CN112947648A - Agricultural greenhouse environment prediction method and system - Google Patents

Agricultural greenhouse environment prediction method and system Download PDF

Info

Publication number
CN112947648A
CN112947648A CN202110239759.5A CN202110239759A CN112947648A CN 112947648 A CN112947648 A CN 112947648A CN 202110239759 A CN202110239759 A CN 202110239759A CN 112947648 A CN112947648 A CN 112947648A
Authority
CN
China
Prior art keywords
model
air
model parameter
greenhouse environment
greenhouse
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.)
Granted
Application number
CN202110239759.5A
Other languages
Chinese (zh)
Other versions
CN112947648B (en
Inventor
苏远平
徐立鸿
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.)
Jiangxi University of Science and Technology
Original Assignee
Jiangxi University of Science and Technology
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 Jiangxi University of Science and Technology filed Critical Jiangxi University of Science and Technology
Priority to CN202110239759.5A priority Critical patent/CN112947648B/en
Publication of CN112947648A publication Critical patent/CN112947648A/en
Application granted granted Critical
Publication of CN112947648B publication Critical patent/CN112947648B/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

Abstract

The invention discloses an agricultural greenhouse environment prediction method and system, firstly, interpolation is carried out according to the measurement data to calculate a reference input track, and an initial greenhouse environment model is updated; secondly, constructing a model parameter identification controller based on a Lyapunov function; calculating optimal solutions of the nonlinear time-varying model parameters at the t moment according to the model parameter identification controller and the initial greenhouse environment model, and determining an identification general model parameter solution according to the optimal solutions of the nonlinear time-varying model parameters in the model parameter set when t is greater than or equal to the control process terminal time; and finally substituting the parameters of the identification general model into the greenhouse environment model to obtain a greenhouse environment predicted value. The method determines the solution of the identification general model parameters based on the Lyapunov function stability, substitutes the solution of the identification general model parameters into the greenhouse environment model to obtain the predicted value of the greenhouse environment, comprehensively considers the nonlinear time-varying model parameters at a plurality of different moments, and further improves the accuracy of prediction.

Description

Agricultural greenhouse environment prediction method and system
Technical Field
The invention relates to the field of environment prediction, in particular to an agricultural greenhouse environment prediction method and system.
Background
Modern greenhouse crop production is an important mode of production for agricultural production. The greenhouse crop production has the advantages of high efficiency and high quality, and can carry out-of-season production to meet the requirements of people on different types of agricultural products throughout the year. However, for modern high-grade agricultural greenhouses, the precondition for realizing high yield, high quality and high-efficiency production is to realize accurate regulation and control of greenhouse environment and water and fertilizer irrigation. Meanwhile, in order to reduce the energy consumption cost in the production process of greenhouse crops and submit the final yield of crops as much as possible, necessary planning and optimization of long-term process regulation and management are generally required before the regulation of greenhouse environment and water and fertilizer irrigation to determine the set values of greenhouse environment regulation in different periods. Therefore, the greenhouse environment and the crop growth in the greenhouse crop production process for a long time must be accurately predicted, and an accurate greenhouse environment dynamic model is an important guarantee for realizing the accurate prediction. On the other hand, from the viewpoint of greenhouse environment control, model-based greenhouse environment control enables energy-efficient optimal control, which also generally requires an accurate greenhouse environment model. Therefore, the construction of an accurate greenhouse environment model has very important practical significance for the production of modern agricultural greenhouse crops.
Changes in greenhouse environment are typically affected by outdoor weather, crop growth, greenhouse structure and materials, and various greenhouse environment conditioning equipment. Therefore, various heat and material exchange processes influencing the change of the greenhouse environment are very complicated, and the existing thermodynamic theory is difficult to accurately simulate the complicated heat and material exchange processes from the mechanism. On the other hand, the heat exchange and substance exchange processes contained in the greenhouses with different structures and materials are greatly different, which causes that the mechanisms of the environment change of the greenhouses with different structures and materials are also different, so that the corresponding greenhouse environment model structures and parameters are also greatly different. This means that it is difficult to find a general greenhouse environment mechanism model to simulate the greenhouse environment change of different structures and materials, and if the greenhouse environment change mechanism model of each structure and material is constructed, it needs to invest a lot of manpower, material resources and time to deeply study the various heat exchange and material exchange processes in the greenhouse, which is not practical for the greenhouse crop production practice. Therefore, the development of a rapid and universal modeling method for greenhouse environment data to meet the requirements of greenhouse environment control and intelligent planning and decision-making in greenhouse crop production practice is urgent.
Case 1 now: the most typical greenhouse environment structured data modeling method is nonlinear modeling based on a neural network, and the method is mainly characterized in that the whole greenhouse environment model is represented by a plurality of nervesA network structure of elements. The performance of the model mainly depends on the connection weights W and V of the network, the number N of neurons and the output threshold theta of the neurons, as shown in FIG. 1, (a) is the network structure of the model, and (b) is the neuron structure, so that the greenhouse environment model based on the neural network can be expressed as
Figure BDA0002961672090000021
For the training set (X, Y), the goal of modeling the greenhouse environment neural network is to continuously adjust the connection weights and neuron thresholds of the network to make the output of the network
Figure BDA0002961672090000023
As close as possible to the actual output Y. The network learning method mainly comprises an off-line learning mode and an on-line rolling learning mode. The off-line learning method is characterized in that the network connection right is adjusted by using experimental data of the greenhouse environment in an off-line state through a gradient descent method, so that the approximation error is made
Figure BDA0002961672090000022
And minimum. If the training data can cover most of the greenhouse environment change dynamics in the greenhouse crop production process, the obtained greenhouse environment neural network model can accurately simulate real greenhouse environment changes generally, but the greenhouse environment is a very complex strong coupling nonlinear system, and because the weather changes have great uncertainty, the experimental data can hardly cover the main dynamics of the greenhouse environment changes generally, and the reliability of the model can hardly be ensured in the dynamics that the training data do not cover by the greenhouse environment model obtained through off-line training. To solve this problem, another neural network modeling approach is to perform rolling training on neural network models of greenhouse environment during the production of greenhouse crops using rolling data sets. Regardless of off-line or on-line modeling, the neural network modeling method of the greenhouse environment is essentially a local modeling method and can only simulate the local dynamics of the greenhouse environment change. In addition, neural network models of greenhouse environments are often able to achieve better simulation on training data setsHowever, because the change of the greenhouse environment is greatly influenced by various uncertain factors such as external weather, crop growth, climate adjustment operation and the like, most of the dynamics of the change of the greenhouse environment are usually difficult to cover by an experimental data set obtained within limited time and cost, so that the generalization capability of a greenhouse environment neural network model is usually not ideal, a relatively ideal effect is difficult to obtain on the dynamic simulation of the greenhouse environment which is not covered by the data set, and even the constructed neural network model cannot be converged on the unknown dynamics. Therefore, the neural network model of the greenhouse environment can only be suitable for simulating the change of the greenhouse environment under specific conditions within a limited time, and the whole nonlinear dynamic state of the change of the greenhouse environment is difficult to simulate, so that the neural network model is a local model. The main reason why the generalization capability of the neural network model in the greenhouse environment is not ideal is that the neural network model is a black box fitting data model and does not have a basic mechanism structure of greenhouse environment change, so that the simulation performance of the neural network model depends heavily on the training data.
Existing case 2: the typical method of modeling greenhouse environment is based on an optimized grey-box modeling method, and the main idea is that certain basic mechanisms of greenhouse environment heat exchange and material exchange are reserved on a model structure, and the model structure is expressed in a typical state differential equation form, as shown in the following,
Figure BDA0002961672090000031
the various components of the model therefore have a well-defined physical meaning. In this mode, the main task of modeling the greenhouse environment is to identify the parameter θ in the model. For a training data sample set (X, Y), the model parameters θ are identified primarily by minimizing model errors
Figure BDA0002961672090000032
To be implemented. This problem can be expressed as follows:
Figure BDA0002961672090000033
wherein theta is*For obtaining the best model parameters, L is the number of training samples, and the common optimization method of parameter identification is mainlyThere are newton gradient descent method and genetic algorithm.
The optimization-based gray box modeling method can obtain more stable optimal model parameters theta for greenhouse environment change under specific conditions*Therefore, the constructed greenhouse environment model can obtain better simulation performance. However, such a method of obtaining optimal model parameters by minimizing model errors generally obtains optimal model parameters θ*Is a set of constants that are independent of changes in the greenhouse environment. However, in practice, the heat transfer coefficients and the exchange coefficients of various heat exchange and material exchange processes involved in greenhouse environment changes are also nonlinear time-varying, and since many mechanisms of greenhouse environment changes are still unknown, unknown dynamics of the greenhouse environment changes are reflected on the heat transfer coefficients and the material exchange coefficients, so that various model parameters in the greenhouse environment model are necessarily nonlinear time-varying parameters with great uncertainty. If the model parameters subjected to the normalization are difficult to ensure that the constructed model obtains better simulation performance outside the training data set, even the convergence of the model can not be ensured, so that the reliability and universality of the model are seriously influenced. Like the black box datamation modeling, the gray box modeling method based on optimization still cannot ensure good generalization capability of the model.
Besides the two modes, the data modeling method of the greenhouse environment further comprises a fuzzy system, a support vector machine and the like. These modeling methods are essentially gray box modeling methods that use a large amount of experimental data to train weight parameters of neural networks and the like to fit changes in the greenhouse environment. However, the model is only an input-output model, the structure and parameters of the model do not have actual physical significance, and the essential processes of various heat exchanges and substance exchanges in the greenhouse cannot be reflected, so that the model does not have wide generalization capability, and is difficult to overcome the disturbance of external weather change and crop growth on the change of the indoor environment. In fact, it can be found from decades of modeling and experience of greenhouse environment mechanism that the greenhouses with different structures and materials have similar heat exchange and material exchange processes inside, such as heat exchange between indoor air and crop canopy, heat exchange between indoor air and ground, etc., and the difference is that the heat transfer coefficients of these heat exchange processes are affected by the greenhouse structures and materials to be changed differently. The difficulty with greenhouse modeling is therefore that various heat transfer coefficients are difficult to determine in addition to complex heat exchange mechanisms.
Disclosure of Invention
The invention aims to provide an agricultural greenhouse environment prediction method and system to improve the accuracy of prediction.
In order to achieve the above object, the present invention provides a method for predicting an agricultural greenhouse environment, the method comprising:
step S1: giving a control process terminal time;
step S2: acquiring measurement data; the measurement data comprises greenhouse environment state variables, control signals and outdoor weather;
step S3: carrying out interpolation calculation according to the measurement data to obtain a reference input track;
step S4: updating an initial greenhouse environment model based on the reference input trajectory;
step S5: constructing a model parameter identification controller based on a Lyapunov function;
step S6: calculating optimal solutions of nonlinear time-varying model parameters at the t moment according to the model parameter identification controller and the initial greenhouse environment model, and storing the optimal solutions to a model parameter set;
step S7: judging whether t is less than the terminal time of the control process; if t is less than the control process terminal time, making t equal to t +1, and returning to step S2; if t is greater than or equal to the control process terminal time, outputting the model parameter set;
step S8: determining an identification general model parameter solution according to the optimal solution of the nonlinear time-varying model parameters at different moments in the model parameter set;
step S9: substituting the identification general model parameter solution into a greenhouse environment model to obtain a greenhouse environment predicted value; the greenhouse environment prediction value comprises the following steps: a predicted value of indoor temperature, a predicted value of indoor humidity, and a predicted value of indoor CO2 concentration.
The invention also provides an agricultural greenhouse environment prediction system, which comprises:
the given module is used for giving the terminal time of the control process;
the acquisition module is used for acquiring measurement data; the measurement data comprises greenhouse environment state variables, control signals and outdoor weather;
the reference input track determining module is used for carrying out interpolation calculation according to the measurement data to obtain a reference input track;
an updating module for updating an initial greenhouse environment model based on the reference input trajectory;
the construction module is used for constructing a model parameter identification controller based on a Lyapunov function;
the calculation module is used for calculating the optimal solution of each nonlinear time-varying model parameter at the t moment according to the model parameter identification controller and the initial greenhouse environment model and storing the optimal solution into a model parameter set;
the judging module is used for judging whether t is less than the terminal time of the control process; if t is less than the terminal time of the control process, making t equal to t +1, and returning to the acquisition module; if t is greater than or equal to the control process terminal time, outputting the model parameter set;
the identification general model parameter solution determining module is used for determining an identification general model parameter solution according to the optimal solution of the nonlinear time-varying model parameters at different moments in the model parameter set;
the prediction module is used for substituting the identification general model parameter solution into a greenhouse environment model to obtain a greenhouse environment prediction value; the greenhouse environment prediction value comprises the following steps: a predicted value of indoor temperature, a predicted value of indoor humidity, and a predicted value of indoor CO2 concentration.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention discloses an agricultural greenhouse environment prediction method and system, firstly, interpolation is carried out according to the measurement data to calculate a reference input track, and an initial greenhouse environment model is updated; secondly, constructing a model parameter identification controller based on a Lyapunov function; calculating optimal solutions of the nonlinear time-varying model parameters at the t moment according to the model parameter identification controller and the initial greenhouse environment model, and determining an identification general model parameter solution according to the optimal solutions of the nonlinear time-varying model parameters in the model parameter set when t is greater than or equal to the control process terminal time; and finally substituting the parameters of the identification general model into the greenhouse environment model to obtain a greenhouse environment predicted value. The method determines the solution of the identification general model parameters based on the Lyapunov function stability, substitutes the solution of the identification general model parameters into the greenhouse environment model to obtain the predicted value of the greenhouse environment, and comprehensively considers a plurality of nonlinear time-varying model parameters, thereby improving the accuracy of prediction.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without inventive exercise.
FIG. 1 is a diagram of a prior art neural network-based greenhouse environment model;
FIG. 2 is a flow chart of a method for predicting agricultural greenhouse environment according to an embodiment of the present invention;
FIG. 3 is a flowchart of an iterative calculation of a crop canopy temperature cycle according to an embodiment of the present invention;
FIG. 4 is a schematic diagram of control input measurement data for greenhouse validation according to an embodiment of the present invention;
FIG. 5 is a diagram illustrating the tracking results of the greenhouse environment of 2016, 9, 10-12 months;
FIG. 6 is a diagram illustrating the tracking results of the greenhouse environment of 12 months and 10-12 days in 2016 according to one embodiment of the present invention;
FIG. 7 is a graph of the nonlinear variation of the model parameter time response identified in the embodiment of the present invention;
FIG. 8 is a diagram illustrating a correlation analysis of model parameters according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention aims to provide an agricultural greenhouse environment prediction method and system to improve the accuracy of prediction.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
As shown in FIG. 1, the invention discloses an agricultural greenhouse environment prediction method, which comprises the following steps:
step S1: given control process terminal time tf。
Step S2: acquiring measurement data; the measurement data includes greenhouse environment state variables, control signals, and outdoor weather.
Step S3: and carrying out interpolation calculation according to the measurement data to obtain a reference input track.
Step S4: updating an initial greenhouse environment model based on the reference input trajectory.
Step S5: and constructing a model parameter identification controller based on the Lyapunov function.
Step S6: calculating optimal solution theta of each nonlinear time-varying model parameter at the t moment according to the model parameter identification controller and the initial greenhouse environment model*And stored to a model parameter set omega, where,
Figure BDA0002961672090000071
Figure BDA0002961672090000072
all represent the optimal solution of the nonlinear time-varying model parameters.
Step S7: judging whether t is less than the terminal time t of the control processf(ii) a If t is less than the control process terminal time tfIf t is t +1, the process returns to step S3; if t is greater than or equal to the control process terminal time tfOutputting the model parameter set omega;
step S8: determining an identification general model parameter solution theta 'according to the optimal solution of each nonlinear time-varying model parameter in the model parameter set omega, wherein theta ═ theta'1,θ′2,θ′3,θ′4,θ′5,θ′6,θ′7]，θ′1～θ′7Both represent the optimal model parameters.
Step S9: substituting the identification general model parameter solution into a greenhouse environment model to obtain a greenhouse environment predicted value; the greenhouse environment prediction value comprises the following steps: a predicted value of indoor temperature, a predicted value of indoor humidity, and a predicted value of indoor CO2 concentration.
The individual steps are discussed in detail below:
the invention adopts a 3-state dynamic model structure, which comprises 3 predicted values of indoor temperature, humidity and CO2 concentration, and other environmental parameters are regarded as certain nonlinear time-varying parameters. Then, according to the basic principles of heat exchange and substance exchange between indoor air and various mechanisms during the change of the greenhouse environment, the greenhouse environment model is represented as follows:
Figure BDA0002961672090000073
wherein, CapairRepresents the volumetric heat capacity, theta, of the indoor air below the inner heat-insulating mesh'1,θ′2,θ′3,θ′4,θ′5,θ′6,θ′7All represent the optimal model parameter, TsoilIndicating the surface soil temperature, TairIndicating a predicted value of the indoor temperature, ToutRepresenting the outdoor temperature, pairDenotes the air density, Cp,airDenotes the specific heat capacity of air, phisideventShowing the side window ventilation rate of the greenhouse after considering the heat preservation net in the greenhouse, the air leakage of the greenhouse and the chimney effectThScrShowing the convective air flow, T, between the upper and lower air of a heat-insulating net in the greenhousetopIndicating the temperature, T, of the air above the inner insulationcanDenotes the canopy temperature, IglobRepresenting solar radiation above the cover, lambda representing the latent heat coefficient, ufogIndicating the actual control quantity of spray, #fogIndicating the rated flow, u, of the spraying systemheatRepresents the actual control quantity of direct air heating of the greenhouse system, phiheatRepresenting the nominal heat flow, p, of the heating systemwaterDenotes the density of water, Cp,waterDenotes the specific heat capacity of water, vwaterIndicating the flow rate of hot water pipe, Tpipe,inIndicates the water temperature T at the inlet of the heating pipepipe,outIndicates the water temperature at the outlet of the heating pipe, AgDenotes the area of the greenhouse, HgIndicating the greenhouse height, wairDenotes a predicted value of indoor humidity, E denotes a canopy transpiration rate, McondDenotes the water vapor condensation coefficient of the cover layer, woutDenotes the outdoor humidity, wtopIndicating the humidity, CO, of the air above the heat-insulating net2,airRepresents a predicted value of the indoor CO2 concentration,
Figure BDA0002961672090000081
indicating the actual amount of CO2 control,
Figure BDA0002961672090000082
indicating the photosynthetic rate of the crop, PgIndicating canopy photosynthetic rate, RmIndicating the sustained respiration rate, CO, of the crop2,outIndicating outdoor CO2 concentration, CO2,topIndicating the CO2 concentration of the air above the heat retention screen.
The invention considers 7 nonlinear time-varying model parameters theta1～θ7. Due to the complexity of greenhouse environment changes, although research on greenhouse environment changes has been conducted for decades, there are still many mechanisms that have not been mastered and, therefore, there are many unmodeled dynamics. The basic thermodynamic dynamics in the model are still not sufficient to fully describe the greenhouse environment changesThe dynamics of the modeling, therefore, the non-modeled dynamics need to be represented by the nonlinear time-varying model parameter theta in the structured data modeling process of the invention1～θ7This is an important reason for these model parameters to have nonlinear time-varying characteristics. However, the basic heat exchange and material exchange mechanism in the model is a basic guarantee for ensuring that the constructed greenhouse environment model has global generalization capability, so the heat exchange and material exchange process model needs to be constructed in a mechanistic simulation mode. The following sections are general descriptions of various basic mechanistic process models.
Many studies are available that give a general model of the greenhouse ventilation process, as shown in equations (2) and (3):
Figure BDA0002961672090000083
Figure BDA0002961672090000084
wherein the content of the first and second substances,
Figure BDA0002961672090000085
specific formulas for calculating the skylight ventilation rate and the side window ventilation rate of the greenhouse considering the heat preservation net in the greenhouse, the greenhouse air leakage and the chimney effect are as follows (4) and (5):
Figure BDA0002961672090000091
Figure BDA0002961672090000092
wherein
Figure BDA0002961672090000093
Figure BDA0002961672090000094
Wherein phi isventroofShowing the ventilation rate of the skylight of the greenhouse after considering the heat preservation net in the greenhouse, the air leakage of the greenhouse and the chimney effect, [ m ]3/(m2·s)]，φsideventShowing the ventilation rate of the side window of the greenhouse after considering the heat preservation net in the greenhouse, the air leakage of the greenhouse and the chimney effect, [ m ]3/(m2·s)]，φ′ventroofShowing the ventilation rate of the skylight of the greenhouse after not considering the heat preservation net in the greenhouse, the air leakage of the greenhouse and the chimney effect, [ m ]3/(m2·s)]，φ′sidtventShowing the side window ventilation rate of the greenhouse after not considering the indoor heat preservation net and the air leakage and chimney effect of the greenhouse, [ m ]3/(m2·s)]，cleakageThe air leakage rate of the greenhouse is shown,
Figure BDA0002961672090000096
represents the average, eta, of indoor and outdoor temperaturesroofAnd ηsideRepresenting the ratio of the ventilation area, eta, of the roof and side windows, respectively, to the greenhouse arearoofthrDenotes the constant of the greenhouse chimney Effect, CwIndicates the ventilation pressure coefficient uroofventRepresenting the actual control variable for skylight ventilation, AroofDenotes the ventilation area of the skylight, CdDenotes the ventilation shape factor, AgDenotes the greenhouse area, g denotes the acceleration of gravity, hventIndicating the height, T, of the greenhousetopIndicating the temperature, T, of the air above the inner insulationoutWhich is indicative of the outdoor temperature of the room,
Figure BDA0002961672090000095
Vwindindicating outdoor wind speed, usideventRepresenting the actual control variable for side window ventilation, AsideIndicates the ventilation area of the side window phileakageIndicates the air leakage rate of the greenhouse, uscrRepresents the actual control quantity h of the inner heat-preservation netsiderooofIndicating the height, T, of the side window to the skylightairIndicating a predicted value of the indoor temperature.
The specific formula for calculating the convection airflow between the upper air and the lower air of the heat-insulating net in the greenhouse is as follows:
φThScr＝uscrKThScr|Tair-Tout|0.66+0.2(1-uscr) (8)；
wherein phi isThScrShowing the convective air flow [ m ] between the upper and lower air of the heat-insulating net in the greenhouse3/(m2·s)]，KThScrAs convection coefficient, (m)3/(m2·K0.66·s))，uscrRepresents the actual control quantity, T, of the internal heat-preservation networkairIndicating a predicted value of the indoor temperature, ToutIndicating the outdoor temperature.
The concrete formula for calculating the transpiration rate of the crops is as follows:
Figure BDA0002961672090000101
wherein E represents the transpiration rate of the crop, [ kg/(m)2·s)]，ρairDenotes the air density, Cp,airDenotes the specific heat capacity of air, LAI denotes the leaf area index, VPcanIndicating the saturated water pressure of the canopy, VPairDenotes the pressure of water in the room, lambda denotes the latent heat coefficient, gamma denotes the thermodynamic constant, rbAir pore impedance (s/m), r representing water vapor transmission of crop canopy boundary layersThe breathing air pore impedance, which represents the boundary layer of the blade, can be calculated from the following equation
rs＝rs,minf1(Rcan)·f2(CO2,air)·f3(VPcan-VPair) (10)；
Wherein the content of the first and second substances,
Figure BDA0002961672090000102
f2(CO2,air)＝1+c3(CO2,air-200)2，f3(VPcan-VPair)＝1+c4(VPcan-VPair)，c1～c4representing four known model parameters, rs,minRepresenting the breathing pore impedance, R, of the minimum blade boundary layercanRepresenting solar radiation, CO, above the canopy2,airPredicted value, VP, representing the indoor CO2 concentrationcanIndicating the saturated water pressure of the canopy, (Pa), VPairIndicating the water pressure in the room.
Canopy saturated water vapor pressure VPcanCalculated from equation (11):
Figure BDA0002961672090000103
wherein, TcanIndicating the canopy temperature.
For a certain absolute humidity wairWater pressure VP of lower, indoor airairCalculated from equation (12):
Figure BDA0002961672090000104
wherein, wairDenotes a predicted value of indoor humidity, R denotes a gas molar constant, TairA predicted value representing the indoor temperature is indicated,
Figure BDA0002961672090000105
represents the molar mass of water.
According to the research result of crop photosynthesis, the photosynthetic rate of the canopy of the greenhouse crop can be considered according to a big leaf model, so that the specific formula for calculating the photosynthetic rate of the crop is as follows:
Figure BDA0002961672090000111
wherein, PgIndicates the photosynthetic rate of the crop, [ umol/m2s]J represents the transmission rate of light quantum, [ umol/m2s]，CO2,stomDenotes the CO2 concentration at the pore, (umol/mol), Γ denotes the CO2 concentration saturation point, Γ ═ cΓTcan(umol/mol), J is calculated from the formula (14):
Figure BDA0002961672090000112
wherein, JPOTRepresents the potential transmission rate of light quantum, [ umol/m2s]And alpha represents a photon conversion coefficient, PARcanRepresenting the photosynthetic effective radiation received by the canopy, theta represents the curvature of the transmission rate of the optical quantum, and the potential transmission rate J of the optical quantumPOTAs in formula (15):
Figure BDA0002961672090000113
wherein the content of the first and second substances,
Figure BDA0002961672090000114
representing the maximum potential photon transmission rate at a canopy temperature of 25 c,
Figure BDA0002961672090000115
Ejactivation energy for transmission of light quanta, Tcan,KDenotes the absolute temperature, T, of the canopy25,KRepresenting the absolute temperature of the canopy at 25 deg.C, R representing the gas molar constant, S and H both representing two model parameters,
Figure BDA0002961672090000116
the maximum potential light quantum transmission rate of the crop leaves at 25 ℃ is shown, and LAI is the leaf area index.
The specific formula for calculating the concentration of CO2 at the air holes of the boundary layer of the canopy is as follows:
CO2,stom＝ηCO2stomCO2,air (16)；
wherein, CO2,stomIndicating the CO2 concentration at the pores of the canopy boundary layer,
Figure BDA0002961672090000119
conversion factor, CO, representing the concentration of CO22,airRepresents a predicted value of the indoor CO2 concentration.
Photosynthetically active radiation PAR received by the canopycanComprises two parts, one part is photosynthetically active radiation PAR above canopyghcanThe other part is photosynthetically active radiation PAR reflected from the ground facing the canopyflrcanIt can be calculated from equations (17) and (18), respectively:
Figure BDA0002961672090000117
Figure BDA0002961672090000118
wherein eta iscovDenotes the light transmittance of the cover layer, IglobRepresenting solar radiation above the overburden, LAI representing the leaf area index, pflrRepresenting the solar radiation reflectance, PAR, of the surface soilghIndicating photosynthetically active radiation of the top layer of the greenhouse, KcanRepresenting the canopy extinction coefficient. Thus having PARcan＝PARghcan+PARflrcan。
If the light respiration and the growth respiration are ignored, the maintenance respiration rate of the crop is calculated by equation (19):
Figure BDA0002961672090000121
wherein R ismIndicates the sustained respiration rate of the crop, [ umol/m2s]，DMfruit,DMstemAnd DMleafRespectively, the dry matter of the fruit, stem and leaf of the crop, and cfruit,cstemAnd cleafEach representing a corresponding maintenance breathing coefficient.
When the temperature in the room is higher than the temperature of the cover layer, moisture in the air condenses on the inner surface of the cover layer, thereby reducing the moisture content in the air, resulting in a decrease in the humidity of the air. Amount of water vapor condensation M on inner surface of upper covering layer per unit areacondCalculated from equation (20):
Figure BDA0002961672090000122
wherein, KrRepresenting the heat exchange coefficient of air with the covering layer, VPairIndicating the pressure of water in the room, VPcovSaturated water pressure of the inner surface of the cover layer, (Pa), calculated from equation (21):
Figure BDA0002961672090000123
wherein, TcovThe temperature of the heat-retaining net is indicated.
Strictly speaking, several environmental factors, namely the temperature of the crop canopy, the temperature above the greenhouse heat-insulating net, the humidity, the concentration of CO2, the temperature of surface soil and the temperature of a covering layer, are also important components of the greenhouse environment, but because the environmental factors do not directly influence the growth of crops, but indirectly influence the growth of crops by changing the environment near the crop canopy, the environmental factors are only regarded as certain time-varying parameters of a greenhouse environment model from the aspects of greenhouse environment control and crop growth simulation. They can be generally expressed as non-linear functions of temperature, humidity and CO2 concentration near the crop canopy.
Canopy temperature TcanCalculated from equation (22):
Figure BDA0002961672090000124
wherein, TairExpressing the predicted value of the indoor temperature, gamma denotes the thermodynamic constant, rvRepresenting the water vapor transmission impedance, rv＝0.92rH，(s/m)，ρairDenotes the air density, Cp,airDenotes the specific heat capacity of air, RcanRepresenting solar radiation above the canopy, rHRepresenting the canopy Heat transfer resistance, (s/m), VPDair＝VPcan-VPairIndicating the saturated water pressure VP of the canopycanAnd the indoor water pressure VPairThe difference between the water vapor pressure and the water vapor pressure, s represents the saturation slope of the water vapor pressure difference and the temperature, (Pa/K),typically set at 145.
Canopy heat transfer impedance rHCalculated from equation (23):
Figure BDA0002961672090000131
wherein alpha ism∈[330,670],βm∈[2.2,7.6]All represent constants, D represents the size constant of the crop leaf, and is generally between 0.05 and 0.15]The value (m) and v represent the air flow velocity near the canopy, and generally an empirical constant, (m/s) and T can be takenairIndicating a predicted value of the indoor temperature, TcanIndicating the canopy temperature.
Due to canopy heat transfer resistance rHIs itself influenced by the temperature of the canopy, and therefore this parameter is estimated by loop iteration, with an initial value
Figure BDA0002961672090000132
Wherein alpha isfDenotes an empirical constant, α f150. Then the canopy temperature and heat transfer resistance rHThe iterative calculation process of the loop in between is shown in fig. 3.
According to the principle of conservation of heat and substances, the equilibrium equation of heat flow, water vapor and CO2 of the air above the greenhouse heat-preservation net is as follows
Qair_top-Qtop_cov-Qroofvent＝0 (24)；
Wair_top-Mtop_cov-Wroofvent＝0 (25)；
Cair_top-Croofvent＝0 (26)；
Wherein Q isair_topRepresents the heat exchange quantity between the air in the upper layer and the air in the lower layer of the heat preservation net, (W/m2), Qtop_covRepresents the amount of heat exchange between the air above the heat-retaining mesh and the cover layer, (W/m2), QroofventExpressed as the amount of heat exchange between the inside and outside of the greenhouse by ventilation, (W/m2), Wair_topRepresents the water vapor between the air at the upper layer and the lower layer of the heat preservation net, (g/m2), Cair_topIndicating the amount of CO2 exchanged between the air in the upper and lower layers of the insulation net, (umol/M2), Mtop_covIndicating water condensation on the cover, (g/m2), WroofventDenotes indoor and outdoor moisture due to ventilation operation, (g/m2), Cair_topIndicating the amount of CO2 exchanged indoors and outdoors (umol/m2) due to the ventilation operation. From the above equation of heat and mass balance, the following static equation can be obtained:
ρairCp,airφThScr(Tair-Ttop)-kr(Ttop-Tcov)-ρairCp,airφventroof(Ttop-Tout)＝0 (27)；
φThScr(wair-wtop)-Mtop_cov-φventroof(wtop-wout)＝0 (28)；
φThScr(CO2,air-CO2,top)-φventroof(CO2,top-CO2,out)＝0 (29)；
wherein k isrIs the heat exchange coefficient between the indoor air and the blanket under steady state conditions. Similarly, the temperature of the coating can be approximated by an algebraic mean of the indoor and outdoor temperatures:
Figure BDA0002961672090000141
where ε is an empirical constant, typically taken to be 3. This results in the temperature, humidity and CO2 concentration of the air above the insulating mesh, namely:
Figure BDA0002961672090000142
Figure BDA0002961672090000143
Figure BDA0002961672090000144
since the thermal capacity of the soil is much greater than that of air, the change in the surface soil temperature always lags behind the change in the air temperature, so that the soil temperature can be estimated using a time-series function without soil sensors or by simulating long-term changes in the greenhouse environment, i.e. the soil temperature is estimated using a time-series function
Tsoil(k+1)＝Tsoil(k)+[ksoil_air(Tair(k)-Tflr(k))+ηsoilIglob-kd(Tsoil(k)-Td)]/Capsoil (34)；
Where k is a discrete time variable, ksoil_airIs the heat exchange coefficient between air and surface soil, kdIs the heat exchange coefficient between surface soil and deep soil (W/m)2K) And T isdThe soil temperature of the constant temperature layer is usually 15 ℃. For surface soil two cm later, its heat capacity CapsoilCan be in the range of [ 1.2X 10 ]5,1.7×105]Taking the value between (J/Kg.K). And the surface soil has an absorptivity of solar radiation of
Figure BDA0002961672090000145
The invention and other greenhouse environment structured data modeling methods are that the invention does not search the possible optimal model parameters of the model through an optimization method, but adopts a Lyapunov stability control-based mode to design an adaptive controller by taking the model parameters as control variables to drive the model to output predicted values to gradually converge on actual measurement data of the greenhouse. Since the measurement data is sampled at a certain sampling period and is discrete data, interpolation calculation must be performed on the measurement data before adaptive control is performed, so that the measurement data forms a greenhouse environment change track with time as a variable on a time axis. To achieve this, the present invention uses a parameterized cubic spline interpolation technique to fit the measured data of the greenhouse environment to maintain the smoothness of the interpolated curve of the measured data, as follows.
Let N be the total number of measurement data samples used for parametric modeling, Ts be the sampling period, and define the following greenhouse environment and control variable data sets:
Tair_set＝{Tair(k),k＝1,2,…,N}
Hair_set＝{Hair(k),k＝1,2,…,N}
CO2,air_set＝{CO2,air(k),k＝1,2,…,N}
Uheat_set＝{uheat(k),k＝1,2,…,N}
Qpipe_set＝{qpipe(k),k＝1,2,…,N}
Usidevent_set＝{usidevent(k),k＝1,2,…,N}；
Uroofvent_set＝{uroofvent(k),k＝1,2,…,N}
Uscr_set＝{uscr(k),k＝1,2,…,N}
where k samples a discrete time index variable, Tair_set、Hair_set、CO2,air_setRespectively measured data sets of indoor temperature, humidity and carbon dioxide concentration, Tout_set、Hout_set、CO2,out_set、Vwind_set、Iglob_setRespectively outdoor temperature, humidity, carbon dioxide concentration, wind speed and solar radiation intensity measurement data sets, Uheat_set、Qpipe_set、Usidevent_set、Uroofvent_set、Uscr_setAnd the control signal measurement data sets are respectively an air direct heating system, a pipeline hot water circulation heating system, side window ventilation, skylight ventilation and an inner heat-preservation network.
Assuming that the measurement data is represented by X ═ { X (k) —, k ═ 1,2, …, N }, the time response curve is
Figure BDA0002961672090000151
Wherein
Figure BDA0002961672090000152
Is a time interval [ Ts.k, Ts. (k +1)]Normalized time variable, which can be calculated from the following equation
Figure BDA0002961672090000161
Where t is the actual process control time.
Polynomial coefficient a according to spline curve end point propertyk、bk、ckAnd dkCan be determined by equation (37):
ak＝x(k)
bk＝Dk
ck＝3[x(k+1)-x(k)]-2Dk-Dk+1 (37)；
dk＝-2[x(k+1)-x(k)]+Dk+Dk+1
wherein D iskThe gradient at point x (k). From the second order differential of the spline starting point x (k) and ending point x (k +1), the relationship shown as (38) can be obtained:
Dk-1+4Dk+Dk+1＝3[x(k+1)-x(k-1)] (38)；
if the second derivative of the first point x (0) and the last point x (N) of the measurement data is set to 0, it can be obtained
2D0+D1＝3[x(1)-x(0)](39)；
2DN-1+DN＝3[x(N)-x(N-1)] (40)；
Then for a measurement data set with N samples, the following interpolation matrix equation can be obtained:
Figure BDA0002961672090000162
therefore, a time response curve, also called a reference input track, of the environmental change of the greenhouse and the control variable of the actuating mechanism in the data measurement time interval can be obtained; the reference input trajectory includes: actual measurement of indoor temperature
Figure BDA0002961672090000163
Actual measurement of indoor humidity
Figure BDA0002961672090000164
Actual measurement of indoor CO2 concentration
Figure BDA0002961672090000165
Actual control quantity u for direct heating of air in greenhouse systemheatActual measurement q of hot water circulation heating of pipelinepipeActual control quantity u of side window ventilationsideventActual control amount u of sunroof ventilationroofventActual control quantity u of inner heat-insulating netscrActual measured value T of outdoor temperatureoutActual measurement H of outdoor humidityoutActual measurement of outdoor carbon dioxide concentration CO2,outActual measurement V of outdoor wind speedwindActual measurement I of solar radiation above the coverglob。
Step S4: updating an initial greenhouse environment model based on the reference input trajectory.
The conventional greenhouse environment model can be generally expressed in the form of nonlinear differential equations, i.e.
Figure BDA0002961672090000171
Wherein the content of the first and second substances,
Figure BDA0002961672090000172
the method comprises the steps of (1) modeling a greenhouse environment, wherein f (x, v) is an internal dynamic nonlinear function of a greenhouse environment system, g (x, v) is a control gain matrix of the greenhouse environment system, x is a greenhouse environment state variable, v is an outdoor weather variable, and u is a control variable of an actuator. The two non-linear functions f (x, v) and g (x, v) are unknown.
The invention considers the mechanism of heat exchange and material exchange process, while other unmodeled dynamics are formed by a nonlinear time-varying model parameter theta ═ theta1,…θl]TTo reflect it. The greenhouse environment model considered by the present invention thus has the following form:
Figure BDA0002961672090000173
in order to obtain the distribution rule of the model parameter theta under different conditions, the model (2) needs to be transformed necessarily before identifying the model parameter, so that the model parameter has the following form:
Figure BDA0002961672090000174
wherein the content of the first and second substances,
Figure BDA0002961672090000175
and
Figure BDA0002961672090000176
each representing a known non-linear function obtained by a term shift, the non-linear time varying model parameters theta being considered as virtual control inputs of the tracking control system,
Figure BDA0002961672090000177
the output variable of the state is represented,
Figure BDA0002961672090000178
Tair,Hair,CO2,airrespectively indicate the predicted values of the greenhouse environment temperature, humidity and carbon dioxide concentration, u indicates the control variable of an actuator, and u ═ u [ u ]heat,qpipe,usidevent,uroofvent,uscr]T，uheat,qpipe,usidevent,uroofvent,uscrRespectively representing the actual control quantity of the greenhouse system, such as direct air heating, pipeline hot water circulation heating, side window ventilation, skylight ventilation and an inner heat-preservation net, wherein v represents the outdoor weather variable, and v is [ T ═ T [out,Hout,CO2,air,Vwind,Iglob]T，Tout,Hout,CO2,air,Vwind,IglobRespectively representing the actually measured outdoor temperature, humidity, carbon dioxide concentration, wind speed and solar radiation.
For a given system reference input, a controller can be designed to obtain a trajectory of change of this virtual control input signal to converge the predicted value on the system reference input. That is, the goal of model parameter identification is to make the model (43) approximate the following equation as much as possible under the optimal model parameters:
Figure BDA0002961672090000181
wherein the content of the first and second substances,
Figure BDA0002961672090000182
an initial greenhouse environment model is represented, and,
Figure BDA0002961672090000183
and
Figure BDA0002961672090000184
all represent known nonlinear functions approximating the actual greenhouse environment state, theta' represents the optimal model parameter variable, xrRepresents the actual greenhouse environment state variable and is provided with a plurality of temperature sensors,
Figure BDA0002961672090000185
respectively, the actual measured greenhouse temperature, humidity and carbon dioxide concentration, u represents the control variable of the actuator, and u ═ u [ u ]heat,qpipe,usidevent,uroofvent,uscr]T，uheat,qpipe,usidevent,uroofvent,uscrRespectively representing the actual control quantity of the greenhouse system, such as direct air heating, pipeline hot water circulation heating, side window ventilation, skylight ventilation and an inner heat-preservation net, wherein v represents the outdoor weather variable, and v is [ T ═ T [out,Hout,CO2,air,Vwind,Iglob]T，Tout,Hout,CO2,air,Vwind,IglobRespectively representing the actually measured outdoor temperature, humidity, carbon dioxide concentration, wind speed and solar radiation.
Equation (44) outputs the model state as much as possible
Figure BDA0002961672090000186
Approaching to the actual greenhouse environment state variable
Figure BDA0002961672090000187
And as a state variable of the control system in the control process, theta' is an optimal value of theta, namely the final output of the greenhouse environment model parameter identification.
Step S5: and constructing a model parameter identification controller based on the Lyapunov function.
Since the greenhouse environment model considered by the invention only has 3 state variables, and the nonlinear time-varying model parameter set theta to be identified has 7 nonlinear time-varying model parameters, the system is a typical over-activator system. Aiming at the system, the invention adopts an adaptive control method based on dynamic control distribution to design a corresponding model parameter identification controller. The greenhouse environment model considered by the invention can be written in the following general form of nonlinear system dynamic differential equation:
Figure BDA0002961672090000188
wherein x ∈ Rn,u∈RmRespectively, a state variable and a control variable of the system, f (·) epsilon RnFor a known nonlinear function, g (-) is the control matrix and θ is the identified set of nonlinear time-varying model parameters. For model parameter identification, the state and control inputs to the system are known measurement data. In fact, the measured greenhouse environment system control input can also be seen as an externally known input with respect to time, just like outdoor weather. In order to identify the nonlinear time-varying model parameters in the model parameter set θ, the model parameters can be regarded as the virtual control input of the system in the virtual control system, so the objective of the virtual control design is to design an effective control law to drive the system to predict the value tracking measurement data. Then the following can be definedTracking error dynamics:
Figure BDA0002961672090000191
wherein e ═ x-xrTo control the tracking error of the system, xrIn order to measure the time-varying trajectory of the state data,
Figure BDA0002961672090000192
given the first differential of the state trajectory, e is shorthand for e (t), x is shorthand for x (t), xrIs xrThe abbreviation of (t) is used,
Figure BDA0002961672090000193
is the derivative of x and is,
Figure BDA0002961672090000194
is xrThe derivative of (c).
Defining τ ═ g (x, u, θ) as the total driving force that the driving model tracks the measurement state trajectory, then we can:
Figure BDA0002961672090000195
it is clear that if there is one nonlinear time-varying model parameter set θ*The following equation:
τ*＝g(x,u,θ*)＝-K·e-f(x,u)+v (48)；
where K is a given positive definite matrix, the systematic error (47) can converge to the origin. However optimum driving force τ*It may not be reachable, which means that the optimum driving force τ is*There is a certain error with g (x, u, theta), i.e.
σ＝g(x,u,θ)-τ* (49)；
To minimize this error, one can find the nonlinear time-varying model parameter set θ by going to the following optimization problem*:
Figure BDA0002961672090000196
Wherein H1、H2Are all positive definite matrixes given by a designer, J represents an error norm, RPA search space is represented in which the search space,
Figure BDA0002961672090000197
to reduce this error, it is generally required to make H1||＞＞||H2L. If it is not
Figure BDA0002961672090000198
Then this optimization problem can take a local minimum. And if g (x, u, theta) has a linear relationship with the model parameter theta, then:
g(x,u,θ)＝g(x,u)θ (51)；
the optimization problem is then a convex optimization problem such that when satisfied
Figure BDA0002961672090000209
The global minimum can be obtained, meaning the obtained optimal model parameter θ*Is globally optimal.
For greenhouse environment models, it is desirable that the model parameters in equation (1) be as positive as possible, since the heat transfer coefficient is generally greater than 0, i.e., θ ≧ 0 is required, based on thermodynamic principles. It is therefore necessary to introduce the following penalty function in the above optimization problem:
Figure BDA0002961672090000201
wherein the content of the first and second substances,θ i、
Figure BDA0002961672090000202
respectively representing the ith model parameter thetaiP represents the total number of nonlinear model parameters.
The above optimization problem becomes as follows:
J＝J1+J2+μS(θ) (53)；
where μ is the Lagrangian coefficient. Thus, if there is a constant χ > 0 satisfied
Figure BDA0002961672090000203
Ip∈Rp×pIs an identity matrix, Rp×pTo search the space, the optimization problem (53) can then achieve global optimality when the following conditions are satisfied.
Figure BDA0002961672090000204
Figure BDA0002961672090000205
Wherein the content of the first and second substances,
Figure BDA0002961672090000206
and is
Figure BDA0002961672090000207
S is a abbreviation for S (θ).
Order to
Figure BDA0002961672090000208
To solve the above adaptive optimization problem, the following quasi-Lyapunov function may be defined:
Figure BDA0002961672090000211
then the first order differential of the Lyapunov function can be obtained, namely:
Figure BDA0002961672090000212
Figure BDA0002961672090000213
by shifting terms, we can get:
Figure BDA0002961672090000214
order to
Figure BDA0002961672090000215
Figure BDA0002961672090000216
Figure BDA0002961672090000217
Assuming that there is a constant ω > 0, then:
Figure BDA0002961672090000218
in order to ensure that the predicted value can converge on the actual measured data change track, the model parameter identification controller is determined as shown in formulas (64) and (65):
Figure BDA0002961672090000219
Figure BDA00029616720900002110
wherein, gamma is1,Γ2Are all given positive definite matrix, ξ1,ξ2Are adaptive terms of the parameter identification controller. Bringing this model parameter control law and lagrangian coefficient update law into (59) availability:
Figure BDA0002961672090000221
in this way, the following Lyapunov function may be selected for the tracking control system:
Figure BDA0002961672090000222
wherein eta is1,η2,η3,η4A normal number given by the designer, and σ1,σ2Is an adaptation term. Then the first order differential of the Lyapunov function described above can be obtained:
Figure BDA0002961672090000223
thus, ξ can be selected as follows1,ξ2σ1And σ2Adaptive law:
Figure BDA0002961672090000224
Figure BDA0002961672090000225
Figure BDA0002961672090000226
Figure BDA0002961672090000227
substituting (69) - (72) into (68) yields:
Figure BDA0002961672090000228
this means that the tracking error of the tracking control system can asymptotically converge to zero.
Statistical analysis of the correlation between nonlinear time-varying model parameters and greenhouse environment factors:
if the model parameters themselves are constant constants, the model parameter variation trajectories generated by the parameter recognition controller will eventually converge on this constant. However, in the greenhouse environment model considered in the present invention, the model parameters need to reflect unmodeled dynamics of greenhouse environment changes, and therefore these model parameters usually have strong nonlinear time-varying characteristics, so the model parameter change trajectory output by the parameter identification controller appears as a certain complex nonlinear curve and is usually affected by changes of greenhouse environment factors. Then the correlation between these model parameters and the environmental factors can be found statistically and the correlation function relationship between the model parameters and the environmental factors can be established.
Nonlinear time-varying model parameter θ1It can be seen as a generalized heat transfer coefficient for heat exchange between the surface soil and the indoor air, and therefore this parameter is usually related to the temperature difference between the surface soil and the indoor air, and therefore the parameter θ can be given1Considered as some non-linear function of this temperature difference, namely:
θ1＝f(Tair-Tsoil) (74)。
therefore, the optimal solution of the nonlinear time-varying model parameters at a plurality of different moments in the model parameter set omega is obtained
Figure BDA0002961672090000231
Performing polynomial fitting to obtain a solution theta of the parameters of the identified general model'1The concrete formula is as follows: theta1′＝p1(Tair-Tsoil)6+p2(Tair-Tsoil)5+p3(Tair-Tsoil)4+p4(Tair-Tsoil)2+p5(Tair-Tsoil)+p6(75)；
Wherein p is1,p2,p3,p4,p5,p6All represent polynomial coefficients, TairIndicating a predicted value of the indoor temperature, TsoilIndicating the surface soil temperature.
Parameter theta3' is a parameter related to greenhouse ventilation, and because the indoor and outdoor heat exchange and the water vapor exchange are usually single, the change of the parameter is relatively smooth and close to a constant, and the parameter theta can be analyzed in a statistical manner3' fixed value by a constant
Figure BDA0002961672090000232
To do so, i.e.
Figure BDA0002961672090000233
Therefore, the optimal solution of the nonlinear time-varying model parameters at a plurality of different moments in the model parameter set omega is obtained
Figure BDA0002961672090000234
Carrying out averaging to obtain a solution theta of the parameters of the identification general model'3The concrete formula is as follows:
Figure BDA0002961672090000235
wherein the content of the first and second substances,
Figure BDA0002961672090000236
representing a constant.
To some extent, the model parameters of the greenhouse environment model are affected by solar radiation. The influence of solar radiation is generally considered when analyzing the correlation between the model parameters and the greenhouse environment factors, and the main reason is that the greenhouse environment change is greatly influenced by the solar radiation actually, so the dynamics of the greenhouse environment change can be greatly different in the daytime with the solar radiation and at night without the solar radiation. E.g. parameter theta4The influence of solar radiation is large in daytime, so that the method determines and identifies the solution theta 'of the parameters of the general model by adopting an exponential fitting method'4But is maintained around a constant value during the night, i.e. fluctuates
Figure BDA0002961672090000237
And constant
Figure BDA0002961672090000238
May be determined by the night parameter θ'4The mean of the trajectories is controlled. Therefore, the invention provides the optimal solution of the nonlinear time-varying model parameters according to a plurality of different moments in the model parameter set omega
Figure BDA0002961672090000239
Determining and identifying theta 'of solution of parameters of general model'4The concrete formula is as follows:
Figure BDA0002961672090000241
wherein, IglobWhich represents the solar radiation above the cover layer,
Figure BDA0002961672090000242
representing a constant.
In daytime, the optimal solution of the nonlinear time-varying model parameters at a plurality of different moments in the model parameter set omega is obtained
Figure BDA0002961672090000243
Performing polynomial fitting to obtain a solution theta of the parameters of the identified general model'5The concrete formula is as follows:
Figure BDA0002961672090000244
wherein q is0～q9All represent polynomial coefficients, TcanIndicating the canopy temperature.
At night, the optimal solution theta of the nonlinear time-varying model parameters at a plurality of different moments in the model parameter set omega is obtained5Averaging to obtain a solution theta 'of the parameters of the identified general model'5The concrete formula is as follows:
Figure BDA0002961672090000245
wherein the content of the first and second substances,
Figure BDA0002961672090000246
representing a constant.
In addition to the correlation between model parameters and greenhouse environment factors, it is possible that there is some correlation between certain model parameters, e.g. θ'2And θ'4There is a highly linear correlation between them, i.e. there is a linear relationship as follows.
According to the invention, theta 'is solved according to the identified general model parameter'4Determining and identifying theta 'of solution of parameters of general model'2The concrete formula is as follows:
θ′2＝1.593θ′4+0.008。
it should be noted that in greenhouses with different structures and materials, the correlation between the model parameters and the greenhouse environment may be different from the above functional relationship, so that statistical analysis needs to be performed again on the model parameter identification results according to the specific greenhouse to find the correlation between the model parameters and the greenhouse environment.
The optimal solution of the nonlinear time-varying model parameters at a plurality of different moments in the model parameter set omega is obtained
Figure BDA0002961672090000247
Carrying out averaging to obtain a solution theta of the parameters of the identification general model'6The concrete formula is as follows:
Figure BDA0002961672090000248
wherein, Cday、CnightTwo different constants are represented, daytime for day and night for night, respectively.
The optimal solution of the nonlinear time-varying model parameters at a plurality of different moments in the model parameter set omega is obtained
Figure BDA0002961672090000251
Carrying out averaging to obtain a solution theta of the parameters of the identification general model'7The concrete formula is as follows:
Figure BDA0002961672090000252
wherein D isday、DnightTwo different constants are represented, daytime for day and night for night.
Figure BDA0002961672090000253
Figure BDA0002961672090000261
The invention also provides an agricultural greenhouse environment prediction system, which comprises:
and the given module is used for giving the terminal time of the control process.
The acquisition module is used for acquiring measurement data; the measurement data includes greenhouse environment state variables, control signals, and outdoor weather.
And the reference input track determining module is used for carrying out interpolation calculation according to the measurement data to obtain a reference input track.
And the updating module is used for updating the initial greenhouse environment model based on the reference input track.
And the construction module is used for constructing the model parameter identification controller based on the Lyapunov function.
And the calculation module is used for calculating the optimal solution of each nonlinear time-varying model parameter at the t moment according to the model parameter identification controller and the initial greenhouse environment model and storing the optimal solution to a model parameter set omega.
The judging module is used for judging whether t is less than the terminal time of the control process; if t is less than the terminal time of the control process, making t equal to t +1, and returning to the acquisition module; and if t is greater than or equal to the terminal time of the control process, outputting the model parameter set omega.
And the identification general model parameter solution determining module is used for determining an identification general model parameter solution according to the optimal solution of each nonlinear time-varying model parameter in the model parameter set omega.
And the prediction module is used for substituting the identification general model parameter solution into the greenhouse environment model to obtain a greenhouse environment prediction value.
As a selectable real-time manner, the module for determining a solution to the parameter of the identification generic model specifically includes:
a first optimal model parameter determining unit, configured to determine an optimal solution of nonlinear time-varying model parameters at multiple different time instants in the model parameter set Ω
Figure BDA0002961672090000271
Performing polynomial fitting to obtain a solution theta of the parameters of the identified general model'1The concrete formula is as follows:
θ′1＝p1(Tair-Tsoil)6+p2(Tair-Tsoil)5+p3(Tair-Tsoil)4+p4(Tair-Tsoil)2+p5(Tair-Tsoil)+p6；
wherein p is1,p2,p3,p4,p5,p6All represent polynomial coefficients, TairIndicating a predicted value of the indoor temperature, TsoilIndicating the surface soil temperature.
A second optimal model parameter determining unit, configured to determine an optimal solution of the nonlinear time-varying model parameters at multiple different times in the model parameter set Ω
Figure BDA0002961672090000272
Carrying out averaging to obtain a solution theta of the parameters of the identification general model'3The concrete formula is as follows:
Figure BDA0002961672090000273
wherein the content of the first and second substances,
Figure BDA0002961672090000274
representing a constant.
A third optimal model parameter determining unit, configured to determine an optimal solution of the nonlinear time-varying model parameters at multiple different times in the model parameter set Ω
Figure BDA0002961672090000275
Determining and identifying theta 'of solution of parameters of general model'4The concrete formula is as follows:
Figure BDA0002961672090000276
wherein, IglobWhich represents the solar radiation above the cover layer,
Figure BDA0002961672090000277
representing a constant.
A fourth optimal model parameter determining unit, configured to determine an optimal solution of the nonlinear time-varying model parameters at multiple different times in the model parameter set Ω
Figure BDA0002961672090000278
Performing polynomial fitting to obtain a solution theta of the parameters of the identified general model'5The concrete formula is as follows:
Figure BDA0002961672090000281
wherein q is0～q9All represent polynomial coefficients, TcanRepresenting canopy temperature, daytime day, and night.
A fifth optimal model parameter determining unit for solving theta 'according to the identified general model parameter'4Determining and identifying theta 'of solution of parameters of general model'2The concrete formula is as follows: theta'2＝1.593θ′4+0.008。
Sixth optimal model parameter determinationA unit for optimizing the solution of the nonlinear time-varying model parameters at different times in the model parameter set omega
Figure BDA0002961672090000282
Carrying out averaging to obtain a solution theta of the parameters of the identification general model'6The concrete formula is as follows:
Figure BDA0002961672090000283
wherein, Cday、CnightRespectively, two different constants.
A seventh optimal model parameter determining unit, configured to determine an optimal solution of the nonlinear time-varying model parameters at multiple different times in the model parameter set Ω
Figure BDA0002961672090000284
Carrying out averaging to obtain a solution theta of the parameters of the identification general model'7The concrete formula is as follows:
Figure BDA0002961672090000285
wherein D isday、DnightTwo different constants are represented.
The greenhouse environment model of the present invention was validated in a Venlo type agricultural greenhouse. The greenhouse is positioned in a Shanghai Chongming modern agriculture demonstration garden. The greenhouse is provided with greenhouse regulating and controlling devices such as spraying, hot water pipeline heating, an inner heat-preservation net, an outer sunshade net, a skylight, side window ventilation and the like. All greenhouse regulation and control devices are centrally regulated by a PRIVA greenhouse production management system. The greenhouse area is 1000 square meters, the height is 6 meters, the distance from the ground to the inner heat-preservation net is 4 meters, and the cultivated crop is cherry tomatoes. The planting time of crops is 2016, 9 and 1 days, the harvesting time is 2017, 5 and 18 days, and the planting density is 3.5. The greenhouse environment and control data acquisition covers the whole crop production period, and the sampling period is 5 minutes. Adaptive law adjusting parameter eta1,η2,η3,η4Is set to 0.01, and1,Γ2the setting positions are as follows:
Figure BDA0002961672090000291
positive definite matrix K, H1,H2Set to K ═ diag ([10, 10), respectively]),H1＝diag([50,50,...,50]),H2diag([1,1,...,1])。
The greenhouse environment model was validated under two weather conditions, cold and warm. The warm day is selected to be between 9 months and 10 days and 30 days, and the cold day is selected to be between 12 months and 10 days and 30 days, within the time covered by the measurement data. Greenhouse environment regulation device control inputs during the warm day validation time are shown below in fig. 4.
The results of tracking the temperature, humidity and CO2 concentration in the greenhouse are shown in fig. 5 and 6, Simulation means Simulation, and Measured means actual measurement. Greenhouse environment model parameters between 9 months and 10-12 days
Figure BDA0002961672090000292
As shown in fig. 7.
For the verified greenhouse environment changes, the statistical distribution of the model parameters and the correlation between the statistical distribution and the environment factors related to the greenhouse environment are shown in fig. 8.
According to the parameter theta1Statistical distribution rule and analysis of statistical correlation with temperature difference between surface soil and air, the temperature difference is [ -12,2 [ -12]Internal time theta1There is a correlation function with this temperature difference as follows:
Figure BDA0002961672090000293
the correlation coefficient was 0.91.
At daytime parameter θ'4The correlation function with solar radiation is as follows:
Figure BDA0002961672090000294
the correlation coefficient was 0.924, while the night fluctuated slightly around a constant of 0.08.
From parameter θ'2And θ'4In terms of the statistical distribution of the two types of the data, the two types of the data are storedIn high linear correlation, the correlation coefficient is as high as 0.994, and theta'2＝1.593θ′4+0.008。
In daytime, parameter θ'5The coefficients of the correlation polynomial function of (a) are shown in table 1, whereas at night this parameter is close to 0.1.
Similarly, a parameter θ 'can be obtained'6And θ'7The correlation function of (a) is as follows:
Figure BDA0002961672090000301
Figure BDA0002961672090000302
the principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.

Claims (10)

1. An agricultural greenhouse environment prediction method, characterized in that the method comprises:
step S1: giving a control process terminal time;
step S2: acquiring measurement data; the measurement data comprises greenhouse environment state variables, control signals and outdoor weather;
step S3: carrying out interpolation calculation according to the measurement data to obtain a reference input track;
step S4: updating an initial greenhouse environment model based on the reference input trajectory;
step S5: constructing a model parameter identification controller based on a Lyapunov function;
step S6: calculating optimal solutions of nonlinear time-varying model parameters at the t moment according to the model parameter identification controller and the initial greenhouse environment model, and storing the optimal solutions to a model parameter set;
step S7: judging whether t is less than the terminal time of the control process; if t is less than the control process terminal time, making t equal to t +1, and returning to step S2; if t is greater than or equal to the control process terminal time, outputting the model parameter set;
step S8: determining an identification general model parameter solution according to the optimal solution of the nonlinear time-varying model parameters at different moments in the model parameter set;
step S9: substituting the identification general model parameter solution into a greenhouse environment model to obtain a greenhouse environment predicted value; the greenhouse environment prediction value comprises the following steps: a predicted value of indoor temperature, a predicted value of indoor humidity, and a predicted value of indoor CO2 concentration.
2. The agricultural greenhouse environment prediction method of claim 1, wherein the greenhouse environment model is specifically formulated as:
Figure FDA0002961672080000011
wherein, CapairRepresents the volumetric heat capacity, theta, of the indoor air below the inner heat-insulating mesh'1,θ′2,θ′3,θ′4,θ′5,θ′6,θ′7All represent the optimal model parameter, TsoilIndicating the surface soil temperature, TairIndicating a predicted value of the indoor temperature, ToutRepresenting the outdoor temperature, pairDenotes the air density, Cp,airDenotes the specific heat capacity of air, phisideventShowing the side window ventilation rate of the greenhouse after considering the heat preservation net in the greenhouse, the air leakage of the greenhouse and the chimney effectThScrShowing the convective air flow, T, between the upper and lower air of a heat-insulating net in the greenhousetopIndicating the temperature, T, of the air above the inner insulationcanDenotes the canopy temperature, IglobRepresenting solar radiation above the cover, lambda representing the latent heat coefficient, ufogIndicating the actual control quantity of spray, #fogIndicating the rated flow, u, of the spraying systemheatIndicating air temperature of greenhouse systemReceiving the actual controlled quantity of heat, phiheatRepresenting the nominal heat flow, p, of the heating systemwaterDenotes the density of water, Cp,waterDenotes the specific heat capacity of water, vwaterIndicating the flow rate of hot water pipe, Tpipe,inIndicates the water temperature T at the inlet of the heating pipepipe,outIndicates the water temperature at the outlet of the heating pipe, AgDenotes the area of the greenhouse, HgIndicating the greenhouse height, wairDenotes a predicted value of indoor humidity, E denotes a canopy transpiration rate, McondDenotes the water vapor condensation coefficient of the cover layer, woutDenotes the outdoor humidity, wtopIndicating the humidity, CO, of the air above the heat-insulating net2,airRepresents a predicted value of the indoor CO2 concentration,
Figure FDA0002961672080000021
indicating the actual amount of CO2 control,
Figure FDA0002961672080000022
indicating the photosynthetic rate of the crop, PgIndicating canopy photosynthetic rate, RmIndicating the sustained respiration rate, CO, of the crop2,outIndicating outdoor CO2 concentration, CO2,topIndicating the CO2 concentration of the air above the heat retention screen.
3. The agricultural greenhouse environment prediction method of claim 1, wherein the determining an identification general model parameter solution according to the optimal solution of the nonlinear time-varying model parameters at different times in the model parameter set specifically comprises:
according to the optimal solution of the nonlinear time-varying model parameters at a plurality of different moments in the model parameter set
Figure FDA0002961672080000023
Performing polynomial fitting to obtain a solution theta of the parameters of the identified general model'1The concrete formula is as follows:
θ′1＝p1(Tair-Tsoil)6+p2(Tair-Tsoil)5+p3(Tair-Tsoil)4+p4(Tair-Tsoil)2+p5(Tair-Tsoil)+p6；
wherein p is1,p2,p3,p4,p5,p6All represent polynomial coefficients, TairIndicating a predicted value of the indoor temperature, TsoilRepresenting the surface soil temperature;
according to the optimal solution of the nonlinear time-varying model parameters at a plurality of different moments in the model parameter set
Figure FDA0002961672080000024
Carrying out averaging to obtain a solution theta of the parameters of the identification general model'3The concrete formula is as follows:
Figure FDA0002961672080000025
wherein the content of the first and second substances,
Figure FDA0002961672080000026
represents a constant;
according to the optimal solution of the nonlinear time-varying model parameters at a plurality of different moments in the model parameter set
Figure FDA0002961672080000027
Determining and identifying theta 'of solution of parameters of general model'4The concrete formula is as follows:
Figure FDA0002961672080000031
wherein, IglobWhich represents the solar radiation above the cover layer,
Figure FDA0002961672080000032
represents a constant;
according to a plurality of differences in the model parameter setTime-of-day nonlinear time-varying model parameter optimal solution
Figure FDA0002961672080000033
Performing polynomial fitting to obtain a solution theta of the parameters of the identified general model'5The concrete formula is as follows:
Figure FDA0002961672080000034
wherein q is0～q9All represent polynomial coefficients, TcanRepresenting canopy temperature, daytime, night;
solving theta 'according to the identified general model parameter'4Determining and identifying theta 'of solution of parameters of general model'2The concrete formula is as follows:
θ′2＝1.593θ′4+0.008；
according to the optimal solution of the nonlinear time-varying model parameters at a plurality of different moments in the model parameter set
Figure FDA0002961672080000037
Carrying out averaging to obtain a solution theta of the parameters of the identification general model'6The concrete formula is as follows:
Figure FDA0002961672080000035
wherein, Cday、CnightRespectively represent two different constants;
according to the optimal solution of the nonlinear time-varying model parameters at a plurality of different moments in the model parameter set
Figure FDA0002961672080000038
Carrying out averaging to obtain a solution theta of the parameters of the identification general model'7The concrete formula is as follows:
Figure FDA0002961672080000036
wherein D isday、DnightTwo different constants are represented.
4. The agricultural greenhouse environment prediction method of claim 1, wherein the model parameter identification controller is specifically formulated as:
Figure FDA0002961672080000041
wherein the content of the first and second substances,
Figure FDA0002961672080000042
the control law of the model parameters is represented,
Figure FDA0002961672080000043
representing the Lagrange coefficient update law, Γ1,Γ2Each representing a given positive definite matrix, ξ1,ξ2Both represent adaptive terms of the parameter recognition controller,
Figure FDA0002961672080000044
μ denotes lagrange coefficients, T denotes transposition,
Figure FDA0002961672080000045
j denotes the error norm and θ denotes the set of nonlinear time-varying model parameters.
5. The agricultural greenhouse environment prediction method of claim 1, wherein the initial greenhouse environment model is specifically formulated as:
Figure FDA0002961672080000046
wherein the content of the first and second substances,
Figure FDA0002961672080000047
an initial greenhouse environment model is represented, and,
Figure FDA0002961672080000048
and
Figure FDA0002961672080000049
all represent known nonlinear functions approximating the actual greenhouse environment state, theta' represents the optimal model parameter variable, xrRepresents the actual greenhouse environment state variable and is provided with a plurality of temperature sensors,
Figure FDA00029616720800000410
Figure FDA00029616720800000411
respectively, the actual measured greenhouse temperature, humidity and carbon dioxide concentration, u represents the control variable of the actuator, and u ═ u [ u ]heat,qpipe,usidevent,uroofvent,uscr]T，uheat,qpipe,usidevent,uroofvent,uscrRespectively representing the actual control quantity of the greenhouse system, such as direct air heating, pipeline hot water circulation heating, side window ventilation, skylight ventilation and an inner heat-preservation net, wherein v represents the outdoor weather variable, and v is [ T ═ T [out,Hout,CO2,air,Vwind,Iglob]T，Tout,Hout,CO2,air,Vwind,IglobRespectively representing the actually measured outdoor temperature, humidity, carbon dioxide concentration, wind speed and solar radiation.
6. An agricultural greenhouse environment prediction system, the system comprising:
the given module is used for giving the terminal time of the control process;
the acquisition module is used for acquiring measurement data; the measurement data comprises greenhouse environment state variables, control signals and outdoor weather;
the reference input track determining module is used for carrying out interpolation calculation according to the measurement data to obtain a reference input track;
an updating module for updating an initial greenhouse environment model based on the reference input trajectory;
the construction module is used for constructing a model parameter identification controller based on a Lyapunov function;
the calculation module is used for calculating the optimal solution of each nonlinear time-varying model parameter at the t moment according to the model parameter identification controller and the initial greenhouse environment model and storing the optimal solution into a model parameter set;
the judging module is used for judging whether t is less than the terminal time of the control process; if t is less than the terminal time of the control process, making t equal to t +1, and returning to the acquisition module; if t is greater than or equal to the control process terminal time, outputting the model parameter set;
the identification general model parameter solution determining module is used for determining an identification general model parameter solution according to the optimal solution of the nonlinear time-varying model parameters at different moments in the model parameter set;
the prediction module is used for substituting the identification general model parameter solution into a greenhouse environment model to obtain a greenhouse environment prediction value; the greenhouse environment prediction value comprises the following steps: a predicted value of indoor temperature, a predicted value of indoor humidity, and a predicted value of indoor CO2 concentration.
7. The agricultural greenhouse environment prediction system of claim 6, wherein the greenhouse environment model is specifically formulated as:
Figure FDA0002961672080000051
wherein, CapairRepresents the volumetric heat capacity, theta, of the indoor air below the inner heat-insulating mesh'1,θ′2,θ′3,θ′4,θ′5,θ′6,θ′7All represent the optimal model parameters，TsoilIndicating the surface soil temperature, TairIndicating a predicted value of the indoor temperature, ToutRepresenting the outdoor temperature, pairDenotes the air density, Cp,airDenotes the specific heat capacity of air, phisideventShowing the side window ventilation rate of the greenhouse after considering the heat preservation net in the greenhouse, the air leakage of the greenhouse and the chimney effectThScrShowing the convective air flow, T, between the upper and lower air of a heat-insulating net in the greenhousetopIndicating the temperature, T, of the air above the inner insulationcanDenotes the canopy temperature, IglobRepresenting solar radiation above the cover, lambda representing the latent heat coefficient, ufogIndicating the actual control quantity of spray, #fogIndicating the rated flow, u, of the spraying systemheatRepresents the actual control quantity of direct air heating of the greenhouse system, phiheatRepresenting the nominal heat flow, p, of the heating systemwaterDenotes the density of water, Cp,waterDenotes the specific heat capacity of water, vwaterIndicating the flow rate of hot water pipe, Tpipe,inIndicates the water temperature T at the inlet of the heating pipepipe,outIndicates the water temperature at the outlet of the heating pipe, AgDenotes the area of the greenhouse, HgIndicating the greenhouse height, wairDenotes a predicted value of indoor humidity, E denotes a canopy transpiration rate, McondDenotes the water vapor condensation coefficient of the cover layer, woutDenotes the outdoor humidity, wtopIndicating the humidity, CO, of the air above the heat-insulating net2,airRepresents a predicted value of the indoor CO2 concentration,
Figure FDA0002961672080000061
indicating the actual amount of CO2 control,
Figure FDA0002961672080000062
indicating the photosynthetic rate of the crop, PgIndicating canopy photosynthetic rate, RmIndicating the sustained respiration rate, CO, of the crop2,outIndicating outdoor CO2 concentration, CO2,topIndicating the CO2 concentration of the air above the heat retention screen.
8. The agricultural greenhouse environment prediction system of claim 6, wherein the identifying general model parameter solution determination module specifically comprises:
a first optimal model parameter determining unit, configured to determine an optimal solution according to the nonlinear time-varying model parameters at multiple different time instants in the model parameter set
Figure FDA0002961672080000063
Performing polynomial fitting to obtain a solution theta of the parameters of the identified general model'1The concrete formula is as follows:
θ′1＝p1(Tair-Tsoil)6+p2(Tair-Tsoil)5+p3(Tair-Tsoil)4+p4(Tair-Tsoil)2+p5(Tair-Tsoil)+p6；
wherein p is1,p2,p3,p4,p5,p6All represent polynomial coefficients, TairIndicating a predicted value of the indoor temperature, TsoilRepresenting the surface soil temperature;
a second optimal model parameter determining unit for determining an optimal solution according to the nonlinear time-varying model parameters at different time instants in the model parameter set
Figure FDA0002961672080000064
Carrying out averaging to obtain a solution theta of the parameters of the identification general model'3The concrete formula is as follows:
Figure FDA0002961672080000065
wherein the content of the first and second substances,
Figure FDA0002961672080000066
represents a constant;
a third optimal model parameter determining unit for determining a nonlinear time-varying model according to a plurality of different time instants in the model parameter setOptimum solution of type parameter
Figure FDA0002961672080000067
Determining and identifying theta 'of solution of parameters of general model'4The concrete formula is as follows:
Figure FDA0002961672080000068
wherein, IglobWhich represents the solar radiation above the cover layer,
Figure FDA0002961672080000069
represents a constant;
a fourth optimal model parameter determining unit, configured to determine an optimal solution according to the nonlinear time-varying model parameters at multiple different time instants in the model parameter set
Figure FDA0002961672080000071
Performing polynomial fitting to obtain a solution theta of the parameters of the identified general model'5The concrete formula is as follows:
Figure FDA0002961672080000072
wherein q is0～q9All represent polynomial coefficients, TcanRepresenting canopy temperature, daytime, night;
a fifth optimal model parameter determining unit for solving theta 'according to the identified general model parameter'4Determining and identifying theta 'of solution of parameters of general model'2The concrete formula is as follows:
θ′2＝1.593θ′4+0.008；
a sixth optimal model parameter determining unit, configured to determine an optimal solution according to the nonlinear time-varying model parameters at multiple different time instants in the model parameter set
Figure FDA0002961672080000073
Carrying out averaging to obtain a solution theta of the parameters of the identification general model'6The concrete formula is as follows:
Figure FDA0002961672080000074
wherein, Cday、CnightRespectively represent two different constants;
a seventh optimal model parameter determining unit, configured to determine an optimal solution according to the nonlinear time-varying model parameters at multiple different time instants in the model parameter set
Figure FDA0002961672080000075
Carrying out averaging to obtain a solution theta of the parameters of the identification general model'7The concrete formula is as follows:
Figure FDA0002961672080000076
wherein D isday、DnightTwo different constants are represented.
9. The agricultural greenhouse environment prediction system of claim 6, wherein the model parameter recognition controller is formulated as:
Figure FDA0002961672080000077
wherein the content of the first and second substances,
Figure FDA0002961672080000081
the control law of the model parameters is represented,
Figure FDA0002961672080000082
representing the Lagrange coefficient update law, Γ1,Γ2All indicate givenPositive definite matrix xi1,ξ2Both represent adaptive terms of the parameter recognition controller,
Figure FDA0002961672080000083
μ denotes lagrange coefficients, T denotes transposition,
Figure FDA0002961672080000084
j denotes the error norm and θ denotes the set of nonlinear time-varying model parameters.
10. The agricultural greenhouse environment prediction system of claim 6, wherein the initial greenhouse environment model is formulated as:
Figure FDA0002961672080000085
wherein the content of the first and second substances,
Figure FDA0002961672080000086
an initial greenhouse environment model is represented, and,
Figure FDA0002961672080000087
and
Figure FDA0002961672080000088
all represent known nonlinear functions approximating the actual greenhouse environment state, theta' represents the optimal model parameter variable, xrRepresents the actual greenhouse environment state variable and is provided with a plurality of temperature sensors,
Figure FDA0002961672080000089
Figure FDA00029616720800000810
respectively, the actual measured greenhouse temperature, humidity and carbon dioxide concentration, u represents the control variable of the actuator, and u ═ u [ u ]heat,qpipe,usidevent,uroofvent,uscr]T，uheat,qpipe,usidevent,uroofvent,uscrRespectively representing the actual control quantity of the greenhouse system, such as direct air heating, pipeline hot water circulation heating, side window ventilation, skylight ventilation and an inner heat-preservation net, wherein v represents the outdoor weather variable, and v is [ T ═ T [out,Hout,CO2,air,Vwind,Iglob]T，Tout,Hout,CO2,air,Vwind,IglobRespectively representing the actually measured outdoor temperature, humidity, carbon dioxide concentration, wind speed and solar radiation.
CN202110239759.5A 2021-03-04 2021-03-04 Agricultural greenhouse environment prediction method and system Active CN112947648B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202110239759.5A CN112947648B (en) 2021-03-04 2021-03-04 Agricultural greenhouse environment prediction method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202110239759.5A CN112947648B (en) 2021-03-04 2021-03-04 Agricultural greenhouse environment prediction method and system

Publications (2)

Publication Number Publication Date
CN112947648A true CN112947648A (en) 2021-06-11
CN112947648B CN112947648B (en) 2021-11-23

Family

ID=76247548

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202110239759.5A Active CN112947648B (en) 2021-03-04 2021-03-04 Agricultural greenhouse environment prediction method and system

Country Status (1)

Country Link
CN (1) CN112947648B (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113867451A (en) * 2021-10-29 2021-12-31 江苏大学 Intelligent temperature control method for livestock and poultry breeding by combining cloud platform and controller
CN114153252A (en) * 2021-11-24 2022-03-08 魏育华 Greenhouse ventilation method and system

Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1885054A1 (en) * 2006-08-03 2008-02-06 STMicroelectronics S.r.l. Method of estimating the state of a system and related device for estimating position and speed of the rotor of a brushless motor
CN101276207A (en) * 2008-05-07 2008-10-01 浙江大学 Multivariable non-linear system prediction function control method based on Hammerstein model
US20090319060A1 (en) * 2000-06-20 2009-12-24 Fisher-Rosemount Systems, Inc. Continuously Scheduled Model Parameter Based Adaptive Controller
CN103064282A (en) * 2012-05-04 2013-04-24 浙江大学 Nonlinear parameter variation (NPV) model identification method
CN103235620A (en) * 2013-04-19 2013-08-07 河北农业大学 Greenhouse environment intelligent control method based on global variable prediction model
CN103268069A (en) * 2013-05-22 2013-08-28 常州大学 Adaptive prediction control method based on Hammerstein model
US20160170384A1 (en) * 2014-12-11 2016-06-16 University Of New Brunswick Model predictive controller and method with correction parameter to compensate for time lag
CN105974801A (en) * 2016-08-03 2016-09-28 苏州睿渲恒晟智能科技有限公司 Greenhouse environment modeling and control technology based on data drive
CN107728481A (en) * 2017-11-14 2018-02-23 江西理工大学 A kind of closed loop modeling method and device based on Model Predictive Control
US20180081348A1 (en) * 2016-09-16 2018-03-22 Honeywell Limited Closed-loop model parameter identification techniques for industrial model-based process controllers
US20180348717A1 (en) * 2017-06-02 2018-12-06 Aspen Technology, Inc. Computer System And Method For Building And Deploying Predictive Inferential Models Online
CN109507882A (en) * 2018-11-20 2019-03-22 中南大学 A kind of fast robust forecast Control Algorithm based on RBF-ARX model
CN110045616A (en) * 2019-05-20 2019-07-23 长沙学院 A kind of Robust Predictive Control method of stirred tank

Patent Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20090319060A1 (en) * 2000-06-20 2009-12-24 Fisher-Rosemount Systems, Inc. Continuously Scheduled Model Parameter Based Adaptive Controller
EP1885054A1 (en) * 2006-08-03 2008-02-06 STMicroelectronics S.r.l. Method of estimating the state of a system and related device for estimating position and speed of the rotor of a brushless motor
CN101276207A (en) * 2008-05-07 2008-10-01 浙江大学 Multivariable non-linear system prediction function control method based on Hammerstein model
CN103064282A (en) * 2012-05-04 2013-04-24 浙江大学 Nonlinear parameter variation (NPV) model identification method
CN103235620A (en) * 2013-04-19 2013-08-07 河北农业大学 Greenhouse environment intelligent control method based on global variable prediction model
CN103268069A (en) * 2013-05-22 2013-08-28 常州大学 Adaptive prediction control method based on Hammerstein model
US20160170384A1 (en) * 2014-12-11 2016-06-16 University Of New Brunswick Model predictive controller and method with correction parameter to compensate for time lag
CN105974801A (en) * 2016-08-03 2016-09-28 苏州睿渲恒晟智能科技有限公司 Greenhouse environment modeling and control technology based on data drive
US20180081348A1 (en) * 2016-09-16 2018-03-22 Honeywell Limited Closed-loop model parameter identification techniques for industrial model-based process controllers
US20180348717A1 (en) * 2017-06-02 2018-12-06 Aspen Technology, Inc. Computer System And Method For Building And Deploying Predictive Inferential Models Online
CN107728481A (en) * 2017-11-14 2018-02-23 江西理工大学 A kind of closed loop modeling method and device based on Model Predictive Control
CN109507882A (en) * 2018-11-20 2019-03-22 中南大学 A kind of fast robust forecast Control Algorithm based on RBF-ARX model
CN110045616A (en) * 2019-05-20 2019-07-23 长沙学院 A kind of Robust Predictive Control method of stirred tank

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
DONG LI: ""Hierarchical model predictive control of greenhouse climate to reduce energy cost"", 《2ND INTERNATIONAL CONFERENCE ON INDUSTRIAL ARTIFICIAL INTERLLIGENCE》 *
H.OUBEHAR: ""Design and real time inplementation of ANFIS controller for greenhouse climate"", 《2018 INTERNATIONAL CONFERENCE ON ELECTRONICS,CONTROL,OPTIMIZATION AND COMPUTER SCIENCE》 *
徐立鸿: ""面向控制的温室系统小气候环境模型要求与现状"", 《农业工程学报》 *
马娇: ""试验温室温度混杂系统的建模与预测控制"", 《中国优秀硕士学位论文全文数据库》 *

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113867451A (en) * 2021-10-29 2021-12-31 江苏大学 Intelligent temperature control method for livestock and poultry breeding by combining cloud platform and controller
CN113867451B (en) * 2021-10-29 2022-08-23 江苏大学 Intelligent temperature control method for livestock and poultry breeding by combining cloud platform and controller
CN114153252A (en) * 2021-11-24 2022-03-08 魏育华 Greenhouse ventilation method and system

Also Published As

Publication number Publication date
CN112947648B (en) 2021-11-23

Similar Documents

Publication Publication Date Title
Van Beveren et al. Optimal control of greenhouse climate using minimal energy and grower defined bounds
Van Henten Greenhouse climate management: an optimal control approach
Van Beveren et al. Minimal heating and cooling in a modern rose greenhouse
del Sagrado et al. Bayesian networks for greenhouse temperature control
CN112947648B (en) Agricultural greenhouse environment prediction method and system
CN107037728A (en) Greenhouse optimal control method based on multiple objective gray particle cluster algorithm
González Pérez et al. Neural networks-based models for greenhouse climate control
Bailey et al. Improving the cost effectiveness of greenhouse climate control
CN108762063A (en) A kind of heliogreenhouse wet curtain fan circulating ventilation cooling system and its control method
Labidi et al. Control of relative humidity inside an agricultural greenhouse
Su et al. A greenhouse climate model for control design
JP6573849B2 (en) Agricultural house environment controller
Van Straten Acceptance of optimal operation and control methods for greenhouse cultivation
CN110073857A (en) A kind of greenhouse facade ventilating and thermal insulating global anti-wind system and control method
CN115796040A (en) Facility greenhouse temperature prediction method based on small samples
Bontsema et al. On-line estimation of the transpiration in greenhouse horticulture
Berenguel et al. Greenhouse diurnal temperature control with natural ventilation based on empirical models
Nazir et al. Design of an automated greenhouse temperature controller; a renewable, robust and cost effective scheme
Li et al. Modeling of natural ventilation using a hierarchical fuzzy control system for a new energy-saving solar greenhouse
Bot A validated physical model of greenhouse climate
LAMMARİ et al. Constrained GA PI sliding mode control of indoor climate coupled MIMO greenhouse model
CN115907366B (en) Agricultural product growth environment optimal regulation and control method and equipment based on flamingo algorithm
Baier et al. Using model predictive control to modulate the humidity in a broiler house and effect on energy consumption
Pérez-Parra et al. Ventilation rate models of mediterranean greenhouses for control purposes
CN118013999A (en) Solar greenhouse building environmental factor adjusting method based on entropy production analysis

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