PREDICTIVE CONTROL SYSTEM AND REGULATORY METHOD FOR TEMPERATURE OF LIVESTOCK HOUSE

Abstract

A predictive control system and regulatory method for a temperature of a livestock house are provided. The predictive control system includes a temperature and humidity sensor, a breeding environment temperature dynamic requirement module, an environmental controller and an environmental regulation implementation mechanism, where the environmental controller is connected to the breeding environment temperature dynamic requirement module and the temperature and humidity sensor; the environmental regulation implementation mechanism is connected to the environmental controller and is configured to perform corresponding environmental regulation according to a command from the environmental controller. The regulatory method controls a temperature of a breeding environment based on a livestock breeding environment temperature dynamic setting model, a livestock breeding environment temperature prediction system based on a GM (1,1) model, and a grey predictive fuzzy system based on the GM (1,1) model.

Claims

1. A predictive control system for a temperature of a livestock house, comprising a temperature and humidity sensor, a breeding environment temperature dynamic requirement module, an environmental controller, and an environmental regulation implementation mechanism, wherein the environmental controller is connected to the breeding environment temperature dynamic requirement module and the temperature and humidity sensor; and the environmental regulation implementation mechanism is connected to the environmental controller and is configured to perform corresponding environmental regulation according to a command from the environmental controller.

2. The predictive control system according to claim 1, wherein the environmental controller comprises a data acquisition module, an environmental temperature prediction module, a predictive fuzzy control module, a predictive fuzzy decision-making module, and an early warning and overhauling module.

3. The predictive control system according to claim 2, wherein the environmental temperature prediction module is connected to the data acquisition module, and is configured to output a predicted temperature of a breeding environment in a livestock house at time N based on a comprehensive grey model (GM) (1,1) integrated with accumulated generating operation (AGO) of a sequence, residual model correction, and equal-dimension new information processing; the predictive fuzzy control module is connected to the environmental temperature prediction module, wherein input signals to the predictive fuzzy control module are a prediction error for the temperature of the breeding environment in the livestock house and a changing rate of the prediction error; a fuzzy decision-making rule is established to determine an output quantity for predictive fuzzy control, and the environmental regulation implementation mechanism is configured to perform control and regulation based on the output quantity for predictive fuzzy control; the predictive fuzzy decision-making module is connected to the predictive fuzzy control module, wherein input signals to the predictive fuzzy decision-making module are an error of the predicted temperature and the output quantity and a changing rate of the error, and an appropriate prediction step is determined; and the early warning and overhauling module is configured to give an early warning based on a number of deviation accumulations, and is connected to the breeding environment temperature dynamic requirement module, the temperature and humidity sensor, and the data acquisition module.

4. The predictive control system according to claim 2, wherein the breeding environment temperature dynamic requirement module is configured to output breeding environment temperature requirement parameters of the livestock at different growth stages based on parameters of a livestock breeding characteristic, a behavioral characteristic, a stress mechanism, a livestock breed, a quality, and an age in days, and is connected to the data acquisition module.

5. The predictive control system according to claim 1, wherein the environmental regulation implementation mechanism comprises a fan, a wet curtain, a small ventilation window, a heating apparatus, and a spraying device, and is configured to control ventilation, cooling, heating, and humidification of the breeding environment of the livestock according to commands from the environmental controller.

6. The predictive control system according to claim 1, wherein a number of temperature and humidity sensors is more than or equal to two, which are disposed inside and outside the livestock house, respectively.

7. A regulatory method for the predictive control system according to claim 2, wherein the regulatory method comprises controlling a temperature of a breeding environment based on a livestock breeding environment temperature dynamic setting model, a livestock breeding environment temperature prediction system based on a GM (1,1) model, and a grey predictive fuzzy system based on the GM (1,1) model, and comprises following steps: S1: determining a required temperature T.sub.x of an environment in a livestock house at time N based on the breeding temperature dynamic requirement model of the livestock, and dynamically setting, by the livestock breeding environment temperature dynamic setting model, a temperature parameter based on a model of parameters of a livestock breeding characteristic, a behavioral characteristic, a stress mechanism, a livestock breed, a quality, and an age in days; S2: predicting a breeding environment temperature of the livestock house based on the GM (1,1) model and acquiring a predicted temperature T.sub.y of the environment in the livestock house at time N, wherein a system and method for predicting an environment in a livestock house by the model based on the comprehensive GM (1,1) model integrated with accumulated generating operation (AGO) of a sequence, residual model correction, and equal-dimension new information processing comprise the following steps: firstly, accumulating historical temperature data from a data acquisition module to generate an accumulated sequence, establishing the GM (1,1) model, and solving the model to obtain a time response function; secondly, obtaining a predicted value using the GM (1,1) model for residual test, and in combination with a changing characteristic of a residual sequence and an advantage of a residual GM (1,1) model, establishing the residual GM (1,1) model for a sequence failing to pass a model test or having low accuracy to correct the original model so as to improve accuracy of the model; thirdly, selecting a number of dimensions of a prediction model by using residual or accuracy test of the model; and finally, performing equal-dimension new information processing to establish an equal-dimension new information predicting GM (1,1) model, and repeating the test and correction of the model; S3: determining whether the required temperature T.sub.x of the breeding environment of the livestock is equal to the predicted temperature T.sub.y; if T.sub.x=T.sub.y, maintaining an original regulatory strategy of the environmental regulation implementation mechanism; and if T.sub.x T.sub.y, regulating the environmental regulation implementation mechanism by the grey predictive fuzzy control system based on the GM (1,1) model, and proceeding to S4; S4: setting a prediction error output by a livestock breeding environment prediction module and a changing rate of the prediction error as input quantities to the predictive fuzzy control module, and setting an error of input and output and a changing rate of the error as input signals to the predictive fuzzy decision-making module, determining an appropriate prediction step, establishing a fuzzy decision-making rule, establishing a grey predictive fuzzy control model based on the GM (1,1) model to determine a final output value for fuzzy control, and controlling an environmental regulatory device based on an output quantity for predictive fuzzy control; when T.sub.y<T.sub.x, the system automatically entering a heating mode; when T.sub.y>T.sub.x, the system automatically entering a wet curtain cooling mode; and causing the temperature of the livestock house to meet a livestock temperature requirement T.sub.x=T.sub.y, and repeating steps S1 and S2; and S5: determining whether the required temperature T.sub.x is equal to a measured temperature T.sub.c; if |T.sub.xT.sub.c|>0.5 C., the system automatically entering an accumulation mode; and when a number of accumulations is more than 5, the system performing early warning, causing a breeding manager to overhaul the temperature and humidity sensor and the environmental regulation implementation mechanism according to the early warning.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0036] FIG. 1 is a structural schematic diagram of a predictive control system for a temperature of a livestock house;

[0037] FIG. 2 is a flowchart of a method for predicting a temperature of a livestock house based on a GM (1,1);

[0038] FIG. 3 is a schematic diagram of a predictive control system for a temperature of a livestock house based on a GM (1,1);

[0039] FIG. 4 is a diagram illustrating comparison curves of a measured value and a predicted value of a temperature of a livestock house;

[0040] FIG. 5 is a percentage map of a difference between a measured value and a predicted value of a temperature of a livestock house; and

[0041] FIG. 6 is a diagram illustrating a variation curve of a temperature of a livestock house under predictive control.

DETAILED DESCRIPTION OF THE EMBODIMENTS

[0042] To make the objectives, technical solutions, and advantages of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be further described in detail below with reference to the accompanying drawings in the embodiments of the present disclosure. Reference numerals which are the same or similar throughout the accompanying drawings represent the same or similar elements or elements with the same or similar functions. The described embodiments are some rather than all of the embodiments of the present disclosure. The embodiments described below with reference to the drawings are illustrative for explaining the present disclosure and are not to be construed as limiting the present disclosure. All other embodiments derived from the embodiments of the present disclosure by a person of ordinary skill in the art without creative efforts shall fall within the protection scope of the present disclosure.

[0043] As shown in FIG. 1 to FIG. 3, the present disclosure provides a predictive control system and method for an environment in a livestock house based on a grey model GM (1,N), mainly including the following steps S1 to S5.

[0044] In step S1: a required temperature of an environment in a livestock house at time N is determined based on a livestock breeding environment temperature dynamic requirement model. High-yield breeds are mostly adopted in modern large-scale high-density livestock breeding production. The livestock has high requirements on the breeding environment and low adaptability to stresses such as temperature fluctuations of the breeding environment. The breeding environment temperatures at different growth stages of the livestock meeting requirements is the basis of guaranteeing the genetic potential and production efficiency of the livestock of high-yield breeds. In the present system, a model of parameters such as a livestock breeding characteristic, a behavioral characteristic, a stress mechanism, a livestock breed, a quality, and an age in days, to dynamically set the requirements T.sub.x for the environment temperature at different growth stages of the livestock, and control the temperature of the livestock house to be an ideal temperature T.sub.x beneficial for the livestock and making the livestock grow healthily.

[0045] In step S2: a predicted temperature T.sub.y of the environment in the livestock house at time N is predicted by a livestock temperature prediction system based on the GM (1,1).

[0046] In step (1): firstly, weather forecast values for a region where the livestock house is located from a data acquisition module of an environmental controller or environmental temperatures inside and outside the livestock house collected by temperature and humidity sensors inside and outside the livestock house are obtained as historical temperature data.

[0047] The weather forecast values for the region where the livestock house is located from the data acquisition module of the environmental controller or the environmental temperatures inside and outside the livestock house collected by the temperature and humidity sensors inside and outside the livestock house are historical data which are set as a data source. A time sequence is constructed with data at a starting point of time and first n data with equal time intervals as an initial arithmetic data set.

[0048] The weather forecast values for the region where the livestock house is located or the environmental temperatures inside and outside the livestock house collected by the temperature and humidity sensors inside and outside the livestock house are sampled at equal time intervals, and an original temperature sequence T.sup.(0) is as follows: [0049] T.sup.(0)=(T.sup.(0) (1), T.sup.(0) (2), T.sup.(0) (3) . . . T.sup.(0) (n)); [0050] In step (2): the historical temperature data is accumulated to generate an accumulated sequence.

[0051] The historical temperature data is accumulated to generate the accumulated sequence. Accumulated generating operation (AGO) is adopted to weaken the fluctuation and randomness of a random sequence of thermal environmental parameter, thus improving the prediction accuracy of the model.

[0052] The first order-accumulated generating operator (1-AGO) is performed on the temperature sequence to weaken the influence of random interferences and obtain a new data sequence T.sup.(1): [0053] T.sup.(1)=(T.sup.(1)(1), T.sup.(1)(2), T.sup.(1)(3) . . . T.sup.(1)(n)), [0054] where, T.sup.(1)(k)=.sub.i=1.sup.kT.sup.(0)(i), k=1, 2, 3 . . . m.

[0055] In step (3): the GM (1,1) is established and solved to obtain a time response function. A whitening differential equation of the GM (1,1) is established with transformed data, and the model is solved to obtain the time response function.

[0056] For T.sup.(1), a first-order variable differential equation is established to form the temperature prediction grey model GM (1,1),

[00001]$\frac{{\mathrm{dT}}^{\left(1\right)}}{dt}+a{T}^{\left(1\right)}=u,$

where and u are parameters to be solved. Letting {circumflex over ()}=[a, u].sup.T, the parameters and u are solved by a method of least squares, {circumflex over ()}=, (B.sup.TB).sup.1B.sup.TY, where B and Y are respectively as follows:

[00002]$B=\left[\begin{array}{cc}-\frac{1}{2}\left(T^{\left(1\right)}\left(1\right)+T^{\left(1\right)}\left(2\right)\right)& 1\\ -\frac{1}{2}\left(T^{\left(1\right)}\left(2\right)+T^{\left(1\right)}\left(3\right)\right)& 1\\ .Math.& .Math.\\ -\frac{1}{2}\left(T^{\left(1\right)}\left(n-1\right)+T^{\left(1\right)}\left(n\right)\right)& 1\end{array}\right]$$Y={\left[T^{\left(0\right)}\left(2\right),T^{\left(0\right)}\left(3\right),T^{\left(0\right)}\left(4\right),.Math.,T^{\left(0\right)}\left(m\right)\right]}^T$

[0057] Assuming T.sup.(1)(0)=T.sup.(0)(1), when the variation of the sequence T.sup.(1) is smooth, the time response sequence of the GM (1,1) model is:

[00003]${\hat{T}}^{\left(1\right)}\left(k+1\right)=\left(T^{\left(0\right)}\left(1\right)-\frac{u}{a}\right)e^{-\mathrm{ak}}+\frac{u}{a}k=1,2,3.Math..Math.m-1,{\hat{T}}^{\left(0\right)}\left(k+1\right)={\hat{T}}^{\left(1\right)}\left(k+1\right)-{\hat{T}}^{\left(1\right)}\left(k\right),$

and the following data sequence is obtained: [0058] {circumflex over (T)}.sup.(1)=({circumflex over (T)}.sup.(1)(1), {circumflex over (T)}.sup.(1)(1), {circumflex over (T)}.sup.(1)(1), . . . , T.sup.(1)(n));

[0059] In step (4): a predicted value is obtained using the GM (1,1) for residual test. Sequence residual test and restoration sequence test are performed on a predicted temperature sequence. If the predicted value is within an accuracy range, when a simulated relative error and an average relative error are less than 1%, a mean square error ratio is less than 0.35, and a small error probability is greater than 0.95, it is considered that a livestock house temperature prediction model meets the accuracy requirement, and the predicted temperature sequence is output.

[0060] An original temperature data sequence of the livestock house is denoted by T.sup.(0)(k) and the predicted temperature sequence is denoted by {circumflex over (T)}.sup.(1)(k). A difference between T.sup.(0)(k) and {circumflex over (T)}.sup.(1) (k) is a residual sequence .sup.(0)(n), .sup.(0)(n)=((1), (2)(3), . . . (n))=(T.sup.(0)(1)-{circumflex over (T)}.sup.(1) (1)T.sup.(0)(2)-{circumflex over (T)}.sup.(1)(2)T.sup.(0)(3)-T.sup.(1)(3), . . . , T.sup.(0)(n)-(n)).

[0061] A relative error sequence is as follows:

[00004]$\left(n\right)=\left(.Math.\frac{\left(1\right)}{T^{\left(0\right)}\left(1\right)}.Math.,.Math.\frac{\left(2\right)}{T^{\left(0\right)}\left(2\right)}.Math.,.Math.\frac{\left(3\right)}{T^{\left(0\right)}\left(3\right)}.Math.,.Math.,.Math.\frac{\left(n\right)}{T^{\left(0\right)}\left(n\right)}.Math.\right)={\left\{\right\}}_1^n.$

[0062] When kn, the simulated relative error of point k is:

[00005]${}_k=.Math.\frac{\left(k\right)}{x^{\left(0\right)}\left(k\right)}.Math.,$

and the average relative error of point k is:

[00006]$\stackrel{\_}{{}_k}=\frac{1}{n}{}_{k=1}^n{}_k.$

[0063] A mean x.sup.(0) of the original data X.sup.(0) is:

[00007]${\stackrel{}{x}}^{\left(0\right)}=\frac{1}{n}{}_{k=1}^n{x}^{\left(0\right)}\left(k\right),$

and a variance S.sub.1.sup.2 of the original data X.sup.(0) is:

[00008]$S_1^2=\frac{1}{n-1}{}_{k=1}^n{\left(x^{\left(0\right)}\left(k\right)-{\stackrel{}{x}}^{\left(0\right)}\right)}^2.$

[0064] A mean .sup.(0) of the residual sequence .sup.(0)(n) is:

[00009]${\stackrel{}{}}^{\left(0\right)}=\frac{1}{n}{}_{k=1}^n{}^{\left(0\right)}\left(k\right);$

and a variance S.sub.2.sup.2 of the residual sequence .sup.(0)(n) is:

[00010]$S_2^2=\frac{1}{n-1}{}_{k=1}^n{\left({}^{\left(0\right)}\left(k\right)-{\stackrel{\_}{}}^{\left(0\right)}\right)}^2.$

[0065] A variance ratio c of posterior-variance-test is:

[00011]$c=\frac{S_2}{S_1}.$

[0066] A small probability error value p is: p=p {|.sup.(0)(k).sup.(0))|<0.6745S.sub.1}.

[0067] An accuracy test is performed according to the calculated values of 4 indicators .sub.k, .sub.k, c, and p.

[0068] In step (5): in combination with a changing characteristic of the residual sequence and an advantage of the residual GM (1,1), the residual GM (1,1) is established for a sequence failing to pass a model test or having low accuracy to correct the original model, so as to improve the accuracy of the model. If failing to pass the model test, a residual correction model is established. The residual sequence of the sequence having low accuracy is chosen and ranked. 1-AGO processing is performed on the residual sequence of the sequence having low accuracy to generate an accumulated sequence. A residual GM (1,N) is established and solved to obtain a time response sequence. A new residual model is obtained and superposed into the correction of the original temperature prediction model. The original model is corrected to obtain a new temperature prediction model. The prediction accuracy of the model is tested. If the accuracy requirement is met, the corrected temperature sequence is output; otherwise, correction continues until the predicted temperature sequence meets the test requirement.

[0069] When the accuracy of the established livestock house temperature prediction model GM (1,1) does not meet the environmental control requirement of the livestock house, the original model needs to be corrected. The residual sequence is utilized for modeling to improve the accuracy of the prediction model GM (1,1).

[0070] The residual sequence is: .sup.(0)(n)=T.sup.(0)(n)-{circumflex over (T)}.sup.(0)(n). The residual sequence of the sequence with low accuracy is chosen and ranked, and 1-AGO processing is performed to obtain sequence .sup.(1)(n). A first-order variable differential equation is established for .sup.(1) to form the residual prediction GM (1,1),

[00012]$\frac{d{}^{\left(1\right)}}{dt}+a^{}{T}^{\left(1\right)}=u^{},$

where and u are parameters to be solved. Letting {circumflex over ()}=[a, u].sup.T, the parameters and u are solved by the method of least squares, {circumflex over ()}=(B.sup.TB).sup.1 B.sup.TY, where B and Y are respectively as follows:

[00013]$B=\left[\begin{array}{cc}-\frac{1}{2}\left({}^{\left(1\right)}\left(1\right)+{}^{\left(1\right)}\left(2\right)\right)& 1\\ -\frac{1}{2}\left({}^{\left(1\right)}\left(2\right)+{}^{\left(1\right)}\left(3\right)\right)& 1\\ .Math.& .Math.\\ -\frac{1}{2}\left({}^{\left(1\right)}\left(n-1\right)+{}^{\left(1\right)}\left(n\right)\right)& 1\end{array}\right]Y={\left[{}^{\left(0\right)}\left(2\right),{}^{\left(0\right)}\left(3\right),{}^{\left(0\right)}\left(4\right),.Math..Math.,{}^{\left(0\right)}\left(m\right)\right]}^T$

[0071] The time response function of the residual model is

[00014]${\hat{}}^{\left(0\right)}\left(k\right)=\left(-a^{}\right)\left({}^{\left(0\right)}\left(1\right)-\frac{u^{}}{a^{}}\right)e^{-a^{}k}.$

[0072] For ease of expression, the time response function is rewritten as

[00015]${\hat{}}^{\left(0\right)}\left(k+1\right)=\left(-a^{}\right)\left({}^{\left(0\right)}\left(1\right)-\frac{u^{}}{a^{}}\right)e^{-a^{}k}.$

[0073] The reciprocal of the time response function {circumflex over (T)}.sup.(1)(k+1) of the livestock house temperature prediction model is corrected. When

[00016]${\hat{T}}^{\left(1\right)}\left(k+1\right)=\left(T^{\left(0\right)}\left(1\right)-\frac{u}{a}\right)e^{-\mathrm{ak}}+\frac{u}{a},k=1,2,3.Math..Math.m-1,T^{\left(0\right)}\left(k+1\right)=\left(-a\right)\left(T^{\left(0\right)}\left(1\right)-\frac{u}{a}\right)e^{-\mathrm{ak}}.$

[0074] The residual GM (1,N) model, {circumflex over ()}.sup.(0)(k+1), is added to the original prediction model for correction to derive

[00017]${\hat{x}}^{\left(0\right)}\left(k+1\right)=\left(-a\right)\left(T^{\left(0\right)}\left(1\right)-\frac{u}{a}\right)e^{-\mathrm{ak}}+\left(k-i\right)\left(-a^{}\right)\left({}^{\left(0\right)}\left(1\right)-\frac{u^{}}{a^{}}\right)e^{-a^{}k},$$\mathrm{where}\left(k-i\right)=\{\begin{array}{cc}1,& ki\\ 0,& k<i\end{array}.$

[0075] In step (6): the number of dimensions of the prediction model is selected by the residual or accuracy test of the model. The number of dimensions m are selected by the residual or accuracy test of the model, and a smaller average absolute error is preferred. The smaller average absolute error is

[00018]$\mathrm{MAPE}\left(\%\right)=\frac{1}{n}{.Math.}_{k=1}^n.Math.\frac{{\hat{x}}_1^{\left(0\right)}\left(k\right)-{\hat{x}}_1^{\left(0\right)}\left(k\right)}{{\hat{x}}_1^{\left(0\right)}\left(k\right)}.Math..$

[0076] When the average absolute error is minimum, namely

[00019]$\underset{1mn}{\min }\mathrm{MAPE}\left(\%\right)=\frac{1}{n}{.Math.}_{k=1}^n.Math.\frac{{\hat{x}}_1^{\left(0\right)}\left(k\right)-{\hat{x}}_1^{\left(0\right)}\left(k\right)}{{\hat{x}}_1^{\left(0\right)}\left(k\right)}.Math.,$

the value of m is the number of dimensions of the prediction model.

[0077] In step (7): equal-dimension new information processing is performed to establish an equal-dimension new information predicting GM (1,1) model, and testing and correction of the model are repeated. With increasing of the livestock house temperature sequence, the predicted temperatures obtained by the temperature prediction model in last step are ranked according to the time sequence. The arithmetic data set is updated. Equal-dimension processing is performed on the sequence to obtain an equal-dimension new information sequence, and the equal-dimension new information GM (1,1) model is established. The steps (3), (4), and (5) are repeated.

[0078] In the livestock house temperature prediction GM (1,1) model, equal-dimension new information processing is performed on T.sup.(0)=(T.sup.(0)(1), T.sup.(0)(2), T.sup.(0)(3), . . . , T.sup.(0)(n1), T.sup.(0)(n)). T.sup.(0)(1) is removed and T.sup.(0)(n+1) is added to obtain: T.sup.(0)=(T.sup.(0)(2), T.sup.(0)(3), . . . , T.sup.(0)(n1), T.sup.(0)(n), T.sup.(0)(n+1)).

[0079] An output sequence at time t.sub.i is T.sub.i, T.sub.i=(T.sub.i(1), T.sub.i(2), T.sub.i(3), . . . , T.sub.i(n)), and an output sequence at time t.sub.i+1 is T.sub.i+1, T.sub.i+1=(T.sub.i+1 (1), T.sub.i+1 (2), T.sub.i+1 (3), . . . , T.sub.i+1 (n)); and it is maintained that T.sub.i+1 (k)=(k+1) and T.sub.i+1 (n1)=T.sub.i(n).

[0080] 1-AGO is performed on the data sequence after the equal-dimension new information processing to generate a new sequence. The first-order variable differential equation is established to form the temperature prediction GM (1,1),

[00020]$\frac{{\mathrm{dT}}^{\left(1\right)}}{\mathrm{dt}}+a_{i+1}{T}^{\left(1\right)}=u_{i+1},$

where .sub.i+1 and u.sub.i+1 are parameters to be solved. Letting =[.sub.i+1, u.sub.i+1].sup.T, the parameters .sub.i+1 and u.sub.i+1 are solved by the method of least squares, =(B.sub.i+1.sup.TB.sub.i+1).sup.1 B.sub.i+1.sup.TY.sub.i+1,N, where B.sub.i+1 and Y.sub.i+1,N are respectively as follows:

[00021]$\left[\begin{array}{c}{a}_{i+1}\\ u_{i+1}\end{array}\right]={\left(B_{i+1}^T{B}_{i+1}\right)}^{-1}{B}_{i+1}^T{Y}_{i+1,N}$

[00022]$B_{i+1}=\left[\begin{array}{cc}-\frac{1}{2}\left(T_i^{\left(1\right)}\left(2\right)+T_i^{\left(1\right)}\left(3\right)\right)& 1\\ -\frac{1}{2}\left(T_i^{\left(1\right)}\left(3\right)+T_i^{\left(1\right)}\left(4\right)\right)& 1\\ .Math.& .Math.\\ -\frac{1}{2}\left(T_i^{\left(1\right)}\left(n\right)+T_{i+1}^{\left(1\right)}\left(n\right)\right)& 1\end{array}\right]Y_{i+1,N}=\left[\begin{array}{cc}{T}_i^{\left(0\right)}& \left(3\right)\\ T_i^{\left(0\right)}& \left(4\right)\\ .Math.& .Math.\\ T_{i+1}^{\left(0\right)}& \left(n\right)\end{array}\right]$

[0081] The corresponding time response model is:

[00023]${\hat{T}}_{i+1}^{\left(1\right)}\left(k+1\right)=\left({\hat{T}}_{i+1}^{\left(1\right)}\left(0\right)-\frac{u_{i+1}}{a_{i+1}}\right)e^{-a_{i}k}+\frac{u_{i+1}}{a_{i+1}},$

k=1, 2, 3 . . . m1. The predicted output value of step m of the system is

[00024]${\hat{T}}^{\left(1\right)}\left(k+m\right)=\left(T^{\left(0\right)}\left(k-n+1\right)-\frac{u}{a}\right)e^{-a\left(k-n+1\right)}+\frac{u}{a},k=1,2,3.Math..Math.m-1.$

[0082] In step (8): step (7) is repeated. By a recursive method, the predicted temperature value at time m is output in sequence. The equation {circumflex over (T)}.sup.(0)(k+1)={circumflex over (T)}.sup.(1)(k+1){circumflex over (T)}.sup.(1) (k) is utilized to obtain the predicted value of the original sequence:

[00025]${\hat{T}}^{\left(0\right)}\left(k+1\right)={\hat{T}}^{\left(1\right)}\left(k+1\right)-{\hat{T}}^{\left(1\right)}\left(k\right)=\left[\left(T^{\left(0\right)}\left(1\right)-\frac{\hat{u}}{\hat{a}}\right)e^{-\hat{a}k}\left(1-e^{-\hat{a}}\right)\right].$

[0083] In step S3: whether the required temperature T.sub.x of the environment in the livestock house is equal to the predicted temperature T.sub.y is determined; if T.sub.x=T.sub.y, an original regulatory strategy of an environmental regulation implementation mechanism is maintained; and if T.sub.x T.sub.y, the environmental regulation implementation mechanism is regulated by the grey predictive fuzzy control system based on a GM (1,1) model.

[0084] S4: the grey predictive fuzzy control system based on a GM (1,1) model obtains a control quantity from the predicted value output from the grey model by fuzzy reasoning. A prediction error and a changing rate of the prediction error are taken as input variables to the grey predictive fuzzy control system. The prediction error (t.sub.i+m) and the changing rate (t.sub.i+m) of the prediction error are (t.sub.i+m)=T.sub.set(t.sub.i+m){circumflex over (T)}.sup.(0)(t.sub.i+m) and (t.sub.i+m)=(t.sub.i+m)(t.sub.i+m 1), respectively. An advanced control quantity of the system is determined to perform advanced control on the controlled object. The prediction error (t.sub.i+m), the changing rate (t.sub.i+m) of the prediction error, and the control quantity U(t.sub.i) are normalized to a basic domain with a scaling factor, and a corresponding fuzzy subset is defined. A fuzzy decision-making rule is established, and Mamdani reasoning method is adopted for a control rule to determine the adaptability of the rule. A weighted average defuzzifying algorithm is utilized. The membership functions of the input and output of a fuzzy controller are both triangular. The calculation is simple and less space is occupied. Adjacent fuzzy numbers cross at a membership angle equal to , and there are two rules at most at a certain time to determine the output of the controller. Quantization levels of the prediction error, the error variation, and the control quantity are 7 levels. An appropriate prediction step is determined. A final output value of fuzzy control is determined. The operation and switching of corresponding ventilation modes, and devices such as a wet curtain cooling system, a heating device, a small window in a side wall are controlled such that the temperature T.sub.x=T.sub.y=T.sub.c of the livestock house meets the requirement.

[0085] S5: whether the required temperature T.sub.x is equal to a measured temperature T.sub.c is determined; if |T.sub.xT.sub.c|>0.5 C., the system automatically enters an accumulation mode; and when the number of accumulations is more than 5, the system performs early warning, causing a breeding manager to overhaul a sensor and an environmental regulation implementation apparatus according to the early warning.

EXAMPLE

[0086] A predictive control example for a temperature of a livestock house in a farm in Rizhao of Shandong is selected for analysis. Predicted values and errors of the temperature of the livestock house under different dimensions are as shown in Table 1. In the study of this example, the number of dimensions of the grey predictive model for the temperature is selected to be 7, and the whitening differential equation of the time response sequence of the GM (1,1) is obtained as follows:

[0087] {circumflex over (T)}.sup.(1)(k+1)=(x).sup.(0)(1)1913.5)e.sup.0.013k+1913.5 k=1, 2, 3 . . . m1. The predicted value of the temperature sequence of the chicken house may be:

[00026]${\hat{T}}^{\left(0\right)}\left(k+1\right)={\hat{T}}^{\left(1\right)}\left(k+1\right)-{\hat{T}}^{\left(1\right)}\left(k\right)=\left[\left(T^{\left(0\right)}\left(1\right)-\frac{\hat{u}}{\hat{a}}\right)e^{-\hat{a}k}\left(1-e^{-\hat{a}}\right)\right].$

[0088] The environmental control level and the regulatory strategy parameters of the experimental house in this example are as shown in Table 1. The predicted value {circumflex over (x)}.sup.(0)(k+1) is utilized in system control decision-making. The variations of the measured values and the predicted values of the temperature at measurement points of the experimental house are as shown in FIG. 4. As can be seen from the figure, the maximum difference between the measured value and the predicted value of the temperature in the chicken house is 0.5 C., and there is no significant difference between the measured value and the predicted value of the temperature of the chicken house (P>0.05). The percentages of the temperature differences between the measured values and the predicted values are as shown in FIG. 5. The percentages of the differences between the measured values and the predicted values of the temperature ranges from 0 to 1.9%. The variation trends of the predicted values and the measured values of the temperature are consistent, and the predicted values of the temperature may well express the variation trend of the temperature of the chicken house. As shown in FIG. 6, the environment of the chicken house is regulated using the grey predictive control strategy for the temperature. The maximum and minimum temperature differences between different measurement positions are 1 C. and 0 C., respectively. Under the predictive control, the system shock and overshoot are weakened.

TABLE-US-00001 TABLE 1 Experimental Environmental Control Level and Strategy Parameters Predicted Temperature ( C.) Level Air Volume (* 10.sup.3 m.sup.3/h) 19.5 0.5 D1 70 20.1 0.6 D2 105 20.8 0.7 D3 140 21.6 0.8 D4 175 22.1 1.0 D5 210 24.1 1.5 D6 280 25.6 1.5 D7 350 27.1 1.5 D8 420 29.1 2.0 D9 560 31.4 2.3 D10 700 33.8 2.4 D11 910 36.8 3.0 D12 1015

[0089] Finally, it needs to be noted that the foregoing embodiments are only used to explain the technical solutions of the present disclosure, and are not intended to limit the same. Although the present disclosure is described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced. Such modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the corresponding technical solutions of the embodiments of the present disclosure.