CN114077269B - Greenhouse environment prediction and optimization control method based on data-driven model - Google Patents

Greenhouse environment prediction and optimization control method based on data-driven model Download PDF

Info

Publication number
CN114077269B
CN114077269B CN202010825260.8A CN202010825260A CN114077269B CN 114077269 B CN114077269 B CN 114077269B CN 202010825260 A CN202010825260 A CN 202010825260A CN 114077269 B CN114077269 B CN 114077269B
Authority
CN
China
Prior art keywords
greenhouse
temperature
carbon dioxide
model
data
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202010825260.8A
Other languages
Chinese (zh)
Other versions
CN114077269A (en
Inventor
华净
康孟珍
王浩宇
王秀娟
王飞跃
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Qingdao Parallel Intelligent Industry Management Co ltd
Original Assignee
Qingdao Parallel Intelligent Industry Management Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Qingdao Parallel Intelligent Industry Management Co ltd filed Critical Qingdao Parallel Intelligent Industry Management Co ltd
Priority to CN202010825260.8A priority Critical patent/CN114077269B/en
Publication of CN114077269A publication Critical patent/CN114077269A/en
Application granted granted Critical
Publication of CN114077269B publication Critical patent/CN114077269B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
    • G05D27/00—Simultaneous control of variables covered by two or more of main groups G05D1/00 - G05D25/00
    • G05D27/02—Simultaneous control of variables covered by two or more of main groups G05D1/00 - G05D25/00 characterised by the use of electric means
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/62—Control or safety arrangements characterised by the type of control or by internal processing, e.g. using fuzzy logic, adaptive control or estimation of values
    • F24F11/63—Electronic processing
    • F24F11/64—Electronic processing using pre-stored data
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/70—Control systems characterised by their outputs; Constructional details thereof
    • F24F11/72—Control systems characterised by their outputs; Constructional details thereof for controlling the supply of treated air, e.g. its pressure
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F11/00—Control or safety arrangements
    • F24F11/88—Electrical aspects, e.g. circuits
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/243—Classification techniques relating to the number of classes
    • G06F18/24323—Tree-organised classifiers
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/10—Temperature
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/20—Humidity
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/30—Velocity
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F24—HEATING; RANGES; VENTILATING
    • F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
    • F24F2110/00—Control inputs relating to air properties
    • F24F2110/50—Air quality properties
    • F24F2110/65—Concentration of specific substances or contaminants
    • F24F2110/70—Carbon dioxide
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
    • Y02A40/00—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production
    • Y02A40/10—Adaptation technologies in agriculture, forestry, livestock or agroalimentary production in agriculture
    • Y02A40/25—Greenhouse technology, e.g. cooling systems therefor

Landscapes

  • Engineering & Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Mechanical Engineering (AREA)
  • Physics & Mathematics (AREA)
  • Combustion & Propulsion (AREA)
  • Chemical & Material Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Signal Processing (AREA)
  • Theoretical Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Fuzzy Systems (AREA)
  • Mathematical Physics (AREA)
  • Automation & Control Theory (AREA)
  • Greenhouses (AREA)

Abstract

The invention relates to the technical field of greenhouse control, in particular to a greenhouse environment prediction and optimization control method. The greenhouse environment prediction and optimization control method based on the data driving model comprises the following steps: establishing a greenhouse environment prediction model by utilizing greenhouse environment monitoring data and equipment state data; setting a current desired temperature and a lower carbon dioxide concentration limit based on planting experience or a crop growth model; searching the most suitable opening and closing degree of the ventilation opening through a greenhouse environment prediction model and optimization calculation so as to achieve an optimization target; and adjusting the opening and closing degree of the ventilation opening according to the result of the optimization calculation. The beneficial effects of the invention are: based on the current situation that the existing agricultural Internet of things is generally applied and data is convenient to acquire, a data-driven method is adopted, the temperature and the carbon dioxide concentration are calculated according to the main influence factors and the control equipment state, a numerical model is constructed without the mutual influence on complex environments, and the method is convenient to apply to an actual greenhouse.

Description

Greenhouse environment prediction and optimization control method based on data-driven model
Technical Field
The invention relates to the technical field of greenhouse control, in particular to a greenhouse environment prediction and optimization control method.
Background
The greenhouse internal environment is a model with complex coupling relation, nonlinearity and high uncertainty, various environmental parameters such as temperature, humidity, illumination intensity and carbon dioxide concentration influence each other, and the establishment of a mathematical model of the greenhouse environment has high challenge. For such a complicated environment, a large amount of manpower is required to be consumed by a common control system, and a good control effect cannot be achieved. In addition, there are many uncertainties in environmental control of greenhouses: the transition probability of the environmental state is unknown, the effect of the control action is uncertain, and the greenhouse environment is partially opened to be influenced by the external environment. The classical control method requires that the system has an accurate mathematical model, and the accuracy is not high when the method is applied to greenhouse environment control because the establishment of the greenhouse environment mathematical model is difficult.
Furthermore, typical greenhouse control only considers achieving a given environmental index, and the need for dynamic changes of crops in the greenhouse is not considered. The accumulation of dry matter during crop growth is mainly derived from crop photosynthesis, and key environmental factors influencing photosynthesis are temperature, carbon dioxide concentration and effective photosynthetic radiation. In the greenhouse, the photosynthetic effective radiation is in monotonous rising relation with the outside illumination intensity of the greenhouse and the intensity of the light supplement lamp inside the greenhouse, and is easy to regulate and control to an expected target. The temperature and the carbon dioxide concentration are influenced by multiple factors and are not easy to regulate to the expected target. Meanwhile, the opening and closing degree of the ventilation opening can influence the temperature and the concentration of carbon dioxide in the greenhouse, and the increase of the ventilation opening can generally reduce the temperature and bring about the loss of the carbon dioxide in the greenhouse. The focus of the invention is to ensure the concentration of carbon dioxide required by crop photosynthesis and regulate and control the temperature to ensure the healthy growth of crops.
The modern facility agriculture starts late, and most greenhouses lack scientific and effective environmental control. Through the accurate control to the environment in the greenhouse, can control the growth process of vegetables comparatively accurately to make the greenhouse crop obtain higher economic benefits, reduce the risk. Therefore, the research of the greenhouse control method is a research in conformity with the agricultural development characteristics of China and is very necessary.
Disclosure of Invention
The invention aims to provide a greenhouse prediction and control method based on a parallel control theory. The optimization objective is to minimize the difference between the desired temperature and the actual temperature by adjustment of the tuyeres and to maintain the concentration of carbon dioxide required for photosynthesis in the greenhouse.
In order to achieve the purpose, the invention adopts the technical scheme that: the greenhouse environment prediction and optimization control method based on the data driving model comprises the following steps:
(1) Establishing a greenhouse environment prediction model based on a lightGBM algorithm by utilizing greenhouse environment monitoring data and equipment state data;
(2) Setting a current desired temperature and a lower carbon dioxide concentration limit based on planting experience or a crop growth model;
(3) Searching the most suitable opening and closing degree of the ventilation opening through a greenhouse environment prediction model and optimization calculation so as to achieve an optimization target;
(4) And adjusting the opening and closing degree of the ventilation opening according to the result of the optimization calculation.
As a preferred mode of the invention, the greenhouse environment prediction model comprises a temperature prediction model and CO 2 The concentration prediction model and the algorithm model are as follows:
1) Initializing decision tree parameters, calculating initial output of the decision tree, and recording as
Figure DEST_PATH_IMAGE001
：
Figure DEST_PATH_IMAGE002
（1）
Wherein the content of the first and second substances,
Figure DEST_PATH_IMAGE003
is a constant value that minimizes the initial loss function.
2) Starting iteration, and in each iteration, creating a new sub-tree and calculating the negative gradient of a loss function relative to each sample based on the current decision tree model to serve as an estimated value of residual error; the current number of iterations is recorded as
Figure DEST_PATH_IMAGE004
Created subtree output as
Figure DEST_PATH_IMAGE005
And the estimated value of the residual error is recorded as
Figure DEST_PATH_IMAGE006
：
Figure DEST_PATH_IMAGE007
（2）
3) Will be provided
Figure DEST_PATH_IMAGE008
As target output of the newly created subtree, X continues to be input set of the newly created subtree, and the new subtree is fitted
Figure 785397DEST_PATH_IMAGE005
And updating the decision tree model, wherein the output of the decision tree is calculated as follows:
Figure DEST_PATH_IMAGE009
（3）
4) And repeating the step 2-3 until the loss function meets the requirement or the iteration frequency reaches the set maximum value, ending the iteration, and finally outputting the decision tree as follows:
Figure DEST_PATH_IMAGE010
（4）。
furthermore, the temperature prediction model predicts the temperature inside the greenhouse at the next moment by taking the temperature inside and outside the greenhouse at the current moment, the illumination intensity, the wind speed and the opening and closing degree of the ventilation opening as the characteristics of input data.
Furthermore, the carbon dioxide concentration prediction model predicts the carbon dioxide concentration in the greenhouse at the next moment by taking the temperature inside and outside the greenhouse, the carbon dioxide concentration, the wind speed and the opening and closing degree of the ventilation opening at the current moment as the characteristics of input data.
As a preferred mode of the present invention, the optimization objective is as follows (5):
Figure DEST_PATH_IMAGE011
（5）
wherein
Figure DEST_PATH_IMAGE012
Is time;
Figure DEST_PATH_IMAGE013
is composed of
Figure 76439DEST_PATH_IMAGE012
Calculating an internal temperature value of the greenhouse by using a greenhouse environment prediction model at a moment;
Figure DEST_PATH_IMAGE014
is composed of
Figure DEST_PATH_IMAGE015
The set value of the expected temperature at the moment is a curve changing along with time and illumination;
Figure DEST_PATH_IMAGE016
is composed of
Figure 59438DEST_PATH_IMAGE012
Greenhouse internal CO calculated by time greenhouse environment prediction model 2 A concentration value;
Figure DEST_PATH_IMAGE017
is composed of
Figure 861172DEST_PATH_IMAGE012
CO in time greenhouse 2 Lower limit of concentration value.
Further preferably, the
Figure 941123DEST_PATH_IMAGE017
And calculating according to the temperature value in the greenhouse and the photosynthetically active radiation at each moment through the photosynthesis curve of the plant.
Compared with the prior art, the invention has the beneficial effects that: based on the current situation that the existing agricultural Internet of things is generally applied and data is convenient to acquire, a data-driven method is adopted, the temperature and the carbon dioxide concentration are calculated according to the main influence factors and the control equipment state, a numerical model is constructed without the mutual influence on complex environments, and the method is convenient to apply to an actual greenhouse. By fusing agronomic experience in data-driven control, different environmental factor control targets are interrelated and together surround a uniform target suitable for crop growth, rather than the common mutually independent control. Thereby the regulation and control result is suitable for the growth of crops, and the waste of resources (such as excessive supplement of carbon dioxide) is reduced.
Drawings
FIG. 1 is a schematic flow chart of one cycle of a greenhouse environment prediction and optimization control method based on a data-driven model according to an embodiment of the present invention;
FIG. 2 is a schematic diagram showing the comparison of the temperatures in the greenhouses before and after the optimization in the embodiment of the invention;
FIG. 3 is a schematic diagram showing the comparison of the carbon dioxide concentrations in the greenhouse before and after the optimization in the embodiment of the present invention;
fig. 4 is a schematic diagram for comparing the opening and closing degree of the front and rear leeward openings.
Detailed Description
In order to facilitate an understanding of the invention, reference will now be made in detail to the present embodiments of the invention, examples of which are illustrated in the accompanying drawings. Preferred embodiments of the present invention are shown in the drawings. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
The accumulation of dry matter during the growth of crops mainly comes from the photosynthesis of crops, and key environmental factors influencing the photosynthesis are temperature, carbon dioxide concentration and effective photosynthetic radiation. In the greenhouse, the photosynthetic effective radiation is in monotonous rising relation with the outside illumination intensity of the greenhouse and the intensity of the light supplement lamp inside the greenhouse, and is easy to regulate and control to an expected target. The temperature and the carbon dioxide concentration are influenced by multiple factors and are not easy to regulate to the expected target. Meanwhile, the opening and closing degree of the ventilation opening can influence the temperature and the concentration of carbon dioxide in the greenhouse, and the increase of the ventilation opening can reduce the temperature and bring the loss of the carbon dioxide in the greenhouse.
The invention realizes temperature regulation and control and carries out carbon dioxide concentration cooperative control at the same time by regulating the opening and closing degree of the ventilation opening through greenhouse environment prediction based on a data driving model, and the flow is shown as figure 1 and specifically comprises the following steps:
1. greenhouse environment prediction model established based on lightGBM algorithm
Establishing a greenhouse environment prediction model based on experimental data obtained by a lightGBM algorithm and a second international autonomous greenhouse challenge match: a temperature prediction model and a carbon dioxide prediction model.
The decision tree can be regarded as an addition model, and the final output of the decision tree is obtained by adding the outputs of a plurality of subtrees. Assuming that the number of training set samples is N, the input to the decision tree is the set𝑋（𝑋 1 ,𝑋 2 ,… ,𝑋 𝑁 ) The system consists of related environment variables in the greenhouse at the current moment and state variables of control equipment; the output is a set
Figure DEST_PATH_IMAGE018
（
Figure 294744DEST_PATH_IMAGE018
1 ,
Figure DEST_PATH_IMAGE019
2 ,… ,
Figure 233882DEST_PATH_IMAGE019
𝑁 ) The temperature in the greenhouse (carbon dioxide concentration) at the next moment predicted for the model; the target output is a set𝑌（𝑌 1 ,𝑌 2 ,… ,𝑌 𝑁 ) The actual temperature (carbon dioxide concentration) in the greenhouse at the next moment. The loss function is noted as𝐿(𝑌̂ 𝑖 ,𝑌 𝑖 ). The maximum number of iterations is denoted as M. The basic algorithm flow of lightGBM is as follows:
1) Initializing decision tree parameters, calculating initial output of decision tree, and recording as
Figure 698361DEST_PATH_IMAGE001
：
Figure 272299DEST_PATH_IMAGE002
（1）
Wherein, the first and the second end of the pipe are connected with each other,
Figure 531242DEST_PATH_IMAGE003
is a constant value that minimizes the initial loss function.
2) Starting iteration, and in each iteration, creating a new sub-tree and calculating the negative gradient of a loss function relative to each sample based on the current decision tree model to serve as an estimated value of residual error; the current number of iterations is recorded as
Figure 551151DEST_PATH_IMAGE004
Created subtree output as
Figure 288163DEST_PATH_IMAGE005
The estimated value of the residual is recorded as
Figure 218073DEST_PATH_IMAGE006
：
Figure 913496DEST_PATH_IMAGE007
（2）
3) Will be provided with
Figure 561647DEST_PATH_IMAGE008
As the target output of the newly created subtree, X continues to be the input set of the newly created subtree, and the new subtree is fitted
Figure 102349DEST_PATH_IMAGE005
And updating the decision tree model, wherein the output of the decision tree is calculated as follows:
Figure 11399DEST_PATH_IMAGE009
（3）
4) And repeating the steps 2) -3) until the loss function meets the requirement or the iteration number reaches the set maximum value, and ending the iteration. After the iteration is completed, the final output of the decision tree is:
Figure 612145DEST_PATH_IMAGE010
（4）。
the temperature prediction model uses the inside and outside temperature, illumination intensity, wind speed, opening and closing degree of a ventilation opening and the like of the greenhouse at the current hour 11
The term factors are characteristic of the input data and predict the temperature inside the greenhouse for the next hour. The specific input features used are shown in table 1.
TABLE 1 temperature prediction model input features
Description of the characteristics Unit of
Internal temperature ℃
Intensity of external solar radiation W·m -2
Outside temperature ℃
Wind speed m·s -1
Photosynthetically active radiation for crop canopy μmol·m -2 ·s -1
Average temperature of heating pipe ℃
Supply side heating line temperature ℃
Degree of opening and closing of leeward opening %
Degree of opening and closing of wind inlet %
Pipeline regulating and controlling temperature set value ℃
Temperature set value regulated and controlled by air vent ℃
Air flow and temperature conduction inside and outside the greenhouse are affected by a number of factors. The temperature inside and outside the greenhouse, the opening and closing degree of the back (opposite) air inlet and the wind speed influence the temperature inside the greenhouse by influencing the conduction of the temperature inside and outside the greenhouse. External solar radiation intensity and crop canopy efficient photosynthesis affect greenhouse internal lighting, and thus greenhouse internal temperature. The heating pipeline is used for heating the greenhouse in the weather with lower temperature and maintaining the temperature in the greenhouse not to be lower than a corresponding set value; the ventilation opening is opened when the temperature in the greenhouse is higher than the corresponding set value, so that the effect of reducing the temperature in the greenhouse is achieved.
The carbon dioxide concentration prediction model predicts the carbon dioxide concentration in the greenhouse in the next hour by taking 10 factors such as the temperature inside and outside the greenhouse in the current hour, the carbon dioxide concentration, the wind speed, the opening and closing degree of a ventilation opening and the like as the characteristics of input data. The specific input features used are shown in table 2.
TABLE 2 carbon dioxide concentration prediction model input features
Description of the features Unit of
Internal carbon dioxide concentration ppm
Temperature inside the greenhouse ℃
Intensity of external solar radiation W·m -2
Outside temperature of greenhouse ℃
Wind speed m·s -1
Photosynthetically active radiation for crop canopy μmol·m -2 ·s -1
Degree of opening and closing of leeward opening %
Degree of opening and closing of wind inlet %
Rate of carbon dioxide supply kg·m -2 ·s -1
Leaf Area Index (LAI) —
The temperature inside and outside the greenhouse, the opening and closing degree of the back (opposite) air inlet and the air speed influence the flow of the air inside and outside the greenhouse so as to influence the concentration of the carbon dioxide inside the greenhouse. The intensity of external solar radiation and effective photosynthesis in the crop canopy affect the rate of crop photosynthesis and the rate of carbon dioxide consumption. The carbon dioxide generator maintains the carbon dioxide concentration by adjusting the feed rate. The leaf area index reflects the growth of the crop, which affects the rate of carbon dioxide consumption.
In this embodiment, a data in a Json file provided by the second international autonomous greenhouse challenge is taken as an example to train and verify a greenhouse environment model. In the virtual experiment, the experiment period is from 12 and 15 days in 2017 to 6 and 1 days in 2018, and the related experiment data are recorded at intervals of hours, so that 4032 pieces of experiment data are recorded in total. It is divided into training and testing sets in chronological order. Wherein, the training set is 3032 data from 12 and 15 days in 2017 to 4 and 20 days in 2018; the test set was 1000 data from 20 months 4 to 1 month 6 in 2018.
In this embodiment, the maximum number of iterations of the decision tree training is set to 10000, and the loss function is an average absolute error function.
The average relative errors of the temperature prediction model and the carbon dioxide concentration prediction model constructed in the embodiment are respectively 2.41% and 1.67%, and the accuracy is high.
2. Optimizing vent control strategies based on the data-driven model established above
1. Setting a current desired temperature based on planting experience or crop growth models
The current suitable temperature is obtained by means of agricultural expert experience, crop photosynthetic models and the like, and is set as the expected temperature.
2. Environmental monitoring data is acquired by using environmental monitoring equipment such as sensors and the like: the temperature inside and outside the greenhouse, the illumination, the carbon dioxide concentration, the wind speed, the leaf area index, the opening degree of the ventilation opening and the like.
3. Predicting the temperature and the carbon dioxide concentration by using the constructed temperature prediction model and the carbon dioxide concentration prediction model, and performing optimization calculation on the predicted temperature and the predicted carbon dioxide concentration in the greenhouse to ensure that CO is generated 2 In the case where the concentration is not lower than the lower limit, the temperature is as close as possible to the desired temperature.
Temperature and carbon dioxide concentration are important factors affecting crop photosynthesis capacity within a greenhouse, and therefore, the vent optimization control strategy being studied is to minimize the difference between the temperature within the greenhouse and the desired temperature while maintaining the necessary carbon dioxide level.
The optimization object is the opening and closing degree of the air vents, and the optimization target is as follows:
Figure 482012DEST_PATH_IMAGE011
（5）
wherein
Figure 91985DEST_PATH_IMAGE012
Is time;
Figure 855542DEST_PATH_IMAGE013
is composed of
Figure 266669DEST_PATH_IMAGE012
A predicted value of the internal temperature of the greenhouse is obtained by the greenhouse environment model at the moment;
Figure 748466DEST_PATH_IMAGE014
is composed of
Figure 630971DEST_PATH_IMAGE015
The set value of the expected temperature at the moment is a curve changing along with time and illumination, and is given by referring to the technical document of the challenge race;
Figure 389980DEST_PATH_IMAGE016
is composed of
Figure 598107DEST_PATH_IMAGE012
Calculating a predicted value of the concentration of carbon dioxide in the greenhouse by using a greenhouse environment model at a moment;
Figure 301621DEST_PATH_IMAGE017
is composed of
Figure 394342DEST_PATH_IMAGE012
Lower limit of carbon dioxide concentration in the greenhouse at that time. In this embodiment, based on the photosynthesis curve of the tomatoes in the TOMSIM model, the lower limit value of the carbon dioxide concentration is calculated according to the temperature value and the photosynthetically active radiation in each hour greenhouse
Figure 132491DEST_PATH_IMAGE017
，The control objective was to keep the total daily photosynthetic rate of the tomatoes at 80% of their saturation value.
4. The optimum greenhouse temperature value and CO calculated according to the optimization 2 Concentration, and the opening and closing degree of the ventilation opening are adjusted, so that the aim of optimizing the environment in the greenhouse is fulfilled.
Fig. 2 shows a comparison of the greenhouse internal temperature of data from 26 days 5 to 1 day 6 in 2018 with the original and optimized vent control strategy. As shown, the desired temperature set point is not constant, and varies with changes in solar radiation and external temperature.
The optimal control strategy based on the data driving model obtains a good control effect. The average difference between the internal temperature of the greenhouse before optimization and the desired temperature was 2.24 ℃. The average difference between the temperature inside the greenhouse after optimization and the expected temperature is 1.24 ℃, and is only 55.35% before optimization. At the end of 5 months, the air vents cannot completely and effectively reduce the indoor temperature due to strong solar radiation at noon, and a large difference exists between the expected temperature and the actual temperature in the greenhouse. The maximum temperature in the greenhouse can reach 34.21 ℃, and the difference value with the expected temperature is more than 10 ℃ at most.
Fig. 3 shows the difference in carbon dioxide concentration in the greenhouse before and after the optimization control from 31/5/2018 to 1/6/1. As can be seen, the optimized carbon dioxide concentration is significantly higher than its lower limit. This indicates that the optimized vent control strategy satisfies the carbon dioxide concentration constraint. Meanwhile, a certain optimization space is reserved between the concentration of the carbon dioxide and the lower limit value of the concentration of the carbon dioxide. Therefore, it is attempted to set the lower limit value thereof to a different value to observe whether the optimization control experiment is still effective. Experimental results show that the optimal control strategy has failed to meet the constraints of the computational experiments when the lower carbon dioxide limit is set to maintain the total daily photosynthetic rate of the crop above 88% of its saturation value. At this point, the required carbon dioxide concentration in the greenhouse is already significantly higher than the actual carbon dioxide concentration.
The simultaneous leeward opening degree is shown in fig. 4, and the opening degree to the tuyere is linearly related to the leeward opening degree. In summer, the opening degree of the back air inlet under the optimized control strategy is obviously smaller than that of the original control strategy.
The greenhouse environment model established based on the lightGBM algorithm becomes more accurate along with the increase of the training data volume. The model basically achieves higher prediction accuracy after a training process of 4 weeks, and the data used for training the model in the embodiment of the invention is recorded in units of hours. In practical greenhouse applications, the period of data acquisition often only requires 15 minutes or even less. Therefore, the time to build an environmental model based on this algorithm is expected to be reduced to a week or even less. In the greenhouse data acquisition process, normal production management and control can be performed according to the prior art and knowledge. And the model is updated through the collected new data, thereby realizing parallel control and further improving the performance of the model. This means that the process of the invention does not require high experimental costs and long experimental periods.
The method of the invention carries out calculation experiments in a data-driven greenhouse environment model, effectively optimizes the greenhouse ventilation opening control strategy under the constraint condition of carbon dioxide concentration, and achieves the purpose of optimizing the environment control in the greenhouse.

Claims (3)

1. The greenhouse environment prediction and optimization control method based on the data-driven model is characterized by comprising the following steps:
(1) Establishing a greenhouse environment prediction model based on a lightGBM algorithm by utilizing greenhouse environment monitoring data and equipment state data;
the greenhouse environment prediction model comprises a temperature prediction model and a carbon dioxide concentration prediction model;
the temperature prediction model predicts the temperature inside the greenhouse at the next moment by taking the temperature inside and outside the greenhouse at the current moment, the illumination intensity, the wind speed, the opening and closing degree of the ventilation opening, the photosynthetic effective radiation of the crop canopy, the average temperature of the heating pipeline and the temperature of the heating pipeline at the supply side as the characteristics of input data;
the carbon dioxide concentration prediction model predicts the carbon dioxide concentration in the greenhouse at the next moment by taking the temperature inside and outside the greenhouse at the current moment, the carbon dioxide concentration, the wind speed, the opening and closing degree of a ventilation opening, the photosynthetic effective radiation of the crop canopy and the carbon dioxide supply rate as the characteristics of input data;
(2) Setting a current desired temperature and a lower carbon dioxide concentration limit based on planting experience or a crop growth model;
(3) Searching the most suitable opening and closing degree of the ventilation opening through a greenhouse environment prediction model and optimization calculation so as to achieve an optimization target;
Figure 903469DEST_PATH_IMAGE001
（5）
wherein the content of the first and second substances,
Figure 820610DEST_PATH_IMAGE002
is time;
Figure 210134DEST_PATH_IMAGE003
is composed of
Figure 605343DEST_PATH_IMAGE004
Calculating an internal temperature value of the greenhouse by using a greenhouse environment prediction model at a moment;
Figure 295082DEST_PATH_IMAGE005
is composed of
Figure 117544DEST_PATH_IMAGE002
The set value of the expected temperature at the moment is a curve changing along with time and illumination;
Figure 525523DEST_PATH_IMAGE006
is composed of
Figure 724423DEST_PATH_IMAGE002
Greenhouse internal CO calculated by time greenhouse environment prediction model 2 A concentration value;
Figure 3089DEST_PATH_IMAGE007
is composed of
Figure 996452DEST_PATH_IMAGE004
CO in greenhouse at any moment 2 A lower limit of the concentration value;
(4) And adjusting the opening and closing degree of the ventilation opening according to the result of the optimization calculation.
2. The greenhouse environment prediction and optimization control method based on the data-driven model as claimed in claim 1, wherein the algorithm models of the temperature prediction model and the carbon dioxide concentration prediction model are:
(1) Initializing decision tree parameters, calculating initial output of the decision tree, and recording as
Figure 891727DEST_PATH_IMAGE008
：
Figure 628739DEST_PATH_IMAGE009
（1）
Wherein, the first and the second end of the pipe are connected with each other,
Figure 27491DEST_PATH_IMAGE010
is a constant value that minimizes the initial loss function;
(2) Starting iteration, and in each iteration, creating a new sub-tree and calculating the negative gradient of a loss function relative to each sample based on the current decision tree model to serve as an estimated value of residual error; the current number of iterations is recorded as
Figure 457335DEST_PATH_IMAGE011
Created subtree output as
Figure 698960DEST_PATH_IMAGE012
And the estimated value of the residual is denoted as r mi ,i＝1,2,...,N：
Figure 227342DEST_PATH_IMAGE014
（2）
(3) Will be provided with
Figure 703454DEST_PATH_IMAGE015
As target output of the newly created subtree, X continues to be input set of the newly created subtree, and the new subtree is fitted
Figure 432375DEST_PATH_IMAGE012
And updating the decision tree model, wherein the output of the decision tree is calculated as follows:
Figure 120977DEST_PATH_IMAGE016
（3）
(4) And (5) repeating the steps (2) to (3) until the loss function meets the requirement or the iteration frequency reaches a set maximum value, ending the iteration, and finally outputting a decision tree as follows:
Figure 618954DEST_PATH_IMAGE017
（4）。
3. the data-driven model-based greenhouse environment prediction and optimization control method as claimed in claim 1, wherein the greenhouse environment prediction and optimization control method is characterized in that
Figure 531546DEST_PATH_IMAGE007
And calculating according to the photosynthesis curve of the crop, the temperature value in the greenhouse at each moment and the photosynthetically active radiation.
CN202010825260.8A 2020-08-17 2020-08-17 Greenhouse environment prediction and optimization control method based on data-driven model Active CN114077269B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010825260.8A CN114077269B (en) 2020-08-17 2020-08-17 Greenhouse environment prediction and optimization control method based on data-driven model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010825260.8A CN114077269B (en) 2020-08-17 2020-08-17 Greenhouse environment prediction and optimization control method based on data-driven model

Publications (2)

Publication Number Publication Date
CN114077269A CN114077269A (en) 2022-02-22
CN114077269B true CN114077269B (en) 2022-10-28

Family

ID=80281149

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010825260.8A Active CN114077269B (en) 2020-08-17 2020-08-17 Greenhouse environment prediction and optimization control method based on data-driven model

Country Status (1)

Country Link
CN (1) CN114077269B (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN115226516B (en) * 2022-07-07 2023-09-26 西北农林科技大学 Cooperative regulation and control method for facility light and carbon dioxide environment
CN115529987B (en) * 2022-11-30 2023-03-24 中化现代农业有限公司 Air port regulating and controlling method, device, equipment and storage medium for crop facility
CN117389355B (en) * 2023-12-07 2024-03-12 凯盛浩丰农业集团有限公司 Intelligent greenhouse temperature control method and system for tomato planting
CN117519354B (en) * 2024-01-08 2024-03-29 山西省农业机械发展中心 Intelligent information remote monitoring system for facility agriculture

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105446142A (en) * 2015-12-25 2016-03-30 中国农业大学 Greenhouse CO2 gas fertilizer increasing method, device and system
CN106875067A (en) * 2017-03-01 2017-06-20 淮阴师范学院 The method of greenhouse carbon dioxide concentration prediction model when building winter fine day
CN108133289A (en) * 2017-12-21 2018-06-08 中国铁建电气化局集团有限公司 Tunnel ventilation control method and system based on environmental forecasting
JP2018099067A (en) * 2016-12-20 2018-06-28 ネポン株式会社 Growth management device, growth management method, and program
CN111368975A (en) * 2020-02-25 2020-07-03 东北农业大学 Method for predicting dynamic change of multi-environment factors in pigsty

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105446142A (en) * 2015-12-25 2016-03-30 中国农业大学 Greenhouse CO2 gas fertilizer increasing method, device and system
JP2018099067A (en) * 2016-12-20 2018-06-28 ネポン株式会社 Growth management device, growth management method, and program
CN106875067A (en) * 2017-03-01 2017-06-20 淮阴师范学院 The method of greenhouse carbon dioxide concentration prediction model when building winter fine day
CN108133289A (en) * 2017-12-21 2018-06-08 中国铁建电气化局集团有限公司 Tunnel ventilation control method and system based on environmental forecasting
CN111368975A (en) * 2020-02-25 2020-07-03 东北农业大学 Method for predicting dynamic change of multi-environment factors in pigsty

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于梯度提升决策树的电力系统短期负荷预测研究;毕云帆;《中国优秀硕士学位论文全文数据库（电子期刊）》;20200331;C042-842 *

Also Published As

Publication number Publication date
CN114077269A (en) 2022-02-22

Similar Documents

Publication Publication Date Title
CN114077269B (en) Greenhouse environment prediction and optimization control method based on data-driven model
Cai et al. A method for modelling greenhouse temperature using gradient boost decision tree
Rodríguez et al. Adaptive hierarchical control of greenhouse crop production
del Sagrado et al. Bayesian networks for greenhouse temperature control
KR20180076766A (en) Artificial Intelligence Based Smart Farm Management System
CN106292802A (en) A kind of Intelligent Prediction Control System for fish and vegetable symbiotic system and method
CN112527037B (en) Greenhouse environment regulation and control method and system with environment factor prediction function
Su et al. Multi-layer hierarchical optimisation of greenhouse climate setpoints for energy conservation and improvement of crop yield
CN115529987B (en) Air port regulating and controlling method, device, equipment and storage medium for crop facility
Moon et al. Estimation of greenhouse CO 2 concentration via an artificial neural network that uses environmental factors
CN114638146A (en) Crop irrigation water demand prediction method based on AquaCrop model and SVR
CN114371748B (en) Greenhouse crop illumination carbon dioxide regulation and control method based on Gaussian curvature
Lešić et al. Rapid plant development modelling system for predictive agriculture based on artificial intelligence
Li et al. Determining optimal CO2 concentration of greenhouse tomato based on PSO-SVM
Bailey et al. Improving the cost effectiveness of greenhouse climate control
CN116449897B (en) Greenhouse environment optimal control method, server and system
Su et al. Greenhouse climate setpoint optimization: An online decision strategy
Ioslovich et al. Approximate seasonal optimization of the greenhouse environment for a multi-state-variable tomato model
CN107918424A (en) A kind of method and system for controlling plant growth environment
Chen et al. Multi-objective optimal regulation model and system based on whole plant photosynthesis and light use efficiency of lettuce
CN115328242B (en) Culture environment intelligent regulation system based on remote control
Boersma et al. Robust sample-based model predictive control of a greenhouse system with parametric uncertainty
CN110073857A (en) A kind of greenhouse facade ventilating and thermal insulating global anti-wind system and control method
CN112034916B (en) Greenhouse Control System
Dong et al. Crop model-based greenhouse optimal control system: survey and perspectives

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant