temperature, humidity, light intensity, crop growth stage, etc.), output action (such as ventilation volume, ventilation time, etc.) and reward function (such as rewards defined based on crop growth conditions, energy consumption, etc.), which constitutes a multi-task learning problem.

[0105] The purpose of multi-task learning is to use the similarities and complementarities between different tasks to learn a common initial model that performs well on different tasks while also making personalized adjustments based on the characteristics of different tasks.

[0106] The meta-gradient update method is used to optimize the parameters of the initial model so that it can quickly adapt and optimize on new tasks.

[0107] The meta-gradient update method refers to using a meta-optimizer to calculate the gradient of the initial model parameters according to the loss functions of different tasks, and then using the meta-learning rate to update the initial model parameters to obtain an updated model, which continues to perform gradient descent on the new task to obtain the final model, which achieves the best performance on the new task. The following is a detailed description of how to use multi-task learning and meta-gradient updates to adapt the greenhouse ventilation decision strategy according to the current greenhouse environment and crop types, including: First, a multi-task learning framework is constructed, including: designing feature vectors that can fully reflect the greenhouse environment and crop status, which will be used as the input of the model; building a shared model structure (such as a neural network) that contains multiple layers to extract useful information from the input features; after the shared model, adding specific layers (such as fully connected layers) for each task to output specific actions for the task; defining loss functions for each task, which are calculated based on the reward function of the task and the actions predicted by the model.

[0108] Then perform meta-learning initialization, including: training a general initial model using meta-learning methods ϕ This initial model ϕ Pre-train on multiple tasks to learn commonalities and basic knowledge between different tasks. Choose a meta-optimizer (such as MAM, Reptile, etc.) to optimize the parameters of the initial model on multiple tasks.

[0109] Then, the meta-gradient update is performed. When faced with a new greenhouse ventilation task, the meta-gradient update method is used to adjust the initial model. ϕ , to quickly adapt to new tasks, including: ϕ To begin, create a task-specific model for each new task i (usually by copying ϕ Get); on each new task, use the task-specific model i Train to minimize the loss function for this task L ( i ) ; Use the meta-optimizer to calculate the initial model parameters based on the loss function on all tasks ϕ The gradient of . This gradient takes into account the initial model of all tasks ϕ The effect of ϕ To better adapt to new tasks; use meta-learning rate α Update initial model parameters ϕ ,Right now ; Use the updated initial model ϕ As a starting point, each new task is further optimized to obtain the final model i ,Right now .

[0110] in, are the initial model parameters, are the updated model parameters, is the loss function of the new task, is the meta-learning rate, is the normal learning rate.

[0111] By training meta-learning, a general initial model is obtained , this initial model Pre-training on multiple tasks to learn commonalities and basic knowledge between different tasks. Rapidly adapt and optimize on different greenhouse ventilation tasks to improve the robustness and generalization of greenhouse ventilation decision strategies.

[0112] Finally, according to the ventilation index of greenhouse ventilation, the following data can be selected as the following output actions: Window opening time, a binary variable indicating whether greenhouse ventilation is performed Indicates that Indicates opening the window. Indicates not to open the window. Indicates that the window is not open.

[0113] The opening size indicates the degree of window opening for greenhouse ventilation, which is a continuous percentage variable. Indicates that Indicates complete shutdown. Indicates fully open. The larger the value, the larger the window opening and the greater the ventilation volume.

[0114] Duration, which indicates the window opening time for greenhouse ventilation, is expressed as a continuous variable in seconds, where Indicates not to open the window. The larger the value, the longer the window is opened and the greater the ventilation volume.

[0115] The multi-objective solar greenhouse ventilation decision-making method provided by the present invention uses the soft behavioral cloning (SAC) algorithm and combines the meta-learning technology to automatically learn and optimize the greenhouse ventilation decision-making strategy. The innovative decision-making optimization method can effectively balance the temperature control and crop growth requirements in the greenhouse, as well as energy consumption and utilization. Compared with the traditional set value method and mathematical model method, it has higher efficiency and performance, and can automatically adapt to the meteorological changes inside and outside the greenhouse and the growth stage of the crop.

[0116] In summary, the present invention uses multi-sensor fusion technology, multi-modal variational autoencoder algorithm, soft behavior cloning algorithm, meta-learning technology, etc., to comprehensively consider multiple objectives such as meteorological parameters inside and outside the greenhouse, physiological needs of crops, and energy benefits of active heat storage and release, so as to achieve multi-objective optimization and coordination of greenhouse ventilation, and can effectively balance the temperature control in the greenhouse and the growth needs of crops, as well as energy consumption and utilization, to achieve long-term optimal control of the temperature in the greenhouse, improve energy utilization efficiency and crop growth quality, increase the economic benefits of agriculture, achieve long-term optimal control of the temperature in the greenhouse, and improve energy utilization efficiency and crop growth quality.

[0117] Based on the content of the above embodiment, as an optional embodiment, the preliminary ventilation decision strategy is optimized by multiple objectives, mainly including: The multi-objective particle swarm optimization algorithm and the multi-objective co-evolution algorithm are used to perform multi-objective optimization on the preliminary decision-making actions, and the objectives of greenhouse temperature control, crop growth requirements, operating costs and effects are comprehensively considered to obtain the optimal ventilation decision strategy.

[0118] The present invention adopts a multi-objective particle swarm optimization algorithm (MOPSO) algorithm, utilizes particle swarm intelligence and a leader selection mechanism, comprehensively considers the temperature control and crop growth requirements in the greenhouse, as well as the operating costs and effects of window ventilation and active heat storage and release devices, establishes and solves the multi-objective function of the greenhouse ventilation optimization problem, and obtains the optimal solution for greenhouse ventilation.

[0119] Specifically, in MOPSO, each particle represents a potential solution, and each particle has its own position and velocity. In the structure of the multi-objective particle swarm optimization algorithm, for Different modalities of temporal and environmental information; is a hidden variable; For encoder; For the decoder; and is the model parameter.

[0120] The goal of the multi-objective particle swarm optimization algorithm is to maximize the marginal log-likelihood of the data, that is: ; Since the integral of the above formula is unsolvable, the variational inference method is used to introduce the approximate posterior distribution , and maximize the variational lower bound (ELBO), that is: ; in, represents the KL divergence, is the prior distribution of the latent variable, which is usually assumed to be a standard normal distribution.

[0121] For ease of calculation, the approximate posterior distribution Constructed in the following way: ; in, is the normalization constant, For a unimodal encoder, it is usually assumed to be a Gaussian distribution, that is: ; in, and is the output of the encoder, representing the mean and variance of the latent variable. The design is based on the data types of different modalities. For example, for numerical data, Gaussian distribution is used, for text data, Bernoulli distribution is used, for image data, multinomial distribution is used, and so on.

[0122] By training the multi-objective particle swarm optimization algorithm, the initial feature vector of greenhouse ventilation is obtained. ,The initial feature vector can contain information of various types inside and outside the greenhouse, reflecting the influencing factors and temperature variation laws of greenhouse ventilation, and providing feature representation for greenhouse ventilation decision-making.

[0123] This step adopts the multi-objective co-evolutionary algorithm (MOCCA), which uses co-evolution and decomposition strategies to decompose the multi-objective functions and decision variables of the greenhouse ventilation optimization problem into multiple sub-problems and sub-components, and optimizes and coordinates them separately to achieve efficient solution of the greenhouse ventilation optimization problem and maintenance of high-quality solutions.

[0124] MOCCA is a multi-objective optimization method based on collaborative evolution. It decomposes the multi-objective optimization problem into multiple cooperative sub-problems and uses different populations to solve different sub-problems, thereby improving the search efficiency and the diversity of solutions.

[0125] In the structure of MOCCA, for The objective function, for decision variables, for A coevolving population, An external archive for storing non-dominated solutions.

[0126] The goal of the multi-objective co-evolutionary algorithm is to solve the following multi-objective optimization problem: ; ; in, For decision space, is the target vector.

[0127] To facilitate calculation, the multi-objective co-evolution algorithm decomposes the multi-objective optimization problem into the following form: ; ; in, is the number of the subproblem, For the The decision variables of each subproblem are For the Decision variables other than the subproblem, For the The decision space of each subproblem is For the Decision space beyond the subproblem, For the The target vector of each subproblem, For the The first sub-question An objective function.

[0128] Each subproblem consists of a population To solve, population The size is ,in is a hyperparameter used to control the search space of each subproblem. Each individual is represented by ,in is the individual number, For the The decision variables of individuals. Each individual has its own fitness value, expressed as ,in For the The target vector of each individual is: ; in, For the The covariate of each individual, Decision variables other than the subproblem, The values ​​of are provided by individuals from other populations, as follows: ; in, is the number of individuals in other populations, ,in and In this way, each individual collaborates with individuals from other populations to form a complete solution, thereby performing fitness evaluation and optimization.

[0129] Furthermore, the co-evolution method is used to make different populations cooperate and compete with each other, thereby improving the search efficiency and the diversity of solutions. The co-evolution method includes the following steps: (1) Initialization: For each population , randomly generated Individuals, that is ,in , and initialize it to a random number that obeys a uniform distribution, that is: ; in, For the The decision space of each subproblem is For uniform distribution.

[0130] Then, for each individual, randomly select individuals from other populations as its co-variables, that is: ; in, is the number of the randomly selected individual; and .

[0131] Then, according to the individual decision variables and collaborative variables, its target vector and fitness value are calculated, namely: ; (2) Evolution: For each population , do the following: Using the Tournament Selection method, Select two individuals as parent individuals, namely: ; ; in, The function selected for the tournament proceeds as follows: From the population Randomly select two individuals from and ,in and .

[0132] Compare the fitness values ​​of two individuals, that is and , according to the Pareto Dominance relationship, select the individual as the winner and return its decision variable, that is: ; in, Represents the Pareto dominance relationship, namely: ; represents the Pareto non-dominated relationship, namely: ; (3) Variation: Using the Gaussian Mutation method, the decision variables of the parent individuals are randomly disturbed to obtain the decision variables of the offspring individuals, namely: ; ; in, and is a random number that obeys a Gaussian distribution, that is: ; ; in, and is the standard deviation of the variation, and is a hyperparameter used to control the intensity of the variation.

[0133] Then, the decision variables of the offspring individuals are checked for boundaries so that they meet the constraints of the decision space, namely: ; ; in, and For the The upper and lower bounds of the decision space of each subproblem are: ; (4) Crossover: Using uniform crossover, the gene at the above position of chromosome 1 of the parent is copied to the same position of offspring 1, and then the missing genes in offspring 1 are filled in order on chromosome 2 of the parent. Another offspring is obtained in a similar manner.

[0134] Furthermore, according to the optimal solution of greenhouse ventilation determined in the above steps, the operating parameters of window ventilation and active heat storage and release device are determined, such as window opening time, opening size and duration, as well as inlet and outlet temperature difference, soil heat storage, water wall heat storage, etc., to achieve coordinated control and balance of window ventilation and active heat storage and release device. The function is to control and coordinate the operation of window ventilation and active heat storage and release device according to the optimal solution of greenhouse ventilation, so as to meet the temperature control and crop growth requirements in the greenhouse.

[0135] In order to improve the coordinated control and balance of window ventilation and active heat storage and release devices, the present invention adopts a multi-agent system (MAS), which uses multiple agents and coordination mechanisms to perform distributed control and collaboration on window ventilation and active heat storage and release devices, so as to coordinate and meet the temperature control in the greenhouse and the needs of crop growth. MAS effectively solves the complexity and dynamics of the coordinated control and balance of window ventilation and active heat storage and release devices.

[0136] Based on the content of the above embodiment, as an optional embodiment, the dynamic adjustment of the optimal ventilation decision strategy to automatically adapt to the meteorological changes inside and outside the greenhouse and the growth stage of the crops includes: By using a multi-agent system and transfer learning technology, the optimal ventilation decision strategy is optimized in real time by utilizing the similarities and differences between the source task and the target task; The target task is determined based on the meteorological changes inside and outside the greenhouse and the growth stage of the crops.

[0137] Specifically, the present invention determines the operating parameters of window ventilation and active heat storage and release device, such as window opening time, opening size and duration, as well as inlet and outlet temperature difference, soil heat storage, water wall heat storage, etc., according to the optimal solution of greenhouse ventilation, so as to realize the coordinated control and balance of window ventilation and active heat storage and release device. The function is to control and coordinate the operation of window ventilation and active heat storage and release device according to the optimal solution of greenhouse ventilation, so as to meet the temperature control and crop growth requirements in the greenhouse.

[0138] In order to improve the coordinated control and balance of window ventilation and active heat storage and release devices, this step adopts a multi-agent system (MAS), which uses multiple agents and coordination mechanisms to perform distributed control and collaboration on window ventilation and active heat storage and release devices, so as to coordinate and meet the temperature control and crop growth needs in the greenhouse. MAS effectively solves the complexity and dynamics of the coordinated control and balance of window ventilation and active heat storage and release devices, and the specific contents include: (1) Transfer learning of reinforcement learning

[0139] The method of transfer learning based on reinforcement learning is adopted to effectively migrate and adapt the greenhouse ventilation decision-making strategy by utilizing the similarities and differences between the source task and the target task, so as to achieve rapid updating and improvement of the greenhouse ventilation decision-making strategy.

[0140] Transfer learning is a learning method based on reinforcement learning that improves learning efficiency and performance by applying knowledge learned in one environment to another related environment.

[0141] In the structure of transfer learning of reinforcement learning, is the source task, For the target task, is the data in the source task, is the data in the target task, is the strategy in the source task, is the strategy in the target task, is the function of transfer learning, is the loss function of transfer learning. The goal of transfer learning in reinforcement learning is to use data from the source task to and strategies , and the data in the target task , learn the function of transfer learning , so that the strategy in the target task Able to achieve optimal performance in the target task, namely: ; ; in, is the loss function of transfer learning, which is used to measure the strategy in the target task With the optimal strategy The gap between is the optimal strategy in the target task, that is: ; in, is the initial state distribution in the target task, is the state-action value function in the target task, indicating that Next action After that, according to the strategy The expected return that can be obtained.

[0142] By training the transfer learning of reinforcement learning, we can obtain the strategy in the target task ,This strategy can utilize the knowledge in the source task and the data in the target task to ,realize the rapid update and improvement of greenhouse ventilation ,decision-making strategy.

[0143] (2) Greenhouse ventilation decision-making strategy

[0144] The greenhouse ventilation decision problem is modeled as a reinforcement learning problem, namely a Markov decision process (MDP), using a four-tuple Indicates. Among them, The state space represents the real-time environmental information inside and outside the greenhouse, including the temperature, humidity, light, wind speed, etc. inside and outside the greenhouse, as well as the crop growth indicators in the greenhouse; is the action space, which represents the control action of greenhouse ventilation, including window opening time, opening size and duration, and the operating parameters of the active heat storage and release device; is the state transition probability, indicating that in state Next action Then, transfer to the next state The probability of ; is the reward function, which means that in the state Next action After that, the immediate reward you can get is ,in It is a reward signal, which is usually related to the error of temperature control in the greenhouse, the degree to which crop growth needs are met, the degree of energy consumption savings, etc.

[0145] Greenhouse ventilation decision strategy is a mapping from state to action, indicating that in state Next select action The probability of .

[0146] The goal of the greenhouse ventilation decision strategy is to maximize the expectation of the cumulative reward, that is: ; in, is the initial state distribution, is the discount factor, For the The reward signal for each time step.

[0147] The present invention dynamically adjusts the greenhouse ventilation decision-making strategy through multi-agent systems and reinforcement learning transfer learning technology, can automatically adapt to meteorological changes inside and outside the greenhouse and the growth stage of crops, improve the greenhouse's adaptability and robustness, reduce the greenhouse's operating risks, achieve energy saving and environmental protection in the greenhouse, realize the intelligence and adaptability of greenhouse ventilation decisions, and improve the greenhouse's adaptability and robustness.

[0148] As an optional embodiment, an embodiment is provided below to illustrate the multi-objective solar greenhouse ventilation decision-making method provided by the present invention, which mainly includes but is not limited to the following three steps: (1) Data acquisition

[0149] Get real-time environmental information inside and outside the greenhouse, including: Temperature is an important factor affecting greenhouse ventilation. Monitor the temperature changes inside and outside the greenhouse, as well as the temperature distribution in different areas of the greenhouse. Use temperature sensors such as DS18B20 and LM35 to collect temperature data. Distribute temperature sensors in appropriate locations inside and outside the greenhouse, such as the top, middle and bottom of the greenhouse, as well as the southeast, northwest and northeast directions of the greenhouse. The number and density of temperature sensors also need to be considered to ensure data coverage and accuracy. The unit symbol of temperature data is degrees Celsius (℃), the value range is -40℃~125℃, and the acquisition frequency is once per minute.

[0150] Humidity is another important factor affecting greenhouse ventilation. Humidity sensors such as DHT11 or HIH-4000 are used to collect humidity data. Humidity sensors are distributed in appropriate locations inside and outside the greenhouse. The unit symbol of humidity data is percentage (%), the value range is 0%~100%, and the collection frequency is every minute.

[0151] Monitor the changes in light radiation inside and outside the greenhouse and the distribution of light radiation in different areas of the greenhouse. Use light sensors to collect light radiation data. Distribute light sensors at appropriate locations inside and outside the greenhouse. The unit symbol is lux (lx), the value range is 0 lx~65535 lx, and the collection frequency is once per minute.

[0152] In addition to the above three data types, you can also choose to obtain other data types based on actual conditions, such as: Carbon dioxide concentration: The unit symbol is parts per million (ppm), the value range is 0 ppm ~ 10000 ppm, and the collection frequency is once per minute.

[0153] Wind speed: The unit symbol is meter per second (m / s), the value range is 0 m / s ~ 30 m / s, and the collection frequency is once per minute.

[0154] The above data can be obtained through the following methods or channels: Build a sensor network yourself: Purchase and install sensor devices by yourself, and transmit sensor data to a data center or cloud server wirelessly or wired. Consider the cost, maintenance, compatibility, stability, and security of sensor devices.

[0155] Use third-party data services: Use third-party data service providers, such as Huawei Cloud, Alibaba Cloud, or Tencent Cloud, to obtain sensor data. Comprehensively consider factors such as the cost, quality, reliability, availability, and privacy of data services.

[0156] (2) Data processing

[0157] The real-time environmental information is processed as follows to serve as the input of the subsequent model: Data cleaning includes quality checks on data related to real-time environmental information, and the removal of invalid, abnormal, duplicate, missing, and erroneous data to improve data accuracy and completeness. Use Python and related libraries such as pandas or numpy to clean data. Use the pandas.DataFrame.dropna() function to delete rows containing null values, and use the pandas.DataFrame.drop_duplicates() function to delete duplicate rows.

[0158] Data conversion includes converting the data format, type, range, scale, etc. to meet the input requirements of the model. Use Python language pandas or scikit-learn to convert data. Use the pandas.DataFrame.astype() function to change the data type, and use the sklearn.preprocessing.MinMaxScaler() function to normalize the data to the [0, 1] interval.

[0159] Data fusion, including the fusion of multi-source, multi-modal, and multi-time series data, improves the amount of information and expression of data. Use pandas or PyTorch to perform data fusion. Use the pandas.DataFrame.merge() function to merge data from different sources, and use the torch.cat() function to splice data from different modes.

[0160] Data storage, including structured or unstructured storage of data, using libraries such as sqlite3 or h5py for data storage. Use the sqlite3.connect() function to create a database file, and use the h5py.File() function to create an HDF5 file.

[0161] (3) Model construction

[0162] In order to realize the multi-objective solar greenhouse ventilation decision of the present invention, the following three model components are mainly constructed: Multimodal variational autoencoder (MVAE), as a deep generative model, performs joint learning and feature extraction on multimodal data to generate feature vectors for greenhouse ventilation. Use PyTorch language and related libraries such as torch and torchvision to build the MVAE model. Define the network structure of the encoder and decoder of MVAE as well as the loss function and optimizer of MVAE. The core parameter configuration of MVAE is shown in the following table:

[0163] The Soft Behavior Cloning (SAC) model uses a policy-based reinforcement learning algorithm to automatically learn and optimize greenhouse ventilation decision strategies. Use PyTorch languages ​​and related libraries such as torch and stable-baselines3 to build the SAC model, define the network structure of the SAC policy network and value function network, as well as the SAC reward function and optimizer. The core parameter configuration of SAC is shown in the following table:

[0164] Meta-Learning enables the model to quickly adapt to new tasks or environments, improving the generalization and robustness of the model. Use PyTorch languages ​​such as torch and related libraries to implement meta-learning technology. Select appropriate meta-learning algorithms such as MAML, Reptile, and appropriate meta-learning tasks such as greenhouse ventilation decision-making tasks or greenhouse temperature prediction tasks. The core parameter configuration of meta-learning is shown in the following table:

[0165] Figure 2 : is a schematic diagram of the structure of the multi-objective solar greenhouse ventilation decision-making device provided by the present invention. Figure 2 As shown, it mainly includes: An information collection unit 21 is used to obtain real-time environmental information inside and outside the greenhouse; An information processing unit 22, used to determine a greenhouse ventilation feature vector according to the real-time environmental information; A strategy generating unit 23, used for generating a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector; A strategy optimization unit 24 is used to perform multi-objective optimization on the preliminary ventilation decision strategy so that the optimized optimal ventilation decision strategy takes into account the temperature control in the greenhouse, the crop growth requirements, the operation cost and the effect; The strategy matching unit 25 is used to dynamically adjust the optimal ventilation decision strategy to automatically adapt to the meteorological changes inside and outside the greenhouse and the growth stage of the crops.

[0166] It should be noted that the multi-objective solar greenhouse ventilation decision-making device provided by the present invention can execute the multi-objective solar greenhouse ventilation decision-making method described in any of the above embodiments during specific operation, which will not be elaborated in this embodiment.

[0167] The multi-objective solar greenhouse ventilation decision-making device provided by the present invention comprehensively utilizes multi-sensor fusion technology, multi-modal variational autoencoder algorithm, soft behavior cloning algorithm, meta-learning technology, multi-objective particle swarm optimization algorithm, multi-objective collaborative evolution algorithm, multi-agent system and reinforcement learning transfer learning technology, etc., which can effectively improve the efficiency of greenhouse ventilation decision-making, reduce greenhouse ventilation energy consumption, and improve the intelligent control of greenhouse temperature and crop growth quality.

[0168] Figure 3 is a schematic diagram of the structure of the electronic device provided by the present invention, such as Figure 3 As shown, the electronic device may include: a processor 310, a communication interface 320, a memory 330 and a communication bus 340, wherein the processor 310, the communication interface 320 and the memory 330 communicate with each other through the communication bus 340. The processor 310 may call the logic instructions in the memory 330 to execute a multi-objective solar greenhouse ventilation decision method, the method comprising: obtaining real-time environmental information inside and outside the greenhouse; determining a greenhouse ventilation feature vector according to the real-time environmental information; generating a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector; performing multi-objective optimization on the preliminary ventilation decision strategy so that the optimized optimal ventilation decision strategy takes into account the temperature control in the greenhouse, the crop growth requirements, the operating cost and the effect; and dynamically adjusting the optimal ventilation decision strategy to automatically adapt to the meteorological changes inside and outside the greenhouse and the growth stage of the crop.

[0169] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0170] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the multi-objective daylight greenhouse ventilation decision method provided by the above-mentioned embodiments, and the method includes: obtaining real-time environmental information inside and outside the greenhouse; determining the greenhouse ventilation characteristic vector based on the real-time environmental information; generating a preliminary ventilation decision strategy based on the greenhouse ventilation characteristic vector; performing multi-objective optimization on the preliminary ventilation decision strategy so that the optimized optimal ventilation decision strategy takes into account the temperature control in the greenhouse, the crop growth requirements, the operating costs and the effects; and dynamically adjusting the optimal ventilation decision strategy to automatically adapt to the meteorological changes inside and outside the greenhouse and the growth stage of the crops.

[0171] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the multi-objective daylight greenhouse ventilation decision method provided in the above-mentioned embodiments, the method comprising: obtaining real-time environmental information inside and outside the greenhouse; determining a greenhouse ventilation characteristic vector based on the real-time environmental information; generating a preliminary ventilation decision strategy based on the greenhouse ventilation characteristic vector; performing multi-objective optimization on the preliminary ventilation decision strategy so that the optimized optimal ventilation decision strategy takes into account temperature control in the greenhouse, crop growth requirements, operating costs and effects; and dynamically adjusting the optimal ventilation decision strategy to automatically adapt to meteorological changes inside and outside the greenhouse and the growth stage of crops.

[0172] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0173] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0174] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A multi-objective solar greenhouse ventilation decision-making method, characterized in that: include: Get real-time environmental information inside and outside the greenhouse; Determining a greenhouse ventilation characteristic vector according to the real-time environmental information; generating a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector; Performing multi-objective optimization on the preliminary ventilation decision strategy so that the optimized optimal ventilation decision strategy takes into account the temperature control in the greenhouse, the crop growth requirements, the operating cost and the effect; The optimal ventilation decision strategy is dynamically adjusted to automatically adapt to the meteorological changes inside and outside the greenhouse and the growth stage of the crops.

2. The multi-objective solar greenhouse ventilation decision-making method according to claim 1 is characterized in that: The real-time environmental information includes at least one of the following environmental data: indoor and outdoor temperature, indoor and outdoor humidity, indoor and outdoor light radiation value, wind speed, wind direction, indoor and outdoor images, greenhouse volume, greenhouse height-span ratio, greenhouse comprehensive thermal resistance, indoor and outdoor temperature difference, light transmittance, and saturated water vapor pressure difference; The real-time environmental information inside and outside the greenhouse is obtained, including: The temperature measurements inside and outside the greenhouse are combined by weighted average method to obtain the temperature inside and outside the greenhouse; The temperature and humidity measurement values ​​inside and outside the greenhouse are integrated by weighted average method to obtain the temperature and humidity inside and outside the greenhouse; The light radiation measurement values ​​inside and outside the greenhouse are merged by weighted average method to obtain the light radiation inside and outside the greenhouse; The wind speed measurements inside and outside the greenhouse are merged by weighted average method to obtain the wind speed inside and outside the greenhouse; The wind direction measurements inside and outside the greenhouse are merged by vector averaging method to obtain the wind direction inside and outside the greenhouse; The real-time images taken inside and outside the greenhouse are fused at the pixel level or feature level to obtain the indoor and outdoor images; Calculate the volume of the greenhouse according to the length, width, height and other dimensions of the greenhouse; Calculating the height-to-span ratio of the greenhouse according to the height and span of the greenhouse; Calculating the comprehensive thermal resistance of the greenhouse based on the material thermal resistance and surface area of ​​each surface of the greenhouse; Calculating the indoor and outdoor temperature difference according to the indoor and outdoor temperatures of the greenhouse; Calculating the light transmittance according to the indoor and outdoor light radiation values ​​of the greenhouse; The saturated water vapor pressure difference is calculated according to the temperature in the greenhouse and the humidity in the greenhouse.

3. The multi-objective solar greenhouse ventilation decision-making method according to claim 1 is characterized in that: Determining the greenhouse ventilation feature vector according to the real-time environmental information includes: Inputting the real-time environmental information into a multimodal variational autoencoder to obtain an initial feature vector related to the influencing factors of greenhouse ventilation and the law of temperature change; The initial feature vector is adaptively weighted and fused using a multimodal attention mechanism to obtain a greenhouse ventilation feature vector.

4. The multi-objective solar greenhouse ventilation decision-making method according to claim 3 is characterized in that: The multimodal attention mechanism is used to adaptively weight and fuse the initial feature vector to obtain a greenhouse ventilation feature vector, including: Determining attention weights and gating mechanisms for different modalities of real-time environmental information constituting the initial feature vector; According to the attention weight and gating mechanism, the real-time environmental information of different modalities is adaptively weighted and fused to obtain the greenhouse ventilation feature vector.

5. The multi-objective solar greenhouse ventilation decision-making method according to claim 1 is characterized in that: The generating a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector comprises: Based on the soft behavior cloning algorithm, the objective function is constructed by maximizing the reward and minimizing the difference between the output action and the expert strategy; Determining a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector and a reward function determined by the objective function; According to the preliminary ventilation decision strategy, an output action for greenhouse ventilation is obtained; The objective function includes a policy network and a value network respectively constructed by using two neural networks; the policy network outputs Gaussian distribution parameters, and the Gaussian distribution parameters are used to characterize the probability distribution of the output action under the greenhouse ventilation feature vector; the value network outputs a scalar, and the scalar is used to characterize the expected reward value under the greenhouse ventilation feature vector.

6. The multi-objective solar greenhouse ventilation decision-making method according to claim 5 is characterized in that: After constructing the objective function, it also includes: The objective function is optimized based on the entropy regularization method, including adding an entropy term to the objective function, and the expression of the entropy term is: ； The expression of the optimized objective function is: ； in, Express expectations about variables; Greenhouse ventilation eigenvector for greenhouse ventilation; Output action for greenhouse ventilation; Reward function for greenhouse ventilation; For strategic networks; and are model parameters; Demonstrate datasets for experts; and is a hyperparameter; is the entropy function; is the KL divergence; For expert strategies; is the entropy term of entropy regularization; is the objective function.

7. The multi-objective solar greenhouse ventilation decision-making method according to any one of claims 5 or 6, characterized in that: After determining the initial ventilation decision strategy, it also includes: Multi-task learning and meta-gradient updating are used to adapt the greenhouse ventilation decision strategy according to the current greenhouse environment and crop types.

8. The multi-objective solar greenhouse ventilation decision-making method according to claim 1 is characterized in that: The preliminary ventilation decision strategy is optimized by multiple objectives, including: The multi-objective particle swarm optimization algorithm and the multi-objective co-evolution algorithm are used to perform multi-objective optimization on the preliminary decision-making actions, and the objectives of greenhouse temperature control, crop growth requirements, operating costs and effects are comprehensively considered to obtain the optimal ventilation decision strategy.

9. The multi-objective solar greenhouse ventilation decision-making method according to claim 1, characterized in that: The optimal ventilation decision strategy is dynamically adjusted to automatically adapt to the meteorological changes inside and outside the greenhouse and the growth stage of the crops, including: By using a multi-agent system and transfer learning technology, the optimal ventilation decision strategy is optimized in real time by utilizing the similarities and differences between the source task and the target task; The target task is determined based on the meteorological changes inside and outside the greenhouse and the growth stage of the crops.

10. A multi-objective solar greenhouse ventilation decision-making device, characterized in that: include: Information collection unit, used to obtain real-time environmental information inside and outside the greenhouse; An information processing unit, used for determining a greenhouse ventilation characteristic vector according to the real-time environmental information; A strategy generating unit, used for generating a preliminary ventilation decision strategy according to the greenhouse ventilation feature vector; A strategy optimization unit, used for performing multi-objective optimization on the preliminary ventilation decision strategy, so that the optimized optimal ventilation decision strategy takes into account the temperature control in the greenhouse, the crop growth requirements, the operation cost and the effect; The strategy matching unit is used to dynamically adjust the optimal ventilation decision strategy to automatically adapt to the meteorological changes inside and outside the greenhouse and the growth stage of the crops.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the multi-objective solar greenhouse ventilation decision-making method as described in any one of claims 1 to 9 is implemented.

12. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the multi-objective solar greenhouse ventilation decision method as described in any one of claims 1 to 9 is implemented.

13. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the multi-objective solar greenhouse ventilation decision method as described in any one of claims 1 to 9 is implemented.

Images