CN106322656B - A kind of air conditioning control method and server and air-conditioning system - Google Patents
A kind of air conditioning control method and server and air-conditioning system Download PDFInfo
- Publication number
- CN106322656B CN106322656B CN201610709749.2A CN201610709749A CN106322656B CN 106322656 B CN106322656 B CN 106322656B CN 201610709749 A CN201610709749 A CN 201610709749A CN 106322656 B CN106322656 B CN 106322656B
- Authority
- CN
- China
- Prior art keywords
- parameter
- input
- output
- air conditioner
- training sample
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Air Conditioning Control Device (AREA)
Abstract
The embodiment of the present invention provides a kind of air conditioning control method and server and air-conditioning system, is related to air-conditioning technical field, can be realized and control without operating status of the user to air-conditioning, realize the intelligence of airconditioning control.The air conditioning control method, comprising: determine the input and output parameter of training sample；Obtain the mathematical model of the neural network of the input parameter and output parameter mapping relations that indicate training sample；The training sample of test is obtained in predefined conditions；The training sample of test is substituted into mathematical model and carries out neural metwork training acquisition nerve network control system；Nerve network control system is configured to air conditioner, real-time output parameter is calculated so that air conditioner acquires input parameter input nerve network control system in real time, the operation of air-conditioning is controlled by real-time output parameter.The embodiment of the present invention is used for airconditioning control.
Description
Technical field
The embodiment of the present invention is related to air-conditioning technical field more particularly to a kind of air conditioning control method and server and air-conditioning
System.
Background technique
Currently, the air conditioner intelligent epoch already arrive, user can be using APP control, voice control, limb action control
Equal intelligentized control methods mode has begun to use, these Intelligentized control methods are mainly by user from traditional remote control control
It frees.For example, summer user can after work by the APP about airconditioning control on mobile phone terminal in family
The operating parameter of air-conditioning carry out remote network control, realize net to the preheating, pre-cooling and humidity of home environment adjustment air
Change etc., or the control by sound or the limb action realization of acquisition user to air conditioner operation parameters at home；But it is above-mentioned
It is still the operating parameter setting for needing user to be actively engaged in air-conditioning in control process, and user can not accurate judgement sky
The operating parameter of tune whether be it is current best, i.e., be still based on what the control of air-conditioning was made in the judgement of user in the prior art
Decision, therefore can not achieve real intelligence.
Summary of the invention
The embodiment of the present invention provides a kind of air conditioning control method and server and air-conditioning system, can be realized without user
The operating status of air-conditioning is controlled, the intelligence of airconditioning control is realized.
In a first aspect, providing a kind of air conditioning control method, comprising:
Determine the input and output parameter of training sample；
Obtain the mathematical model of the neural network of the input parameter and output parameter mapping relations that indicate the training sample；
The training sample of test is obtained in predefined conditions；
The training sample of the test is substituted into the mathematical model and carries out neural metwork training acquisition ANN Control
System；
The nerve network control system is configured to air conditioner, so that input parameter is defeated in real time for air conditioner acquisition
Enter the nerve network control system and calculate real-time output parameter, the fortune of air-conditioning is controlled by the real-time output parameter
Row；Wherein the input parameter includes environmental parameter, user characteristics parameter and users'comfort parameter, wherein the user
Comfort quantity is used to characterize the user that the user characteristics parameter indicates comfortable in the environment that the environmental parameter indicates
Degree；The output parameter includes air conditioner operation parameters.
Second aspect provides a kind of server, comprising:
It is pre-configured unit, for determining the input and output parameter of training sample；
Model selection unit, for obtaining the mind of the input parameter Yu output parameter mapping relations that indicate the training sample
Mathematical model through network；
Sample collection unit, for obtaining the training sample of test in predefined conditions；
Training unit carries out neural metwork training acquisition for the training sample of the test to be substituted into the mathematical model
Nerve network control system；
Transmission unit, for configuring the nerve network control system to air conditioner, so that the air conditioner acquires in fact
When input parameter input the nerve network control system and calculate real-time output parameter, pass through the real-time output parameter
Control the operation of air-conditioning；Wherein the input parameter includes environmental parameter, user characteristics parameter and users'comfort parameter,
Wherein the users'comfort parameter is used to characterize the user characteristics parameter what the user indicated indicated in the environmental parameter
Comfort level in environment；The output parameter includes air conditioner operation parameters.
The third aspect provides a kind of air-conditioning system, including air-conditioning and any of the above-described server.
Server can determine the input and output parameter of training sample in above scheme；Obtaining indicates the training
The mathematical model of the neural network of the input parameter and output parameter mapping relations of sample；The instruction of test is obtained in predefined conditions
Practice sample；The training sample of test is substituted into the mathematical model and carries out neural metwork training acquisition nerve network control system；
Nerve network control system is configured to air conditioner, inputs the neural network control so that air conditioner acquires input parameter in real time
The real-time output parameter of system-computed processed controls the operation of air-conditioning by real-time output parameter；The wherein input parameter packet
Environmental parameter, user characteristics parameter and users'comfort parameter are included, wherein the users'comfort parameter is described for characterizing
Comfort level of the user that user characteristics parameter indicates in the environment that the environmental parameter indicates；The output parameter includes air-conditioning
Operating parameter.
Since server can export the air conditioner operation parameters for airconditioning control to air conditioner, and eventually by the sky
Allocate and transport the operation of row state modulator air conditioner；To realize the real-time input parameter of air conditioner direct basis air conditioner acquisition
Air conditioner operating parameter is adjusted, and controls the operation of air conditioner by air conditioner operation parameters, without user to air-conditioning
The operating status of device is controlled, and the intelligence of airconditioning control is realized.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
It obtains other drawings based on these drawings.
Fig. 1 is a kind of structural schematic diagram for air-conditioning system that the embodiment of the present invention provides；
Fig. 2 is a kind of flow diagram for air conditioning control method that the embodiment of the present invention provides；
Fig. 3 is a kind of whole implementation model schematic diagram for air-conditioning system that the embodiment of the present invention provides；
Fig. 4 is the research object and purpose schematic diagram for a kind of air conditioning control method that the embodiment of the present invention provides；
Fig. 5 is a kind of test fundamental schematic diagram for air conditioning control method that the embodiment of the present invention provides；
Fig. 6 is a kind of BP neural network schematic diagram that the embodiment of the present invention provides；
Fig. 7 is a kind of neuron models schematic diagram for RBF neural that the embodiment of the present invention provides；
Fig. 8 is a kind of RBF neural schematic diagram that the embodiment of the present invention provides；
Fig. 9 is a kind of flow diagram for air-conditioner control method that another embodiment of the present invention provides；
Figure 10 is a kind of structural schematic diagram for server that the embodiment of the present invention provides.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
The system architecture and business scenario of description of the embodiment of the present invention are to more clearly illustrate implementation of the present invention
The technical solution of example, does not constitute the restriction for technical solution provided in an embodiment of the present invention, those of ordinary skill in the art
It is found that technical solution provided in an embodiment of the present invention is for similar with the differentiation of system architecture and the appearance of new business scene
The technical issues of, it is equally applicable.
The basic principle for the technical solution that the embodiment of the present invention provides are as follows: server can determine the input of training sample
Parameter and output parameter；Obtain the mathematical modulo of the neural network of the input parameter and output parameter mapping relations that indicate training sample
Type；The training sample of test is obtained in predefined conditions；The training sample of test is substituted into mathematical model and carries out neural network instruction
Practice and obtains nerve network control system；Then the nerve network control system that will acquire is configured to air conditioner, and air conditioner acquisition is real
When input parameter input nerve network control system calculate real-time output parameter, pass through real-time output parameter and control air-conditioning
Operation.Compared with the prior art, user can be completely free of the operation of remote controler or client, i.e., without user to air-conditioning
Operating status is controlled, and the intelligence of airconditioning control is realized.
Specifically scheme provided by the invention is illustrated referring to following embodiment:
The embodiment of the present invention is applied to air-conditioning system as shown in Figure 1, including air-conditioning 11 and positioned at the service in cloud
Device 12.
Based on above-mentioned air-conditioning system, referring to shown in Fig. 2, the embodiment of the present invention provides a kind of air conditioning control method, packet
It includes:
101, server determines the input and output parameter of training sample.
102, server obtains the neural network of the input parameter and output parameter mapping relations that indicate the training sample
Mathematical model.
103, server obtains the training sample of test in predefined conditions.
104, the training sample of the test is substituted into the mathematical model and carries out neural metwork training acquisition mind by server
Through network control system.
105, server configures the nerve network control system to air conditioner, so that air conditioner acquisition is real-time
It inputs parameter and inputs the real-time output parameter of the nerve network control system calculating, pass through the real-time output parameter and control
The operation of air-conditioning；Wherein the input parameter includes environmental parameter, user characteristics parameter and users'comfort parameter, wherein
The users'comfort parameter is used to characterize the user characteristics parameter environment that the user indicated indicates in the environmental parameter
In comfort level；The output parameter includes air conditioner operation parameters.
Server can determine the input and output parameter of training sample in above scheme；Obtaining indicates the training
The mathematical model of the neural network of the input parameter and output parameter mapping relations of sample；The instruction of test is obtained in predefined conditions
Practice sample；The training sample of test is substituted into the mathematical model and carries out neural metwork training acquisition nerve network control system；
Nerve network control system is configured to air conditioner, inputs the neural network control so that air conditioner acquires input parameter in real time
The real-time output parameter of system-computed processed controls the operation of air-conditioning by real-time output parameter；The wherein input parameter packet
Environmental parameter, user characteristics parameter and users'comfort parameter are included, wherein the users'comfort parameter is described for characterizing
Comfort level of the user that user characteristics parameter indicates in the environment that the environmental parameter indicates；The output parameter includes air-conditioning
Operating parameter.
Since server can export the air conditioner operation parameters for airconditioning control to air conditioner, and eventually by the sky
Allocate and transport the operation of row state modulator air conditioner；To realize the real-time input parameter of air conditioner direct basis air conditioner acquisition
Air conditioner operating parameter is adjusted, and controls the operation of air conditioner by air conditioner operation parameters, without user to air-conditioning
The operating status of device is controlled, and the intelligence of airconditioning control is realized.
Specifically referring to whole implementation model shown in Fig. 3, be described as follows to the basic realization principle of this programme: its is hollow
Adjusting device 11 is to realize that user improves the tool of indoor environment, therefore can be in server 12 beyond the clouds by user environment demand
(mainly being embodied in the form of inputting parameter to be environmental parameter, user characteristics parameter and users'comfort parameter) and sky
It adjusts operating status (mainly embodying in the form of output parameter to be air conditioner operation parameters) as research object, establish mathematical modulo
Determine that the collection rule of training sample of test, the purpose of sample collection are exactly to obtain the user environment demand of ordinary user after type
Data (training sample tested) and corresponding running state of air conditioner (can be characterized using air conditioner operation parameters), by right
The data processing of a large amount of collecting samples selects neural network to carry out off-line training, the nerve network control system obtained after training
With predictive ability, in conjunction with the control system of corresponding software and hardware configuration composition manual intelligent air conditioner, the manual intelligent
Air conditioner can automatically adjust output when detecting nerve network control system input, adjust without user.Certain user makes
Used time will be transmitted back to the server in cloud if you need to adjust air-conditioning, modified data, and the data passed back combine existing training sample
On the basis of re-start neural metwork training, the nerve network control system of update is downloaded to the sky of the user after on-line training
Adjust device, a kind of nerve network control system adaptive use of the user of the new nerve network control system as customization
Habit；A large number of users modified running state of air conditioner is collected by cloud, the data augmentation largely fed back the sky of training sample
Between, further progress neural metwork training upgraded after nerve network control system, nerve network control system matching
It is more accurate when user demand, so that the air-conditioning of next batch is more intelligent.As the data of user feedback are continuously increased, pass through
Adjustment several times, new manual intelligent air conditioner is lower and lower using matching degree error with user, the output symbol of air conditioner
User demand is closed, user is without amendment, to realize the artificial intelligence of air-conditioning.Specifically implementation of the invention is carried out specifically
It is bright as follows:
201, the mapping set relationship of the input and output parameter of the training sample of neural network: f (P is determinedi,Ei,
Si)→Y(Ti)；Wherein, PiFor user characteristics parameter, EiFor environmental parameter, SiFor users'comfort parameter；TiFor air-conditioning fortune
Row parameter.
Step 201 is intended to determine research object and target, and the input ginseng of the training sample of neural network is confirmed in the step
Several and output parameter.Wherein, human body is environmental degree of comfort parameter for the major demands of air conditioning function, environmental degree of comfort parameter
Specific manifestation index includes but is not limited to following several: hotness index, dry damp index, ride number.Distinct group
Body is different for the demand of environmental degree of comfort, and population difference is in particular in gender, age, posture and special population (pregnant woman, body
Body is uncomfortable).Further, human body is different for the index of the environmental degree of comfort demand under varying environment, can be marked according to country
The enthalpy difference experiment operating condition required in standard tests environmental degree of comfort demand as Typical Representative environment.
After human body determines the demand of the environmental degree of comfort under specific environment, air-conditioning is needed to maintain current ambient conditions,
So air-conditioning needs to be adjusted to certain working condition, shows different air conditioner operation parameters, such as: compressor operation frequency
The parameters such as rate plus/moisture removal, wind speed, air outlet form.
A kind of mapping relations, i.e. different crowd P can be taken out by above descriptioniFor varying environment EiEnvironment relax
Appropriate SiThere are certain demand Y, and this demand passes through running state of air conditioner TiIt maintains, shows as operation of air conditioner in corresponding ginseng
Number state.Mapping set relationship can be described as:
f(Pi,Ei,Si)→Y(Ti)
According to the above analysis, the corresponding relationship in mapping set can be in certain known user PiPhysiological characteristic and reality
Environment EiLower anticipating user is for environmental degree of comfort SiRequirement, and then control air conditioner operation parameters Ti。
Complexity is adapted to for crowd, this mapping relations are a kind of multivariable, the Complex Nonlinear System of close coupling, sheet
Invention analyzes fit non-linear system using neural metwork training.The crowd that sample space covers different characteristic as far as possible is corresponding
Environmental amenity degree demand, it is therefore desirable to carry out environmental testing for different crowd to obtain enough training sample spaces, protect
Demonstrate,prove the forecasting accuracy of nerve network system.
Referring to shown in Fig. 4, what above-mentioned steps 201 determined: the object of test are as follows: different crowd PiFor varying environment Ei's
Environmental degree of comfort SiThere are the corresponding running state of air conditioner parameter T of certain demand Yi；The target of test are as follows: instructed using neural network
Practice analysis fitting mapping relationship f (Pi,Ei,Si)→Y(Ti), this mapping relations are obtained by neural metwork training, are being detected
Air conditioner is directly predicted output air conditioner operation parameters and is run to required state when input.
202, the input parameter and output parameter mapping relations of the training sample are indicated by the neural network
Mathematical model y (t)=Παx(t)；
Wherein, ΠαIndicate neural network；xd(t)=[xd1,xd2,...xdp]TIndicate the p dimensional input vector of input parameter, y
(t)=[y1,y2,...yq]TIndicate that the q of output parameter ties up output vector；xd1,xd2,...xdpIndicate p input parameter；y1,
y2,...yqIndicate q output parameter.
After clear research object and target, acquired training sample can train neural network, general nerve net
Network ∏αMapping relations in step 201, ∏ can be described by founding mathematical modelsαInput is p dimensional vector xd(t)=[xd1,
xd2,...xdp]T, export q dimensional vector y (t)=[y1,y2,...yq]T, general
Y (t)=Παx(t)
Specifically combine intelligent air conditioner actual conditions:
Input parameter X=[x1,x2,x3]T=[Pi,Ei,Si]T,
Input parameter Y=[y1,y2,y3,y4]T=[Fi,Ji,Mi,Ai]T
Further Pi=[ai,si,hi,ki]；Ei=[t(indoor)i,t(outdoor)i,h(indoor)i,ei]；Si=[Hi,Wi,Ci]
Then mathematical model may be summarized to be:
[Fi,Ji,Mi,Ai]T=f { [Pi,Ei,Si]T}
Wherein: ai-- tester's age；
si-- tester's gender；
hi-- tester's posture；
ki-- tester's health status；
t(indoor)i-- operating condition room temperature；
t(outdoor)i-- operating condition outdoor temperature；
h(indoor)i-- humidity in operating condition room；
ei-- other environmental parameters of operating condition；
Hi-- hot comfort；
Wi-- dry and wet comfort level；
Ci-- subjective assessment index；
Fi-- operation of air conditioner compressor frequency；
Ji-- operation of air conditioner adds the operating parameter of dehumidification device；
Mi-- operation of air conditioner fan motor rotational speed；
Ai-- operation of air conditioner wind deflector angle；
P dimensional vector x can must be inputted by analyzing aboved(t)=[xd1,xd2,...xdp]TFor 11 dimensional input vectors: xd(t)=
[ai,si,hi,ki,t(indoor)i,t(outdoor)i,h(indoor)i,ei,Hi,Wi,Ci]T
Q dimensional vector y (t)=[y1,y2,...yq]TOutput vectors: y (t)=[F are tieed up for 4i,Ji,Mi,Ai]T
It is possible thereby to determine the complex nonlinear system for needing to obtain this 11 dimension input, 4 dimension outputs by neural metwork training
System, and this system can input corresponding output with Accurate Prediction.Furthermore main affecting parameters are only provided in model of the present invention,
But these parameters are not limited to, such as user characteristics parameter can also include height, shell temperature, degrees of motion etc., environmental parameter
It can also can also be the further subdivision to environmental degree of comfort including other factors, the users'comfort parameter in environment.Such as
After increasing relevant parameter, it is only necessary to which adjustment inputs the model for updating neural network with the dimension of output, and which is not described herein again.
203, the training sample of test is obtained in predefined conditions.
The mathematical model of conclusion needs neural metwork training to lead to generally for the precision of prediction for guaranteeing nerve network system
Training sample acquisition should have popularity and universal adaptability in normal situation.
According to the research purpose and object of mathematical model, the test fundamental such as attached drawing 5 of the training sample of test is obtained:
Operating condition of test selection, tester's mass selection are selected, evaluation index selects, air conditioner operation parameters selection:
(1) determine tester's distribution proportion, the testee (men and women) of different sexes require the distribution of covering different age group,
Different posture distributions, special population distribution and different health conditions etc..
The confirmation of tester's radix: a certain the smallest number of class testing person accounting for meeting setting is more than or equal to 1, such as accounting
The smallest is P=0.001%, therefore sum is at least 1000, to guarantee to cover whole features.
Classify in specific implementation process not limited to this, classification can be refined further, required tester's number after refinement
Amount increases, and is contemplated that overall cost selects tester's radix early period, generality, the popularity of tester will affect using nerve net
Network prediction, sample size is bigger in principle, and it is more accurate to predict.It is most by testing accurate acquisition crowd couple and the demand of comfort level
Main target corrects data by user in the later period and updates neural network control certainly in the case where tester's limited amount
System processed can also be improved the accuracy of prediction.
(2) operating condition of test can choose the enthalpy difference experiment operating condition in national standard: specified refrigeration, specified heating, maximum
(small) operation heating, maximum (small) operation heating, high-temperature refrigeration, low-temperature heating ..., due to these operating conditions be it is discrete, further
The precision for improving neural network prediction can further segment operating condition of test.
(3) evaluation index is characterization user for environmental degree of comfort demand (in mathematical model users'comfort parameter list
Sign) form, index, comfort level subjective assessment, which refer to, mainly to be felt for the hotness index under specific environment, dry and wet from crowd
Number etc. is evaluated, and the termination condition of test is the best demand for reaching user for environmental degree of comfort.
(4) demand of user can be maintained by running state of air conditioner, and the operating status of air-conditioning can be compressed by control
The running frequency of machine plus/dehumidification device plus/the air conditioner operation parameters such as dehumidified state, fan motor wind speed, wind deflector angle protect
Card.
204, the training sample of the test is substituted into the mathematical model and carries out neural metwork training acquisition neural network
Control system.
Artificial neural network (Artificial neural networks, ANN) be also referred to as neural network (NNs) or
Make link model, it is a kind of algorithm mathematics of simulation human nerve network behavior feature progress distributed parallel information processing
Type.Neural network is a kind of operational model, is constituted by being coupled to each other between a large amount of node (or neuron).Each node generation
A kind of specific output function of table, referred to as excitation function (activation function).Connection all generations between every two node
One, the table weighted value for passing through the connection signal, referred to as weight, this is equivalent to the memory of artificial neural network.Network
Output then relies on the connection type of network, the difference of weighted value and excitation function and it is different.And network itself is usually all to certainly
Right certain algorithm of boundary or function approach, it is also possible to the expression to a kind of logic strategy.Data are carried out using neural network
Off-line training, the purpose of neural network is to obtain the complicated system of people, air-conditioning, multivariable nonlinearity close coupling between environment
System.The present invention mainly uses BP neural network and RBF neural method training test sample, BP neural network and RBF nerve
Network is the neural network method for typically solving the problems, such as Nonlinear Multivariable complication system, and other forms equally also can be used
Neural network structure be trained, the present invention not one by one carry out detailed analysis description.
(1) BP neural network is a kind of error back propagation feedforward neural network, the typical basic structure of BP neural network
As shown in the figure.Usually given n dimensional input vector P=(p1,p2,...,pn)TIt is normalized to obtain n dimensional input vector X ∈ (x1,
x2,...,xn)T；Hidden layer has m neuron, obtains hidden layer output H=(h1,h2,...,hm)T；Output layer has k nerve
Member obtains output layer output Y=(y1,y2,...,yk)T；Output is subjected to anti-normalization processing and obtains Q=(q1,q2,...,qk)T
Sample training output；Weight between input layer and hidden layer is wij, threshold value θj, the weight between hidden layer and output layer is
vjh, threshold value τh。
Wherein the output of each layer neuron meets:
BP neural network basic structure is as shown in Figure 6；
The selection of BP neural network node in hidden layer:
Known BP neural network outputs and inputs as 11-4, and node in hidden layer is undetermined, and hidden layer node number is straight
The performance indicator for influencing training is connect, and influences certain network generalization.Actually the selection of node in hidden layer is excessively
Complexity does not still provide accurate theory deduction, can only estimate general range, need by being constantly trying to final determination
Reasonable number.In general, node in hidden layer is chosen excessive, can extend learning time and delay efficiency, otherwise can reduce again
The precision and fault-tolerance of neural network.Hidden layer neuron number m depend on input vector dimension n and can number of partitions M, for
The optimal hidden layer node number of fitting system is referred to formula, and k is output vector dimension, and a is constant, generally less than
10。
(2) quasi- to train using another form of neural network, that is, radial basis function neural network (RBF neural)
Close inverse system.The difference of the building and design of RBF and BP neural network is: curve matching of the RBF based on a higher dimensional space
Problem rather than weight and threshold value press the stochastic approximation of gradient negative direction decline.The former is that higher dimensional space interpolation is sought most preferably
The curved surface of fitting and the latter is the best fit in statistical significance.In addition, RBF has also been proved to that fitting arbitrary continuation can be approached
Nonlinear function, RBF neural increasingly favored in nonlinear Identification field.
Radial basis function neural network abbreviation RBF neural, the same with BP neural network is all common feedforward network,
It is mainly used for the mathematical model that fitting has nonlinear function controlled device, due to its characteristic with partial approximation,
Certain advantage is shown when the analytical function of study and training complication system.
RBF neural is made of multiple radial base neurons, and wherein the transfer function of hidden layer uses radial basis function,
It is the function general name to Euclidean distance dullness, and radial base neuron models are as shown in Figure 7.
RBF neural is generally three-decker, and typical basic structure is as shown in the figure.Usually given n dimension input to
Measure P=(p1,p2,...,pn)TIt is normalized to obtain n dimensional input vector X ∈ (x1,x2,...,xn)T；Hidden layer neuron simultaneously
Number h, output layer have n neuron, and the weight of hidden layer to output layer is Whn, the threshold value of hidden layer to output layer is B=
(b1,b2,...,bn)T, Ri(||x-ci| |) be i-th of concealed nodes transfer function, obtain hidden layer output H=(h1,
h2,...,hm)T, obtain output layer output Y=(y1,y2,...,yn)T；Output is subjected to anti-normalization processing and obtains Q=(q1,
q2,...,qn)TSample training output.Transfer function Euclidean distance in neuron models | | dist | | it is equal to radial base neural net
Input vector P=(p1,p2,...,pn)TWith weight WhnThe distance between Ri(||x-ci| |), radial base neural net structure is as schemed
Shown in 8.
The general step of training BP/RBF neural network and training program are as follows in step 204:
A. determine that n group is used as the input vector collection for inputting parameter and n group as defeated in the training sample of the test
The output vector collection of parameter out.
That is: the input and output parameter of training sample is defined: p represent n group by [x1, x2, x3, x4, x5,
X6, x7, x8, x9, x10, x11]=[ai,si,hi,ki,t(indoor)i,t(outdoor)i,h(indoor)i,ei,Hi,Wi,Ci] composition instruction
Practice input vector collection used, t represents n group by [y1, y2, y3, y4]=[Fi,Ji,Mi,Ai] composition training used in output vector
Collection, program are as follows:
P=[x1 (1:n) '；x2(1:n)'；x3(1:n)'；x4(1:n)'；x5(1:n)'；x6(1:n)'；x7(1:n)'；x8
(1:n)'；x9(1:n)'；x10(1:n)'；x11(1:n)']；
T=[y1 (1:n) '；y2(1:n)'；y3(1:n)'；y4(1:n)']；
B. the input vector collection and output vector collection are normalized
Pn, tn represent the data of p, t after the processing that maximin method is normalized, and program is as follows:
[pn, minp, maxp]=premnmx (p)；
[tn, mint, maxt]=premnmx (t).
E. the operational model of neural metwork training is chosen for the mathematical model, and initializes the structure of neural network,
By the neural network of initialization to after normalized input vector collection and output vector collection carry out neural metwork training and obtain
Take nerve network control system；Wherein the operational model includes including at least below with the mind of fit non-linear system capability
Through network: feed-forward neural network, radial basis function neural network.
For BP neural network: net=newff (pn, tn, [node in hidden layer, imply the number of plies], ' input layer is to hidden
Transmission function containing layer ', ' hidden layer to output layer transmission function ', ' training algorithm ', ' e-learning function ')；
For RBF neural net=newrbe, (pn, tn, training objective extend constant, neuron maximum number, brush
New rate) or net=newrbe (pn, tn extend constant).
Wherein, the determination of above-mentioned neural network and the program in training process are to realize the specific example of corresponding method, journey
Sequence logic can be by different software realizations, here with no restrictions.
205, the nerve network control system is configured to air conditioner, so that the air conditioner acquires input ginseng in real time
Number inputs the nerve network control system and calculates real-time output parameter, controls air-conditioning by the real-time output parameter
Operation.
In above process by the numerical value software for calculation neural metwork training such as MATLAB/DSPACE after, mind can be obtained
Through network control system, wherein above-mentioned training process is usually to be instructed in the enough situations of training sample using offline mode
Practice, the numerical module of nerve network control system is generated after training, the composing softwares such as MATLAB/DSPACE can be passed through
The CPU main control chip that program code is downloaded to air conditioner is converted by nerve network control system.
In step 205, the air conditioner acquisition input parameter input in real time nerve network control system calculating is real-time
Output parameter, realized by the operation that the real-time output parameter controls air-conditioning especially by following steps:
A, air conditioner acquisition inputs parameter in real time；
B, the corresponding real-time input vector collection of the real-time input parameter of the nerve network control system, Yi Jishi are determined
When the corresponding real-time output vector collection of output parameter；
Real-time input parameter and real-time output parameter are defined: p2, t2 represent 100 groups [x1, x2, x3,
X4, x5, x6, x7, x8, x9, x10, x11], input vector collection, in real time in real time used in the forecast test of [y1, y2, y3, y4] composition
Output vector collection, program are as follows:
P2=[x1 (n+1:n+100) '；x2(n+1:n+100)'；x3(n+1:n+100'；x4(n+1:n+100)'；x5(n+
1:n+100)'；x6(n+1:n+100)'；x7(n+1:n+100)'；x8(n+1:n+100)'；x9(n+1:n+100)']；
T2=[y1 (n+1:n+100) '；y2(n+1:n+100)'；y3(n+1:n+100)'；y4(n+1:n+100)']；
C, the real-time input vector collection is normalized；
Illustratively, p2n represents data of the p2 after the processing that maximin method is normalized, and program is such as
Under:
[p2n]=tramnmx (p2, minp, maxp)；
D, the real-time input vector collection after normalized is substituted into the nerve network control system and calculates reality output
Vector set；
By taking BP neural network as an example, trained BP neural network control system predicts output: it is defeated in real time that an represents normalization
The trained obtained reality output of BP neural network control system of incoming vector collection p2n:
An=sim (net, p2n)；
E, renormalization is carried out to the reality output vector set and calculates the acquisition real-time output vector collection.
Trained nerve network control system prediction output renormalization: pr represents reality output an by renormalization
The real-time output vector collection that postmnmx order obtains:
[pr]=postmnmx (an, mint, maxt)；
After nerve network system program is downloaded to CPU main control chip, the air conditioner containing nerve network control system program
It needs to meet certain condition in hardware configuration, overall arrangement includes acquisition unit, which needs to be equipped with camera, infrared
The various kinds of sensors such as sensor, dry and wet sensor, temperature sensor, and this configuration is in addition to camera, other desired sensings
Device configuration is typically all the basic configuration of air-conditioning.The analog signal of acquisition unit is converted digital signal by input unit, such as
The users such as characterization user's gender, age, body temperature, posture will be converted to by the analog signal of camera, infrared sensor acquisition
The analog signal that temperature sensor detects is converted to indoor environment parameter or Outdoor Air Parameters etc. by the digital signal of feature
Deng；Processing unit is to export processing unit using nerve network control system as the processor of core, can be by ANN Control
The real-time airconditioning control Parameter Switch of system output is that driving signal drives air conditioner to adjust running frequency, humidification degree, wind speed, go out
Air port form etc..
Air conditioner is installed to user in specific above-mentioned air-conditioning system, and the scheme of specific air conditioner control is referring to attached drawing
9 are described as follows:
(1) user when in use, only needs electrifying startup air conditioner, by imaging sensor (as imaged after air conditioner booting
Head is infrared), the discriminations such as infrared sensor using the user characteristics such as the gender of user, age and posture parameter and feed back；It is logical
It crosses the processing indoor and outdoor surroundings parameter such as humidity sensor, temperature sensor and air velocity transducer and feeds back, comfort quantity is silent
Recognize is that ride number is best (=0)；
(2) analog signal of various kinds of sensors feedback is that digital signal feeds back to air conditioner CPU master by analog-digital conversion a/d
Chip is controlled, after air conditioner CPU obtains input signal, is transported according to the nerve network control system for being downloaded to chip
It calculates, output signal is obtained after operation, output signal is the user of nerve network control system prediction in current environment
Under air conditioner operation parameters corresponding for suitable environment comfort level demand, output signal directly controls compressor, fan, plus removes
The ambient condition that the operation such as wet device is needed to user；
(3) if the air conditioner operation parameters that exports according to nerve network control system of certain user booting with actual user most
There are certain deviation, user can be finely adjusted to optimum reelability quality state good comfort level demand by air conditioner operation panel, be used
Optimum reelability quality state corresponding air conditioner operation parameters in family feed back to the CPU of air conditioner, and the CPU of air conditioner passes through data are corrected
Server is passed in cloud back, and the modified data of the user combine the training sample of former test to re-start neural metwork training, is updated
The nerve network control system of version is re-downloaded by cloud service to the air conditioner of the user, to meet user for comfort level
Best demand.Each adjustment of the user can all be recognized by CPU and carry out online adaptive adjustment, to realize air-conditioning
The function of commonly using family and thinking, by recycling several times, n.0 (personal customization version) nerve network control system can be exactly matched
The user demand, to realize that air-conditioning is private customized and intelligent；
(4) N number of different user can pass server back by cloud for the adjustment of operation of air conditioner, to the data passed back into
Row taxonomic revision, as the training sample of next batch air conditioner off-line training, training sample is anti-with marketing users data
Present it is increasing, it is a large amount of to correct data and the trained new nerve network control system of legacy data N.0 version (general-purpose version) is next
The air conditioner operation parameters of batch are more excellent for the generalization and error resilience performance of different user, and user needs to adjust general
Rate becomes smaller.As the operation of multiple batches of air conditioner enters the market, in the case where training sample abundance enough, makes after air conditioner unlatching
Judgement complies fully with the actual demand of user, therefore the air conditioner intelligent under the big data realized, in this case operation of air conditioner
No longer user is needed to adjust, the variation of parameter is inputted according to systems such as environmental parameters, air conditioner is independently adjusted, and air conditioner will
Intelligent robot as adjustment user environment comfort level.
The embodiment of the present invention provides a kind of server, for implementing above-mentioned method, referring to Fig.1 shown in 0, comprising:
It is pre-configured unit 11, for determining the input and output parameter of training sample；
Model selection unit 12, for obtaining the input parameter and output parameter mapping relations that indicate the training sample
The mathematical model of neural network；
Sample collection unit 13, for obtaining the training sample of test in predefined conditions；
Training unit 14 is obtained for the training sample of the test to be substituted into the mathematical model progress neural metwork training
Take nerve network control system；
Transmission unit 15, for configuring the nerve network control system to air conditioner, so as to air conditioner acquisition
Input parameter inputs the nerve network control system and calculates real-time output parameter in real time, is joined by the real-time output
The operation of number control air-conditioning；Wherein the input parameter includes environmental parameter, user characteristics parameter and users'comfort ginseng
Number, wherein the users'comfort parameter is used to characterize the user characteristics parameter user indicated to be indicated in the environmental parameter
Environment in comfort level；The output parameter includes air conditioner operation parameters.
A kind of specific embodiment is provided, unit 11 is pre-configured and is specifically used for determining the defeated of the training sample of neural network
Enter the mapping set relationship of parameter and output parameter: f (Pi,Ei,Si)→Y(Ti)；
Wherein, PiFor user characteristics parameter, EiFor environmental parameter, SiFor users'comfort parameter；TiFor operation of air conditioner ginseng
Number.
A kind of specific embodiment is provided, model selection unit 12 is specifically used for indicating by the neural network
The input parameter of the training sample and mathematical model y (t)=Π of output parameter mapping relationsαx(t)；
Wherein, ΠαIndicate neural network；xd(t)=[xd1,xd2,...xdp]TIndicate the p dimensional input vector of input parameter, y
(t)=[y1,y2,...yq]TIndicate that the q of output parameter ties up output vector；xd1,xd2,...xdpIndicate p input parameter；y1,
y2,...yqIndicate q output parameter.
A kind of specific embodiment is provided, training unit 14 is specifically used for determining n in the training sample of the test
Group is used as the output vector collection of input vector collection and n group as output parameter of input parameter；To the input vector collection and
Output vector collection is normalized；The operational model of neural metwork training is chosen for the mathematical model, and is initialized
The structure of neural network, by the neural network of initialization to after normalized input vector collection and output vector collection carry out
Neural metwork training obtains nerve network control system；Wherein the operational model includes including at least to have fitting non-thread below
The neural network of property system capability: feed-forward neural network, radial basis function neural network.
It should be noted that transmission unit can be for using the realization of following form in above scheme: transceiver, sending and receiving end
Mouth, transmission circuit etc., being pre-configured unit, model selection unit, sample collection unit, training unit can individually set up
Processor also can integrate and realize in some processor of server, in addition it is also possible to be stored in the form of program code
In the memory of server, the above pre-configuration unit, model choosing are called by some processor of air-conditioner controller and executed
Take the function of unit, sample collection unit, training unit.Processor described here can be a central processing unit
(Central Processing Unit, CPU) or specific integrated circuit (Application Specific
Integrated Circuit, ASIC), or be arranged to implement one or more integrated circuits of the embodiment of the present invention.
The above description is merely a specific embodiment, but scope of protection of the present invention is not limited thereto, any
Those familiar with the art in the technical scope disclosed by the present invention, can easily think of the change or the replacement, and should all contain
Lid is within protection scope of the present invention.Therefore, protection scope of the present invention should be based on the protection scope of the described claims.
Claims (8)
1. a kind of air conditioning control method characterized by comprising
Determine the input and output parameter of training sample；
Obtain the mathematical model of the neural network of the input parameter and output parameter mapping relations that indicate the training sample；
The training sample of test is obtained in predefined conditions；
The training sample of the test is substituted into the mathematical model and carries out neural metwork training acquisition nerve network control system；
The nerve network control system is configured to air conditioner, inputs institute so that the air conditioner acquires input parameter in real time
It states nerve network control system and calculates real-time output parameter, the operation of air-conditioning is controlled by the real-time output parameter；Its
Described in input parameter include environmental parameter, user characteristics parameter and users'comfort parameter, wherein the users'comfort
Parameter is used to characterize comfort level of the user of the user characteristics parameter expression in the environment that the environmental parameter indicates；It is described
Output parameter includes air conditioner operation parameters；
Wherein, the mathematical modulo of the neural network of the input parameter and output parameter mapping relations that indicate the training sample is obtained
Type, comprising:
The input parameter of the training sample and the mathematical modulo of output parameter mapping relations are indicated by the neural network
Type: y (t)=Παx(t)；
Wherein, ΠαIndicate neural network, xd(t)=[xd1,xd2,...xdp]TIndicate the p dimensional input vector of input parameter, y (t)
=[y1,y2,...yq]TIndicate that the q of output parameter ties up output vector, xd1,xd2,...xdpIndicate p input parameter, y1,y2,
...yqIndicate q output parameter；
It is specific:
Input parameter X=[x1,x2,x3]T=[Pi,Ei,Si]T,
Input parameter Y=[y1,y2,y3,y4]T=[Fi,Ji,Mi,Ai]T,
Further Pi=[ai,si,hi,ki], Ei=[t(indoor)i,t(outdoor)i,h(indoor)i,ei], Si=[Hi,Wi,Ci]；
Then mathematical model are as follows:
[Fi,Ji,Mi,Ai]T=f { [Pi,Ei,Si]T}；
Wherein: ai-- tester's age；
si-- tester's gender；
hi-- tester's posture；
ki-- tester's health status；
t(indoor)i-- operating condition room temperature；
t(outdoor)i-- operating condition outdoor temperature；
h(indoor)i-- humidity in operating condition room；
ei-- other environmental parameters of operating condition；
Hi-- hot comfort；
Wi-- dry and wet comfort level；
Ci-- subjective assessment index；
Fi-- operation of air conditioner compressor frequency；
Ji-- operation of air conditioner adds the operating parameter of dehumidification device；
Mi-- operation of air conditioner fan motor rotational speed；
Ai-- operation of air conditioner wind deflector angle；
Obtain input p dimensional vector xd(t)=[xd1,xd2,...xdp]TFor 11 dimensional input vectors:
xd(t)=[ai,si,hi,ki,t(indoor)i,t(outdoor)i,h(indoor)i,ei,Hi,Wi,Ci]T,
Q dimensional vector y (t)=[y1,y2,...yq]TOutput vectors: y (t)=[F are tieed up for 4i,Ji,Mi,Ai]T。
2. being wrapped the method according to claim 1, wherein determining the input and output parameter of training sample
It includes:
Determine the mapping set relationship of the input and output parameter of the training sample of neural network: f (Pi,Ei,Si)→Y
(Ti)；
Wherein, PiFor user characteristics parameter, EiFor environmental parameter, SiFor users'comfort parameter；TiFor air conditioner operation parameters.
3. the method according to claim 1, wherein the training sample of the test is substituted into the mathematical model
It carries out neural metwork training and obtains nerve network control system, comprising:
Determine that n group is used as the input vector collection of input parameter and n group is used as output parameter in the training sample of the test
Output vector collection；
The input vector collection and output vector collection are normalized；
The operational model of neural metwork training is chosen for the mathematical model, and initializes the structure of neural network, by first
The neural network of beginningization is to the input vector collection and output vector collection progress neural metwork training acquisition nerve after normalized
Network control system；Wherein the operational model includes including at least below with the nerve net of fit non-linear system capability
Network: feed-forward neural network, radial basis function neural network.
4. the method according to claim 1, wherein air conditioner acquisition is inputted in real time described in parameter input
Nerve network control system calculates real-time output parameter；Include:
The air conditioner acquisition inputs parameter in real time；
Determine the corresponding real-time input vector collection of the real-time input parameter of the nerve network control system, and defeated in real time
The corresponding real-time output vector collection of parameter out；
The real-time input vector collection is normalized；
Real-time input vector collection after normalized is substituted into the nerve network control system and calculates reality output vector set；
Renormalization is carried out to the reality output vector set and calculates the acquisition real-time output vector collection.
5. a kind of server characterized by comprising
It is pre-configured unit, for determining the input and output parameter of training sample；
Model selection unit, for obtaining the nerve net of the input parameter Yu output parameter mapping relations that indicate the training sample
The mathematical model of network；
Sample collection unit, for obtaining the training sample of test in predefined conditions；
Training unit carries out neural metwork training acquisition nerve for the training sample of the test to be substituted into the mathematical model
Network control system；
Transmission unit, for configuring the nerve network control system to air conditioner, so that air conditioner acquisition is real-time
It inputs parameter and inputs the real-time output parameter of the nerve network control system calculating, pass through the real-time output parameter and control
The operation of air-conditioning；Wherein the input parameter includes environmental parameter, user characteristics parameter and users'comfort parameter, wherein
The users'comfort parameter is used to characterize the user characteristics parameter environment that the user indicated indicates in the environmental parameter
In comfort level；The output parameter includes air conditioner operation parameters；
Wherein, the model selection unit is specifically used for indicating that the input of the training sample is joined by the neural network
Several mathematical models with output parameter mapping relations:
Y (t)=Παx(t)；
Wherein, ΠαIndicate neural network, xd(t)=[xd1,xd2,...xdp]TIndicate the p dimensional input vector of input parameter, y (t)
=[y1,y2,...yq]TIndicate that the q of output parameter ties up output vector, xd1,xd2,...xdpIndicate p input parameter, y1,y2,
...yqIndicate q output parameter；
It is specific:
Input parameter X=[x1,x2,x3]T=[Pi,Ei,Si]T,
Input parameter Y=[y1,y2,y3,y4]T=[Fi,Ji,Mi,Ai]T,
Further Pi=[ai,si,hi,ki]；Ei=[t(indoor)i,t(outdoor)i,h(indoor)i,ei], Si=[Hi,Wi,Ci]；
Then mathematical model are as follows:
[Fi,Ji,Mi,Ai]T=f { [Pi,Ei,Si]T}；
Wherein: ai-- tester's age；
si-- tester's gender；
hi-- tester's posture；
ki-- tester's health status；
t(indoor)i-- operating condition room temperature；
t(outdoor)i-- operating condition outdoor temperature；
h(indoor)i-- humidity in operating condition room；
ei-- other environmental parameters of operating condition；
Hi-- hot comfort；
Wi-- dry and wet comfort level；
Ci-- subjective assessment index；
Fi-- operation of air conditioner compressor frequency；
Ji-- operation of air conditioner adds the operating parameter of dehumidification device；
Mi-- operation of air conditioner fan motor rotational speed；
Ai-- operation of air conditioner wind deflector angle；
Obtain input p dimensional vector xd(t)=[xd1,xd2,...xdp]TFor 11 dimensional input vectors:
xd(t)=[ai,si,hi,ki,t(indoor)i,t(outdoor)i,h(indoor)i,ei,Hi,Wi,Ci]T,
Q dimensional vector y (t)=[y1,y2,...yq]TOutput vectors: y (t)=[F are tieed up for 4i,Ji,Mi,Ai]T。
6. server according to claim 5, which is characterized in that
The unit that is pre-configured is specifically used for determining the mapping ensemblen of the input and output parameter of the training sample of neural network
Conjunction relationship: f (Pi,Ei,Si)→Y(Ti)；
Wherein, PiFor user characteristics parameter, EiFor environmental parameter, SiFor users'comfort parameter；TiFor air conditioner operation parameters.
7. server according to claim 5, which is characterized in that
The training unit is specifically used for determining that n group is used as the input vector of input parameter in the training sample of the test
Collection and n group are used as the output vector collection of output parameter；
The input vector collection and output vector collection are normalized；
The operational model of neural metwork training is chosen for the mathematical model, and initializes the structure of neural network, by first
The neural network of beginningization is to the input vector collection and output vector collection progress neural metwork training acquisition nerve after normalized
Network control system；Wherein the operational model includes including at least below with the nerve net of fit non-linear system capability
Network: feed-forward neural network, radial basis function neural network.
8. a kind of air-conditioning system, which is characterized in that including air-conditioning and such as the described in any item servers of claim 5-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610709749.2A CN106322656B (en) | 2016-08-23 | 2016-08-23 | A kind of air conditioning control method and server and air-conditioning system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610709749.2A CN106322656B (en) | 2016-08-23 | 2016-08-23 | A kind of air conditioning control method and server and air-conditioning system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106322656A CN106322656A (en) | 2017-01-11 |
| CN106322656B true CN106322656B (en) | 2019-05-14 |
Family
ID=57742425
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610709749.2A Active CN106322656B (en) | 2016-08-23 | 2016-08-23 | A kind of air conditioning control method and server and air-conditioning system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106322656B (en) |
Families Citing this family (44)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106705384A (en) * | 2017-02-09 | 2017-05-24 | 美的集团股份有限公司 | Refrigerant leakage reminding method and device and air conditioner |
| DE102017213247A1 (en) * | 2017-06-30 | 2019-01-03 | Conti Temic Microelectronic Gmbh | Knowledge transfer between different deep-learning architectures |
| CN109491253B (en) * | 2017-09-11 | 2021-12-21 | 安徽师范大学 | Online learning type personalized intelligent home system and control method thereof |
| CN107797459B (en) * | 2017-09-15 | 2020-09-25 | 珠海格力电器股份有限公司 | Control method and device of terminal equipment, storage medium and processor |
| CN107844073B (en) * | 2017-09-15 | 2018-12-07 | 珠海格力电器股份有限公司 | Control method of electronic device, device, system, storage medium and processor |
| CN107730000A (en) * | 2017-09-15 | 2018-02-23 | 珠海格力电器股份有限公司 | Coolant leakage detection method and device for air-conditioning |
| CN107741084B (en) * | 2017-09-30 | 2019-09-27 | 广东美的制冷设备有限公司 | The recommended method of air conditioner and its operating parameter, system and big data server |
| CN110059819B (en) * | 2018-01-17 | 2022-11-25 | 腾讯科技（深圳）有限公司 | Device work control method, device, system, control equipment and storage medium |
| CN108361927A (en) * | 2018-02-08 | 2018-08-03 | 广东美的暖通设备有限公司 | A kind of air-conditioner control method, device and air conditioner based on machine learning |
| CN110631221A (en) * | 2018-06-05 | 2019-12-31 | 珠海格力电器股份有限公司 | Control method and device of air conditioner, terminal and storage medium |
| CN109028480B (en) * | 2018-06-08 | 2021-08-20 | 杭州博联智能科技股份有限公司 | Constant temperature and humidity control system and method thereof |
| CN110578994B (en) * | 2018-06-11 | 2021-01-29 | 珠海格力电器股份有限公司 | Operation method and device |
| CN109060115B (en) * | 2018-07-31 | 2020-05-26 | 珠海格力电器股份有限公司 | Noise analysis method, device, storage medium and system for equipment |
| CN110779175A (en) * | 2018-07-31 | 2020-02-11 | 珠海格力电器股份有限公司 | Operation mode control method and device |
| CN110836515A (en) * | 2018-08-17 | 2020-02-25 | 珠海格力电器股份有限公司 | Household equipment control method and device, control equipment and readable storage medium |
| CN110836509A (en) * | 2018-08-17 | 2020-02-25 | 珠海格力电器股份有限公司 | Household equipment control method and device, control equipment and readable storage medium |
| CN110953679A (en) * | 2018-09-26 | 2020-04-03 | 珠海格力电器股份有限公司 | Method and device for adjusting air conditioner and air conditioner |
| CN109164707A (en) * | 2018-09-28 | 2019-01-08 | 苏州市建筑科学研究院集团股份有限公司 | A kind of indoor environment negative-feedback regu- lation system based on artificial neural network algorithm |
| CN110966731B (en) * | 2018-09-28 | 2020-12-22 | 珠海格力电器股份有限公司 | Method for regulating operating parameters |
| CN109297140A (en) * | 2018-10-15 | 2019-02-01 | 宁波溪棠信息科技有限公司 | A kind of air conditioning control method based on artificial intelligence |
| CN111090238A (en) * | 2018-10-24 | 2020-05-01 | 宝马股份公司 | Control method, device, equipment, system and storage medium of intelligent household equipment |
| CN109405195A (en) * | 2018-10-31 | 2019-03-01 | 四川长虹电器股份有限公司 | Air conditioner intelligent control system and method |
| CN109492771A (en) * | 2018-11-12 | 2019-03-19 | 北京百度网讯科技有限公司 | Exchange method, device and system |
| CN109595765A (en) * | 2018-12-10 | 2019-04-09 | 珠海格力电器股份有限公司 | Air-conditioner control method, device, storage medium and air conditioner |
| CN109631255A (en) * | 2018-12-10 | 2019-04-16 | 珠海格力电器股份有限公司 | A kind of air conditioning control method, device, storage medium and air-conditioning |
| CN109827292A (en) * | 2019-01-16 | 2019-05-31 | 珠海格力电器股份有限公司 | Construction method, control method, the household electrical appliances of household electrical appliances adaptive power conservation Controlling model |
| CN109901406A (en) * | 2019-02-22 | 2019-06-18 | 珠海格力电器股份有限公司 | Monitoring method, device, equipment and the storage medium of building environment |
| CN109899937A (en) * | 2019-03-12 | 2019-06-18 | 王馨仪 | Comfort level and the foreseeable air conditioning system of section and method based on LSTM model |
| CN111765604B (en) * | 2019-04-01 | 2021-10-08 | 珠海格力电器股份有限公司 | Control method and device of air conditioner |
| CN112146240B (en) * | 2019-06-28 | 2023-10-03 | 广东美的制冷设备有限公司 | Operation control method, control device, drive-by-wire apparatus, server, and storage medium |
| CN110456647B (en) * | 2019-07-02 | 2020-11-27 | 珠海格力电器股份有限公司 | Intelligent household control method and intelligent household control device |
| CN112443943A (en) * | 2019-08-30 | 2021-03-05 | 珠海格力电器股份有限公司 | Model training method based on small amount of labeled data, control system and air conditioner |
| CN110595012B (en) * | 2019-09-23 | 2021-08-31 | 海信（广东）空调有限公司 | Recommendation method and device for air conditioner operation mode |
| CN110726215B (en) * | 2019-10-29 | 2020-11-03 | 珠海格力电器股份有限公司 | Air conditioner, control method and device thereof, storage medium and processor |
| CN110966733A (en) * | 2019-11-22 | 2020-04-07 | 广州医科大学附属第一医院（广州呼吸中心） | Intelligent control method and system for indoor environment and storage medium |
| CN110762799B (en) * | 2019-12-26 | 2020-05-29 | 星络智能科技有限公司 | Storage medium, intelligent panel and air conditioner control method based on intelligent panel |
| CN111142399A (en) * | 2020-01-09 | 2020-05-12 | 四川轻化工大学 | Embedded intelligent home automation control test system based on computer |
| CN111649443A (en) * | 2020-06-17 | 2020-09-11 | 宁波奥克斯电气股份有限公司 | Air conditioner control method and device, air conditioner and storage medium |
| CN111780376B (en) * | 2020-07-22 | 2022-06-14 | 海尔优家智能科技（北京）有限公司 | Method and device for starting air conditioner and air conditioner |
| CN114079585A (en) * | 2020-08-19 | 2022-02-22 | 佛山市顺德区美的电热电器制造有限公司 | Control method of household appliance, household appliance and storage medium |
| CN112432229B (en) * | 2020-11-16 | 2022-03-01 | 东南大学 | HVAC (heating ventilation and ventilation air conditioning) online monitoring system based on indoor humidity source information and control method |
| CN112460741B (en) * | 2020-11-23 | 2021-11-26 | 香港中文大学（深圳） | Control method of building heating, ventilation and air conditioning system |
| CN113706029A (en) * | 2021-08-31 | 2021-11-26 | 广东电网有限责任公司 | Self-adaptive management method, device and system for household energy |
| CN114777325B (en) * | 2022-04-11 | 2024-04-05 | 富联智能工坊(郑州)有限公司 | Boiler system regulation and control method, model building method, related equipment and medium |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH05126380A (en) * | 1991-11-05 | 1993-05-21 | Toshiba Corp | Air conditioning controller |
| CN102866684A (en) * | 2012-08-24 | 2013-01-09 | 清华大学 | Indoor environment integrated control system and method based on user comfort |
| CN104374053A (en) * | 2014-11-25 | 2015-02-25 | 珠海格力电器股份有限公司 | Intelligent control method, device and system |
| CN104598765A (en) * | 2015-02-16 | 2015-05-06 | 常州瑞信电子科技有限公司 | Building energy consumption prediction method based on elastic adaptive neural network |
| CN204853818U (en) * | 2015-06-26 | 2015-12-09 | 苏州中科院全周期绿色建筑研究院有限公司 | Interactive air conditioner environment comfort level detection device |
| CN105588290A (en) * | 2016-03-02 | 2016-05-18 | 广东工业大学 | Control terminal, system and method for adjusting temperature of air-conditioner |
-
2016
- 2016-08-23 CN CN201610709749.2A patent/CN106322656B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH05126380A (en) * | 1991-11-05 | 1993-05-21 | Toshiba Corp | Air conditioning controller |
| CN102866684A (en) * | 2012-08-24 | 2013-01-09 | 清华大学 | Indoor environment integrated control system and method based on user comfort |
| CN104374053A (en) * | 2014-11-25 | 2015-02-25 | 珠海格力电器股份有限公司 | Intelligent control method, device and system |
| CN104598765A (en) * | 2015-02-16 | 2015-05-06 | 常州瑞信电子科技有限公司 | Building energy consumption prediction method based on elastic adaptive neural network |
| CN204853818U (en) * | 2015-06-26 | 2015-12-09 | 苏州中科院全周期绿色建筑研究院有限公司 | Interactive air conditioner environment comfort level detection device |
| CN105588290A (en) * | 2016-03-02 | 2016-05-18 | 广东工业大学 | Control terminal, system and method for adjusting temperature of air-conditioner |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106322656A (en) | 2017-01-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106322656B (en) | A kind of air conditioning control method and server and air-conditioning system | |
| KR102553299B1 (en) | Data learning server and method for generating and using thereof | |
| CN107367016B (en) | A kind of air conditioner intelligent control method and its device, air-conditioning | |
| CN110836514B (en) | Control method and device of air conditioning unit | |
| CN104374053B (en) | Intelligent control method, device and system | |
| CN108361927A (en) | A kind of air-conditioner control method, device and air conditioner based on machine learning | |
| CN110726218B (en) | Air conditioner, control method and device thereof, storage medium and processor | |
| JP2020177344A (en) | Learning method, learning program and learning device | |
| CN105652677B (en) | A kind of intelligent home furnishing control method based on user behavior analysis, device and system | |
| CN108413588B (en) | Personalized air conditioner control system and method based on thermal imaging and BP neural network | |
| CN106369739A (en) | Air conditioner control method, air conditioner controller and air conditioner system | |
| JP6897767B2 (en) | Set value calculation system, method and program | |
| CN106322657A (en) | Air conditioner control method, air conditioner controller and air conditioner system | |
| CN110598906A (en) | Method and system for controlling energy consumption of superstores in real time based on deep reinforcement learning | |
| CN114662201B (en) | Optimizing method for intelligent regulation and control of natural ventilation | |
| CN110781595A (en) | Energy use efficiency PUE prediction method, device, terminal and medium | |
| CN110986249B (en) | Self-adjustment control method and system of air conditioner and air conditioner | |
| US20220019186A1 (en) | Method and system for smart environment management | |
| CN110738120A (en) | Environment data adjusting method and device, electronic equipment and medium | |
| CN112815487B (en) | Control method and device of humidifier, equipment and storage medium | |
| Aparicio-Ruiz et al. | KNN and adaptive comfort applied in decision making for HVAC systems | |
| JP7378497B2 (en) | Model sharing system, model management device, and air conditioner control device | |
| CN110726216B (en) | Air conditioner, control method, device and system thereof, storage medium and processor | |
| CN110779175A (en) | Operation mode control method and device | |
| Jin et al. | Prediction of Indoor Air Temperature Based on Deep Learning. |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information |
Address after: 266736 Hisense Road 1, South Village Town, Pingdu City, Qingdao, Shandong Applicant after: Hisense (Shandong) Air-conditioning Co., Ltd. Address before: 266100 Zhuzhou Road, Laoshan District, Qingdao, Shandong Province, No. 151 Applicant before: Hisense (Shandong) Air-conditioning Co., Ltd. |
|
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: 266100 Zhuzhou Road, Laoshan District, Qingdao, Shandong Province, No. 151 Applicant after: Hisense (Shandong) Air-conditioning Co., Ltd. Address before: 266736 Hisense Road 1, South Village Town, Pingdu City, Qingdao, Shandong Applicant before: Hisense (Shandong) Air-conditioning Co., Ltd. |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |