CN104374053B - Intelligent control method, device and system - Google Patents
Intelligent control method, device and system Download PDFInfo
- Publication number
- CN104374053B CN104374053B CN201410691067.4A CN201410691067A CN104374053B CN 104374053 B CN104374053 B CN 104374053B CN 201410691067 A CN201410691067 A CN 201410691067A CN 104374053 B CN104374053 B CN 104374053B
- Authority
- CN
- China
- Prior art keywords
- value
- comfort level
- running status
- sensor
- network model
- 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
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F2120/00—Control inputs relating to users or occupants
- F24F2120/10—Occupancy
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F11/00—Control or safety arrangements
- F24F11/30—Control or safety arrangements for purposes related to the operation of the system, e.g. for safety or monitoring
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F11/00—Control or safety arrangements
- F24F11/62—Control or safety arrangements characterised by the type of control or by internal processing, e.g. using fuzzy logic, adaptive control or estimation of values
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F2110/00—Control inputs relating to air properties
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F11/00—Control or safety arrangements
- F24F11/50—Control or safety arrangements characterised by user interfaces or communication
- F24F11/56—Remote control
- F24F11/58—Remote control using Internet communication
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F2120/00—Control inputs relating to users or occupants
- F24F2120/10—Occupancy
- F24F2120/14—Activity of occupants
Landscapes
- Engineering & Computer Science (AREA)
- Chemical & Material Sciences (AREA)
- Combustion & Propulsion (AREA)
- Mechanical Engineering (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Atmospheric Sciences (AREA)
- Physics & Mathematics (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Signal Processing (AREA)
- Air Conditioning Control Device (AREA)
Abstract
The invention provides an intelligent control method, device and system. A plurality of sensors are arranged indoors and outdoors to be used for collecting indoor environment information, human body status information and outdoor environment information, and multisource information constituted by the information collected by different types of the sensors is input a pre-established preset neural network model. Because data input into the preset neural network model come from an indoor space and an outdoor space and are used for environment measurement and human body status measurement, measurement of the ambient environment information and the human body status information can be achieved in a diversification mode through the sensors. The multisource information of the sensors can be fused, the ambient environment of an air conditioner and human body status can be accurately measured, running status includes the human body comfort degree, an air conditioning system is controlled according to the human body comfort degree, and therefore people-oriented control over the air conditioner is achieved, and it is guaranteed that indoor temperature, humidity and wind force are always kept under the state which is optimally suitable for human body dwelling.
Description
Technical field
The present invention relates to technical field of automation, more particularly, to a kind of intelligent control method, apparatus and system.
Background technology
At present, most of air-conditioning is mainly acquired to the humiture of living environment with single Temperature Humidity Sensor, and with
Single humiture, as major regulatory object, carries out control of lowering the temperature, carries out a liter temperature control when the temperature is low when temperature is higher
System, but single humidity temperature pickup can not accurately reflect the humiture situation of living environment, so can not control well
The humiture of living environment, generally requires user and through the multiple temperature adjusting air-conditioning, humidity and sweeps landscape condition, just can determine that ratio
Indoor environment conveniently.
For example:When user just returns from outdoor hot environment, air-conditioner temperature is adjusted to 20 degree, user after a period of time
Can feel colder, need to adjust air-conditioner temperature again to 25 degree, after a period of time may having been spent, can again air-conditioner temperature be adjusted again
To 27 degree.So airconditioning control of the prior art be main on the basis of temperature, there is no that people-oriented.
So needing a kind of method of the Based Intelligent Control indoor temperature and humidity that people-oriented, to ensure indoor temperature, humidity
And wind-force remains at the state of human body optimum inhabitation.
Content of the invention
The invention provides a kind of intelligent control method, apparatus and system, ensure that indoor temperature, humidity and wind-force begin
It is maintained at the state of human body optimum inhabitation eventually.
To achieve these goals, the invention provides following technological means:
A kind of intelligent control method, is applied to intelligence control system, and described system includes:It is arranged at indoor at least one
The sensor of collection indoor environment state and the sensor of at least one collection body state, are arranged at outdoor at least one and adopt
The sensor of collection outdoor environment state, the processor being connected with multiple sensors of indoor and outdoor, methods described includes:
Obtain multigroup sensing data of multiple sensor collections of described indoor and outdoor, corresponding one group of one of sensor
Sensing data;
Described multigroup sensing data is inputted default neural network model, described default neural network model is warp in advance
The training of least one set training sample, with the running status of air-conditioning for the model of output, wherein running status includes human comfort;
Export current running status after described default neural network model computing, adjust by described current running status
Whole air-conditioning system.
Preferably, described human comfort includes at least one comfort level.
Preferably, when described human comfort includes a comfort level, using this comfort level as currently relaxing
Appropriate grade, is included by current running status adjustment air-conditioning system:
Judge whether described current comfort level is more than predetermined level;
If described current comfort level is more than described predetermined level, maintain the current running status of air-conditioning;
If described current comfort level is less than described predetermined level, default with described by described current comfort level
The gap size of grade carries out corresponding to adjustment to described air-conditioning system.
Preferably, when described human comfort includes at least two comfort level, by described current running status
Adjustment air-conditioning system includes:
Obtain at least two comfort level in described current running status;
Obtain multiple basic confidence levels corresponding with each comfort level in different sample spaces respectively, one of
Sample space corresponds to one group of sensing data;
Corresponding at least two comfort level multiple basic confidence levels are carried out by Dempster-Shafer formula respectively
Merge, obtain at least two pooled functions, one of pooled function is corresponding with a comfort level;
When meeting pre-conditioned, obtain the functional value of at least two pooled functions, by the pooled function that functional value is maximum
Corresponding comfort level is as current comfort level;
Judge whether described current comfort level is more than predetermined level;
If described current comfort level is more than described predetermined level, maintain the current running status of air-conditioning;
If described current comfort level is less than described predetermined level, default with described by described current comfort level
The gap size of grade carries out corresponding to adjustment to described air-conditioning system.
Preferably, judge that meeting pre-conditioned process includes:
Solve belief function and the plausibility function of each comfort level respectively, wherein said belief function represents comfortable
Degree level results are genuine trusting degree, and described plausibility function represents the non-false trusting degree of comfort level result;
Using the difference between described plausibility function value and described belief function value as nondeterministic function value;
When described nondeterministic function value is less than preset value, and when the value of belief function is more than nondeterministic function value, judge full
Foot is pre-conditioned.
Preferably, the described running status of default neural network model also includes:
Optimum temperature value, optimal wet angle value and optimal wind-force value;
Preferably, include by described current running status adjustment air-conditioning system:
Regulate and control described air-conditioning system by the optimum temperature value in current running status, optimal wet angle value and optimal wind-force value
System.
Preferably, the building process of described default neural network model includes:
Set the neuronal quantity of input layer, hidden layer and output layer in neural network model, sensing data, described defeated
Enter initial weights between layer, described hidden layer and described output layer, judge initial threshold value and iterations at the end of training,
Wherein, the quantity of the input layer of described neural network model is consistent with the quantity of multiple sensors, output layer neuron
Quantity be four, respectively corresponding temperature value, humidity value, wind-force value and human comfort;
Obtain one group of training sample, described training sample includes the raw sample data of multiple sensors and original at this
The target operation state of air-conditioning under sample data, described target operation state includes optimum temperature value, optimal wet angle value, optimal wind
Force value and current human comfort under raw sample data;
By the raw sample data input neural network model of one group of training sample, through described input layer, described hidden layer
And exporting running status undetermined after the weighted calculation of described output layer, described running status undetermined includes temperature value undetermined, undetermined
Humidity value, wind-force value undetermined and human comfort undetermined;
If running status undetermined is less than threshold value with the error of target operation state, sample training terminates, and otherwise changes weights
With restart sample training after threshold value, until the error of running status undetermined and target operation state is less than threshold value or reaches repeatedly
Generation number;
The neural network model that current weights and threshold value are built is as described default neural network model.
Preferably, the plurality of sensing data includes:
It is arranged at the warm and humid angle value of at least one Temperature Humidity Sensor collection, at least one infrared temperature-test sensor of interior
The temperature value of collection;
It is arranged at the warm and humid angle value of at least one Temperature Humidity Sensor collection of outdoor and at least one intensity of illumination senses
The illumination intensity value of device collection.
A kind of intelligent controlling device, is applied to intelligence control system, and described system includes:It is arranged at indoor at least one
The sensor of collection indoor environment state and the sensor of at least one collection body state, are arranged at outdoor at least one and adopt
The sensor of collection outdoor environment state, the processor being connected with multiple sensors of indoor and outdoor, described device includes:
Obtain data cell, for obtaining multigroup sensing data of multiple sensor collections of described indoor and outdoor, wherein
One sensor corresponds to one group of sensing data;
Input block, for inputting default neural network model, described default nerve net by described multigroup sensing data
Network model is to train through least one set training sample in advance, with the running status of air-conditioning for the model of output;
Processing unit, for exporting current running status after described default neural network model computing, is worked as described
Front running status adjustment air-conditioning system, wherein said running status includes human comfort.
Preferably, described processing unit includes:
First processing units, during for judging that described human comfort includes a comfort level, judge described current
Whether comfort level is more than predetermined level;If described current comfort level is more than described predetermined level, maintain air-conditioning
Current running status;If described current comfort level is less than described predetermined level, by described current comfort level and
The error size of described predetermined level carries out corresponding to adjustment to described air-conditioning system;
Second processing unit, when including at least two comfort level for described running status, obtains described operation shape
At least two comfort level in state;Multiple bases corresponding with each comfort level are obtained respectively in different sample spaces
This confidence level, one of sample space corresponds to one group of sensing data;By corresponding at least two comfort level multiple bases
This confidence level is merged by Dempster-Shafer formula respectively, obtains at least two pooled functions, one of merging letter
Number is corresponding with a comfort level;When meeting pre-conditioned, obtain the functional value of at least two pooled functions, by functional value
The maximum corresponding comfort level of pooled function is as current comfort level;Judge whether described current comfort level is big
In predetermined level;If described current comfort level is more than described predetermined level, maintain the current running status of air-conditioning;If
Described current comfort level is less than described predetermined level, then press the error of described current comfort level and described predetermined level
Size carries out corresponding to adjustment to described air-conditioning system;
3rd processing unit, for when running status also includes optimum temperature value, optimal wet angle value and optimal wind-force value,
As the optimum temperature value in current running status, optimal wet angle value and air-conditioning system described in optimal wind-force value control.
Preferably, second processing unit includes:Condition judgment unit, for solving the reliability of each comfort level respectively
Function and plausibility function, wherein said belief function represents that comfort level result is genuine trusting degree, described likelihood degree
The non-false trusting degree of function representation comfort level result;By between described plausibility function value and described belief function value
Difference is as nondeterministic function value;When described nondeterministic function value is less than preset value, and the value of belief function is more than uncertain letter
During numerical value, judgement meets pre-conditioned.
Preferably, also include:
Build the construction unit of default neural network model;Described construction unit includes:
Initialization unit, for setting the neuronal quantity of input layer in neural network model, hidden layer and output layer, passes
Initial weights between sensor data, described input layer, described hidden layer and described output layer, initial at the end of judging training
Threshold value and iterations, wherein, the quantity one of the quantity of the input layer of described neural network model and multiple sensors
Cause, the quantity of output layer neuron is four, respectively corresponding temperature value, humidity value, wind-force value and human comfort;
Obtain sample unit, for obtaining one group of training sample, described training sample includes the original sample of multiple sensors
Notebook data and under this raw sample data air-conditioning target operation state, described target operation state includes optimum temperature
Value, optimal wet angle value, optimal wind-force value and current human comfort under raw sample data;
Integrated unit, for by the raw sample data input neural network model of one group of training sample, through described input
Running status undetermined is exported, described running status undetermined includes treating after the weighted calculation of layer, described hidden layer and described output layer
Constant temperature angle value, humidity value undetermined, wind-force value undetermined and human comfort undetermined;
Judging unit, if the error for running status undetermined and target operation state is less than threshold value, sample training is tied
Bundle, otherwise restarts sample training, until the error of running status undetermined and target operation state after modification weights and threshold value
Less than threshold value or reach iterations;
Complete unit, for the neural network model that builds current weights and threshold value as described default neutral net
Model.
A kind of intelligence control system, including:
Multiple sensors, the processor being connected with the plurality of sensor, the plurality of sensor includes being arranged at room
The interior sensor of at least one collection indoor environment state and the sensor of at least one collection body state, are arranged at outdoor
At least one collection outdoor environment state sensor;
Described processor, multigroup sensing data that the multiple sensors for obtaining described indoor and outdoor gather, wherein one
Individual sensor corresponds to one group of sensing data;Described multigroup sensing data is inputted default neural network model, described default
Neural network model is to train through least one set training sample in advance, with the running status of air-conditioning for the model of output;Through described
Current running status is exported after default neural network model computing, by described current running status adjustment air-conditioning system, its
Described in running status include human comfort.
Preferably, multiple sensor interfaces are also included, multiple sensor interfaces are used for connecting each sensor and processor.
The invention provides a kind of intelligent control method, the method is applied to intelligence control system, and the system is outer indoors
Deployed to ensure effective monitoring and control of illegal activities multiple sensors for gathering the environmental information of the environmental information, body state information and outdoor of interior, by difference
The multi-source information of the information composition of type sensor collection, the default neural network model that multi-source information input is built in advance,
The current running status of air-conditioning will be exported through model, using current running status as regulation and control air-conditioning system after Multi-source Information Fusion
The foundation of system.
Also have outdoor due to what the data of input to default neural network model not only had an interior, not only measuring environment
Also have measurement body state, so sensor of the invention be capable of equably, diversely measurement surrounding environment letter
For breath and body state information, with traditional single Temperature Humidity Sensor, the present invention can merge the multi-source of multiple sensors
Information is such that it is able to accurately measure air-conditioning surrounding environment and body state, and current running status includes human comfort,
With human comfort regulate and control air-conditioning system, thus realize people-oriented regulation and control air-conditioning purpose, thus ensureing indoor temperature, humidity
And wind-force remains at the state of human body optimum inhabitation.
Brief description
In order to be illustrated more clearly that the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing
Have technology description in required use accompanying drawing be briefly described it should be apparent that, drawings in the following description be only this
Some embodiments of invention, for those of ordinary skill in the art, on the premise of not paying creative work, acceptable
Other accompanying drawings are obtained according to these accompanying drawings.
Fig. 1 is a kind of structural representation of intelligence control system disclosed in the embodiment of the present invention;
Fig. 2 is a kind of flow chart of intelligent control method disclosed in the embodiment of the present invention;
Fig. 3 is the flow chart of the embodiment of the present invention another intelligent control method disclosed;
Fig. 4 is the flow chart of the embodiment of the present invention another intelligent control method disclosed;
Fig. 5 is the flow chart of the embodiment of the present invention another intelligent control method disclosed;
Fig. 6 is the flow chart of the embodiment of the present invention another intelligent control method disclosed;
Fig. 7 is the structural representation of the embodiment of the present invention another intelligence control system disclosed;
Fig. 8 is the schematic diagram of default neural network model in a kind of intelligent control method disclosed in the embodiment of the present invention;
Fig. 9 is the structural representation of the embodiment of the present invention another intelligence control system disclosed;
Figure 10 is a kind of structural representation of intelligent controlling device disclosed in the embodiment of the present invention;
Figure 11 is the structural representation of the embodiment of the present invention another intelligent controlling device disclosed;
Figure 12 is the structural representation of the embodiment of the present invention another intelligent controlling device disclosed.
Specific embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete
Site preparation description is it is clear that described embodiment is only a part of embodiment of the present invention, rather than whole embodiments.It is based on
Embodiment in the present invention, it is every other that those of ordinary skill in the art are obtained under the premise of not making creative work
Embodiment, broadly falls into the scope of protection of the invention.
As shown in figure 1, the invention provides a kind of intelligence control system, system includes:
It is arranged at the sensor 101 of at least one collection indoor environment state of interior, abbreviation sensor group A;
It is arranged at the sensor 102 of at least one collection body state of interior, abbreviation sensor group B;
It is arranged at the sensor 103 of at least one collection outdoor environment state of outdoor, abbreviation sensor group C;
The processor 100 being connected with multiple sensors of indoor and outdoor.
When implementing, the sensor of indoor and outdoor surroundingses can pass for temperature sensor, humidity sensor or humiture
Sensor, is mainly used to gather the humiture of indoor and outdoor, and outdoor in addition also have intensity of illumination sensor to be used for detecting the illumination of outdoor
Intensity, the sensor of collection body state includes infrared temperature-test sensor, for gathering the temperature of human body, additionally can have detection
The sensor of human body humidity and/or ccd sensor, are mainly used to gather the humidity of human body and/or the facial expression of human body, institute
The data that some sensors collect is referred to as multiple sensing datas.
All of sensor is all passed through sensor interface and is connected with processor 100, and processor 100 is in advance to multiple sensors
Data is pre-processed, such as AD conversion and filtering process, carries out follow-up process for convenience, typically also needs to multiple biographies
Sensor data is normalized.
As shown in Fig. 2 the present invention additionally provides a kind of intelligent control method, institute on the basis of above-mentioned intelligence control system
The method of stating includes:
Step S101:Obtain multigroup sensing data of multiple sensor collections of described indoor and outdoor, one of sensing
Device corresponds to one group of sensing data;
Multigroup sensing data includes:The sensing data of sign indoor environment state is, characterize the biography of outdoor environment state
Sensor data and the sensing data characterizing body state.
Include when implementing:The indoor warm and humid angle value of at least one Temperature Humidity Sensor collection, at least one is red
The temperature value of outer temperature transducer collection, the outdoor warm and humid angle value of at least one Temperature Humidity Sensor collection and at least one light
Illumination intensity value according to intensity sensor collection;When the sensor of the humidity being provided with detection human body and the biography of human body face expression
During sensor, multiple sensing datas also include human body humidity value and facial expression data.
Step S102:Described multigroup sensing data is inputted default neural network model, described default neutral net mould
Type is to train through least one set training sample in advance, with the running status of air-conditioning for the model of output;
Built default neural network model in advance before present invention execution, and using training sample to default neutral net
Model is trained so that default neural network model can export the current running status of air-conditioning according to sensing data.
Wherein, training sample includes training sensing data and the training running status under training sensing data, using instruction
Practice sample training neural network model so that neural network model can export air-conditioning after input training sensing data
Training running status.
Step S103:Export current running status after described default neural network model computing, by described current
Running status adjusts air-conditioning system, and wherein said running status includes human comfort.
When specifically used, multiple sensing datas of multiple sensor collections of indoor and outdoor are inputted to default nerve net
After default neural network model calculates, in network model, export the current running status of this model prediction, then processor
Adjust the parameters of air-conditioning system according to current running status, so that regulation and control air-conditioning reaches the more suitable operation shape of human body
State.
Running status includes human comfort, exports current human comfort after default neural network model computing,
Size using gap between current human comfort and default human comfort carries out corresponding adjustment to air-conditioning system.
The invention provides a kind of method of Based Intelligent Control temperature, the method is applied to intelligence control system, and the system exists
Indoor and outdoor has equably deployed to ensure effective monitoring and control of illegal activities multiple sensors for gathering the environment of the environmental information, body state information and outdoor of interior
Information, the multi-source information of the information composition that dissimilar sensor is gathered, the default god that multi-source information input is built in advance
Through network model, the current running status of air-conditioning will be exported after Multi-source Information Fusion through model, current running status is made
For regulating and controlling the foundation of air-conditioning system.
Also have outdoor due to what the data of input to default neural network model not only had an interior, not only measuring environment
Also have measurement body state, so sensor of the invention be capable of equably, diversely measurement surrounding environment letter
For breath and body state information, with traditional single Temperature Humidity Sensor, the present invention can merge the multi-source of multiple sensors
Information is such that it is able to accurately measure air-conditioning surrounding environment and body state, and running status includes human comfort, with human body
Comfort level regulate and control air-conditioning system, thus realize people-oriented regulation and control air-conditioning purpose, thus ensureing indoor temperature, humidity and wind-force
Remain at the state of human body optimum inhabitation.
Obtain human comfort in step S103 in above-mentioned Fig. 2 and include at least one comfort level, comfort level is used
In expression under multiple sensing datas of input, the comfortable degree of human body sensory, comfort level is higher, represents human comfort
Degree is better, and wherein, the total quantity of comfort level can be set according to situation by technical staff it is to be understood that setting
The total quantity of comfort level is more, and accuracy is higher.
The process that by described running status adjust air-conditioning system is described in detail below, as shown in figure 3, including:
Step S201:When described human comfort includes a comfort level, using this comfort level as current
Comfort level;
When only having a comfort level in human comfort, show that this neural network model is receiving multigroup sensing
After device data, it is only capable of exporting a comfort level according to multigroup sensing data, this comfort level can react air-conditioning
Current running status and the relation of human comfort, that is, current comfort level is higher, represents that human comfort is better, currently
Comfort level is lower, represents that human comfort is poorer.
Step S202:Judge whether described current comfort level is more than predetermined level;If being more than, enter step
S203, if less than entrance step S204;
Judge whether current comfort level is more than predetermined level, predetermined level is to preset one more comfortable one
Individual comfort level, if current comfort level is more than predetermined level, represents that the current running status of air-conditioning is good, need not
Adjustment, if current comfort level is less than predetermined level then it represents that the current running status of air-conditioning is poor, needs to be adjusted
Running status, so that the output of air-conditioning can reach the more more suitable state of human body.
Step S203:If described current comfort level is more than described predetermined level, maintain the current operation of air-conditioning
State;
Step S204:If described current comfort level is less than described predetermined level, by described current comfort level
With the gap size of described predetermined level, corresponding adjustment is carried out to described air-conditioning system.
When being adjusted, if the gap between current comfort level and predetermined level is bigger, represent air-conditioning
Now running status is poor, then carry out reversely larger adjustment to air-conditioning, if between current comfort level and predetermined level
Gap is less, represents air-conditioning running status elementary errors now, then air-conditioning is carried out with reversely less adjustment.
Obtain human comfort grade after default neural network model in the present invention and can be set as one or many
Individual, it act as predicting current running status due to default neural network model, prediction has inaccurate process, if setting
Human comfort only one of which grade in fixed running status, then because correct comfort level may be ignored by error, from
And produce error, at least two comfort level will be set in running status in order to reduce error, so that neural network model is first
First predict at least two comfort level, then carry out further computing again at least two comfort level again, from
And obtain accurate comfort level, to improve the accuracy of comfort level.
It is described in detail below in the process carrying out further computing at least two comfort level, as shown in figure 4, bag
Include:
Step S301:Obtain at least two comfort level in described current running status;
The comfort level obtaining after being that the data of all the sensors is merged in the corresponding method of Fig. 2, in order to
Convenient narration, the comfort level in current running status is referred to as S, including S1, S2 ... etc..
Step S302:Obtain corresponding with each comfort level multiple substantially credible respectively in different sample spaces
Degree, one of sample space corresponds to one group of sensing data;
In order to verify that whether comfort level S is correct, presses to each sensing data in multiple sensing datas below
The method of Fig. 2 carries out many experiments, and a sensing data corresponds to a sample space, obtains in a sample space Y1
Multiple comfort level M of many experiments output, including M1, M2 ... etc., the probability S1 in M is referred to as this sample empty
Between the basic confidence level for S1, by the probability that S2 occurs in M be referred to as in this sample space be directed to S2 basic confidence level, according to
Secondary analogize.
Basic confidence level is the probability of comfort level S in sample space, and probability is higher to represent that comfort level is more accurate.
In the same way, it is possible to obtain the basic confidence level in each sample space.
Obtain the comfort level in different sample spaces as stated above, Y1S1 represents comfort level in Y1 sample
The corresponding basic confidence level of S1, Y1S2 represents the corresponding basic confidence level of comfort level S2 in Y1 sample, the like, obtain
Obtain the basic confidence level of each comfort level in different sample spaces.
Step S303:Corresponding at least two comfort level multiple basic confidence levels are pressed Dempster- respectively
Shafer formula merges, and obtains at least two pooled functions, and one of pooled function is corresponding with a comfort level;
The basic reliability function of different sample spaces is merged, tries to achieve all samples for a comfort level
The combined value of this space middle grade.Taking comfort level S1 as a example, Y1S1, Y2S1, Y3S1 ... YNS1 is merged, obtain
Comfort level is the pooled function of S1, in the same manner, it is possible to obtain comfort level is the pooled function of S2, the like, obtain
The pooled function of each comfort level.
Step S304:When meeting pre-conditioned, obtain the functional value of at least two pooled functions, functional value is maximum
The corresponding comfort level of pooled function is as current comfort level.
The corresponding pooled function of each comfort level in step S303 is ranked up by size, by pooled function
It is worth greatly corresponding comfort level as current human's comfort level.The maximum corresponding comfort level of pooled function is in difference
Maximum probability in sample space, is this comfort level and disclosure satisfy that each sample space more.
Wherein, judge to meet pre-conditioned process, comprise the following steps as shown in Figure 5:
Step S401:Solve belief function and the plausibility function of each comfort level, wherein said reliability letter respectively
Number represents that comfort level result is genuine trusting degree, and described plausibility function represents the non-false trust of comfort level result
Degree;
Step S402:Using the difference between described plausibility function value and described belief function value as nondeterministic function
Value;
Step S403:When described nondeterministic function value is less than preset value, and the value of belief function is more than nondeterministic function value
When, judgement meets pre-conditioned.
Obtain qualitative probabilistic really for pooled function and must can confirm that this is comfortable more than uncertain probability
Degree grade is comfort level that can determine, satisfactory.
After obtaining current comfort level, execution adjusts air-conditioning system by described current running status, specifically
Implementation procedure consistent with the step shown in Fig. 3, will not be described here.In above-mentioned control process, processor is the place of air-conditioning system
Reason device, the processor of air-conditioning system directly gathers sensing data and is processed, additionally, processor can also be independent of sky
Adjusting system and the processor that exists, after processor obtains current running status, current running status are sent to air-conditioning
Controller, so that air-conditioner controller is controlled to itself according to current running status, no matter which kind of mode can be realized
The present invention.
Above-described embodiment is the process introducing to regulate and control air-conditioning system with human comfort, and another control is described below
Mode, the running status of default neural network model also includes:Optimum temperature value, optimal wet angle value and optimal wind-force value, that is, exist
Multiple sensing datas are inputted to after default neural network model, neural network model can be according to multiple sensing datas
Calculated, exported optimum temperature value, optimal wet angle value and the optimal wind-force value of now Neural Network model predictive, directly utilized
Optimum temperature value, optimal wet angle value and optimal wind-force value regulator control system can achieve that regulation and control air-conditioning is in the purpose of optimum state.
The building process of default neural network model in Fig. 2 step S102 is described in detail below, as shown in fig. 6, including following walking
Suddenly:
Step S501:Neural network model is initialized, initialized process includes setting in neural network model
The neuronal quantity of input layer, hidden layer and output layer, sensing data, described input layer, described hidden layer and described output
Initial weights between layer, judge initial threshold value and iterations at the end of training, wherein, described neural network model defeated
The quantity entering layer neuron is consistent with the quantity of multiple sensors, and the quantity of output layer neuron is four, respectively corresponding temperature
Value, humidity value, wind-force value and human comfort;
Before carrying out the present invention, obtain the model of air-conditioning;Sensing corresponding with this model is inquired about in presetting database
Device quantity;Using the initial data consistent with number of sensors as this air-conditioning training sample.Quantity and sky due to sensor
The refrigerating capacity adjusted corresponds, so after model determines, just can obtain should making of this product in presetting database
Number of sensors, after number of sensors determines, just can search consistent with number of sensors original in presetting database
Sample data, to carry out neural network computing for this product.
First neural network model is initialized, set the quantity of the neuron of input layer, input layer number and this
Multiple number of sensors used in invention are consistent, then set the quantity of output layer, because the present invention needs to export
Temperature value, humidity value, wind-force value and four parameters of human comfort, so output layer neuronal quantity is four, middle implicit
Depending on layer number can be according to the use algorithm of technical staff, reset the weights between each layer.
For example:Weights between input layer and hidden layer are A, then hidden layer=input layer * A, and the size of weights represents
Relation between two-layer, in the case that input is constant, can adjust mould by the weights in the default neural network model of adjustment
The output result of type.
Step S502:Obtain one group of training sample, described training sample include the raw sample data of multiple sensors with
And under this raw sample data air-conditioning target operation state, described target operation state includes optimum temperature value, optimal wet
Angle value, optimal wind-force value and current human comfort under raw sample data;
By can get training sample in step S501, training sample includes the operation shape of raw sample data and air-conditioning
State, wherein running status are to judge, according to Expert Rules, the running status that air-conditioning should have under raw sample data, wherein
Including the value of temperature, the value of humidity, the value of wind-force and human comfort.
Step S503:By the raw sample data input neural network model of one group of training sample, through described input layer, institute
Running status undetermined is exported, described running status undetermined includes temperature undetermined after stating hidden layer and the weighted calculation of described output layer
Value, humidity value undetermined, wind-force value undetermined and human comfort undetermined;
Preferably, before raw sample data inputs to neutral net, described raw sample data can be carried out pre-
Process and normalized, the process that described raw sample data is pre-processed includes AD conversion and filtering process, logarithm
According to carrying out after pretreatment and normalized, follow-up processing procedure can be significantly facilitated it is possible to improve accuracy.
Using raw sample data as neutral net input, default neural network model is by raw sample data through defeated
Enter layer, hidden layer and output layer, and be weighted merging between each layer, operation shape undetermined will be obtained after Multi-source Information Fusion
Four output results of state, temperature value respectively undetermined, humidity value undetermined, wind-force value undetermined and human comfort undetermined.
Step S504:Judge whether running status undetermined and the error of target operation state are less than threshold value, or current number of times
Reach iterations, as long as meeting one of condition then enter step S505, if two conditions are all unsatisfactory for, iterations
Step S503 is entered after plus one;
Step S505:Training terminates.
Wherein, judge that running status undetermined and the error of target operation state include less than threshold process:Treat described in judgement
Error between constant temperature angle value and described optimum temperature value is less than described threshold value;Judge that described humidity value undetermined is optimal with described
Error between humidity value is less than described threshold value;Judge that the error between described wind-force value undetermined and described optimal wind-force value is less than
Described threshold value.
If running status undetermined is less than described threshold value with the error of optimal operational condition, sample training terminates, and otherwise changes
Restart sample training, until the error of running status undetermined and target operation state is less than threshold value or reaches after weights and threshold value
To iterations;The neural network model of the current weights after training is terminated and threshold value structure is as described default nerve net
Network model.
If the error of running status undetermined and target operation state is less than threshold value, show error between the two permissible
In the range of reception, just can terminate training process, otherwise it is assumed that error is too big continuing neural network model to be trained, directly
To error in preset range when terminate to train, the purpose of training is to revise weights and threshold value so that running status undetermined and fortune
Error in preset range, after training certain number of times, between running status undetermined and running status for the error of row state
Also it is not up in preset range it was demonstrated that this time training process cannot reach error in preset range, so terminating this training.
After one group of training sample terminates, also include:Obtain the initial data of multigroup training sample;To described multigroup training
The initial data of sample presses the training method in Fig. 6 respectively, one by one described default neural network model is trained, to revise
Described weights and threshold value.
In order to ensure that presetting neural network model can be applied to different sensing datas it is possible to obtain multigroup again
Training sample, and using training sample, default neutral net is trained one by one, revise weights and threshold value so that being somebody's turn to do with continuous
Default neural network model can be more accurate, being capable of the corresponding running status of the polytype sensing data of Accurate Prediction.
The process that neural network model preset by above-mentioned structure is the process that multiple sensing datas carry out information fusion, shape
The default neural network model becoming can be in the case that input determines, the running status of prediction output.
The specific embodiment of the present invention is described below:
As shown in fig. 7, the general structure block diagram providing for the present invention.
The N number of sensor 104 of outer distribution indoors, is uniformly distributed some Temperature Humidity Sensors according to indoor design condition environment, uses
To detect indoor temperature and humidity conditions, it is uniformly distributed temperature and humidity sensing according to running state of air conditioner in air-conditioning internal machine and outer machine
Device, in order to detect running state of air conditioner;It is uniformly distributed Temperature Humidity Sensor and optical sensor according to outdoor situations in outdoor, use
In detection outdoor temperature and humidity conditions;Actual conditions according to people's life install some infrared temperature-test sensors indoors, use
To detect human body temperature situation.
Above-mentioned N number of sensor signal needs to be pre-processed and normalize, and wherein pretreatment includes A/D and changes, filters, puts
Big and reject the computings such as gross error, pretreated data is normalized again.
Normalization algorithm can adopt linear normalization algorithm, that is,:
Wherein x, y are the value before and after normalization, xmax、xminIn sample for all the sensors output
Maximum and minimum of a value.
Sample data after normalization carries out Algorithm Analysis and use processing by processor 100.After training
Intelligent algorithm can make different predictions to actually entering data, thus exporting the temperature of optimum, humidity and wind-force value, builds
The environment that optimum people lives.When processor 100 carries out decision data according to multiple sensor signals, can be in conjunction with expert system
The suggestion providing carries out decision-making.
, by testing collection, the neural network model after primary data sample training can for primary data sample data
Can be variant with practical application, therefore intelligent air condition can carry out secondary study according to practical operation situation, constantly calibrates nerve net
Network model parameter, the state making intelligent air condition not only be in predict but also learn, sustained improvement parameter, constantly need most life close to people
Environment.
As shown in figure 8, being BP neural network model in the present embodiment, training sample set is obtained by experiment, according to type
Coupling, it may be determined that the quantity of system sensor, so that it is determined that the number of input sample of sampling every time, that is, determines input layer
Number.
If training sample set is X=[X1,X2,…,Xk..., XM], a certain training sample:Xk=[xk1,xk2,…,xkN]T,
(k=1,2 ..., N), reality output is:Yk=[yk1,yk2,yk3,yk4]TIt is desirable to be output as dk=[dk1,dk2,dk3,dk4]T.If n
For iterations, weights and reality output are the functions of n.
Weights for neural network model and threshold value assign initial value, and initial value is it may be said that be less than 1 random number.Input sample number
According to note iterations is n=0, calculates input signal u and output signal v of BP every layer of neuron of network, if hidden layer and output
The excitation function of layer is respectively f1() and f2(), then neutral net be output as
The error signal of q-th neuron is ekq(n)=dkq(n)-ykq(n), wherein q=1,2,3,4.
Now choosing excitation function is S type function, that is,
According to error signal ekjN () revises the weights between hidden layer and output layer, on input layer any node with implicit
The weights of any node on layer.Until error is no longer iterated training after meeting zone of reasonableness.Between hidden layer and output layer
Modified weight amount be Δ wjp(n)==η δj(n)vj(n), the weights of any node on any node and hidden layer on input layer
Correction is Δ wmi(n)=- η δi(n)xkm(n).
Neutral net output layer has four nodes, i.e. Yk=[yk1,yk2,yk3,yk4]T, represent temperature, humidity, wind respectively
Power and human comfort.
It is the D-S evidence theory algorithm stream in the control strategy of intelligent air condition and the implementation method based on Multi-source Information Fusion
Cheng Tu, comprises the following steps that:
(1) to human comfort grade classification be some grades Si(i=1,2,3 ..., N), constitutes and assumes that space is S=
{S1,S2…,SN, wherein N is divided rank number.
(2) by aforementioned neurological training sample data, obtain human comfort grade S in certain sample spacei, obtain
Basic confidence level m of different sample spacesj(Si), that is, obtain mass function.Obtain its uncertain probability m simultaneouslyj(Ui).
(3) according to Dempster-Shafer composite formula, merge mass function mj(Si), that is,Its
(4) by above-mentioned merging mass function, belief function and plausibility function bel (S are solvedi) and pl (Si).Solution formula
As follows:
(5) judge the rule of human comfort:
1. uncertain probability mj(Ui) it is necessarily less than a certain threshold gamma;
2. belief function bel (Si) have to be larger than uncertain probability mj(Ui);
1. and 2. 3. meet under the conditions of, merge mass function mj(Si) in maximum be judged to current human's comfort level etc.
Level Si.
(6) according to judged result, Intelligent air conditioner control system could be made that corresponding adjustment.Adjustment grade is according to human comfort
Depending on grade.If human comfort is very poor, intelligent air-conditioning system will make larger tune to temperature, humidity, wind-force and wind direction
Whole;If conversely, human comfort, adjusts less very well or keeps current operating conditions.
The hardware connection diagram that Fig. 9 provides for the present invention.As shown in figure 9, the sensor interface 105 of processor is permissible
Extension, carry out n extension according to real sensor quantity and species, each sensor interface 105 can connect i sensor,
Sensor signal is after A/D, filtering and normalization scheduling algorithm are processed, then carries out intelligent algorithm calculating, draws optimum control side
Method, to control the operation of air-conditioning.
The number of sensors of whole intelligent air-conditioning system and species are different because actual application environment is different, therefore airconditioning control
Device hardware and software design needs to make corresponding adjustment, designs multiple sensor acquisition interface, and carries out A/D, filtering and normalizing
Change scheduling algorithm to process.In order to without loss of generality, air-conditioner controller peripheral interface circuit can carry out hardware expanding as needed, and
Software programming will have corresponding redundancy program.
As shown in Figure 10, present invention also offers a kind of intelligent controlling device, it is applied to the intelligence control system shown in Fig. 1
, including:
Obtain data cell 11, for obtaining multigroup sensing data of multiple sensor collections of described indoor and outdoor, its
In sensor correspond to one group of sensing data;
Input block 12, for inputting default neural network model, described default nerve by described multigroup sensing data
Network model is to train through least one set training sample in advance, with the running status of air-conditioning for the model of output;
Processing unit 13, for exporting current running status after described default neural network model computing, by described
Current running status adjustment air-conditioning system, wherein said running status includes human comfort.
As shown in figure 11, described processing unit 13 includes:
First processing units 131, during for judging that described human comfort includes a comfort level, judge described working as
Whether front comfort level is more than predetermined level;If described current comfort level is more than described predetermined level, maintain air-conditioning
Current running status;If described current comfort level is less than described predetermined level, by described current comfort level
With the error size of described predetermined level, corresponding adjustment is carried out to described air-conditioning system;
Second processing unit 132, when including at least two comfort level for described running status, obtains described operation
At least two comfort level in state;Obtain corresponding with each comfort level multiple respectively in different sample spaces
Basic confidence level, one of sample space corresponds to one group of sensing data;Will be corresponding at least two comfort level multiple
Basic confidence level is merged by Dempster-Shafer formula respectively, obtains at least two pooled functions, one of merging
Function is corresponding with a comfort level;When meeting pre-conditioned, obtain the functional value of at least two pooled functions, by function
The maximum corresponding comfort level of pooled function of value is as current comfort level;Whether judge described current comfort level
More than predetermined level;If described current comfort level is more than described predetermined level, maintain the current operation shape of air-conditioning
State;If described current comfort level is less than described predetermined level, by described current comfort level and described predetermined level
Error size described air-conditioning system is carried out corresponding to adjustment;
3rd processing unit 133, for also including optimum temperature value, optimal wet angle value and optimal wind-force value in running status
When, as the optimum temperature value in current running status, optimal wet angle value and air-conditioning system described in optimal wind-force value control.
Wherein, second processing unit 132 includes:Condition judgment unit, for solving the letter of each comfort level respectively
Degree function and plausibility function, wherein said belief function represents that comfort level result is genuine trusting degree, described likelihood
The non-false trusting degree of degree function representation comfort level result;By between described plausibility function value and described belief function value
Difference as nondeterministic function value;When described nondeterministic function value is less than preset value, and the value of belief function is more than uncertain
During functional value, judgement meets pre-conditioned.
As shown in figure 12, a kind of intelligent controlling device of the present invention also offer also includes building default neural network model
Construction unit 14;Described construction unit 14 includes:
Initialization unit 21, for setting the neuronal quantity of input layer in neural network model, hidden layer and output layer,
Initial weights between sensing data, described input layer, described hidden layer and described output layer, initial at the end of judging training
Threshold value and iterations, wherein, the quantity of the quantity of the input layer of described neural network model and multiple sensors
Unanimously, the quantity of output layer neuron is four, respectively corresponding temperature value, humidity value, wind-force value and human comfort;
Obtain sample unit 22, for obtaining one group of training sample, described training sample includes the original of multiple sensors
Sample data and under this raw sample data air-conditioning target operation state, described target operation state includes optimum temperature
Value, optimal wet angle value, optimal wind-force value and current human comfort under raw sample data;
Integrated unit 23, for by the raw sample data input neural network model of one group of training sample, through described defeated
Running status undetermined is exported, described running status undetermined includes after the weighted calculation entering layer, described hidden layer and described output layer
Temperature value undetermined, humidity value undetermined, wind-force value undetermined and human comfort undetermined;
Judging unit 24, if the error for running status undetermined and target operation state is less than threshold value, sample training is tied
Bundle, otherwise restarts sample training, until the error of running status undetermined and target operation state after modification weights and threshold value
Less than threshold value or reach iterations;
Complete unit 25, for the neural network model that builds current weights and threshold value as described default nerve net
Network model.
As shown in figure 1, present invention also offers a kind of intelligence control system, including:
Multiple sensors 101 (102,103), the processor 100 being connected with the plurality of sensor, the plurality of sensor
Sensor including at least one the collection indoor environment state being arranged at interior and the biography of at least one collection body state
Sensor, is arranged at the sensor of at least one collection outdoor environment state of outdoor;
Described processor 100, for obtaining multigroup sensing data of multiple sensor collections of described indoor and outdoor, wherein
One sensor corresponds to one group of sensing data;Described multigroup sensing data is inputted default neural network model, described pre-
If neural network model is to train through least one set training sample in advance, with the running status of air-conditioning for the model of output;Through institute
Current running status is exported after stating default neural network model computing, by described current running status adjustment air-conditioning system,
Wherein said running status includes human comfort.
Wherein, it is connected by sensor interface between the plurality of sensor and described processor.Described system also includes
Multiple sensor interfaces, each sensor interface is connected with multiple sensors, for transmitting the sensor of multiple sensor collections
Data.
If the function described in the present embodiment method is realized and as independent product pin using in the form of SFU software functional unit
When selling or using, can be stored in a computing device read/write memory medium.Based on such understanding, the embodiment of the present invention
Partly being embodied in the form of software product of part that prior art is contributed or this technical scheme, this is soft
Part product is stored in a storage medium, including some instructions with so that computing device (can be personal computer,
Server, mobile computing device or network equipment etc.) execution each embodiment methods described of the present invention all or part step
Suddenly.And aforesaid storage medium includes:USB flash disk, portable hard drive, read-only storage (ROM, Read-Only Memory), deposit at random
Access to memory (RAM, Random Access Memory), magnetic disc or CD etc. are various can be with the medium of store program codes.
In this specification, each embodiment is described by the way of going forward one by one, and what each embodiment stressed is and other
The difference of embodiment, between each embodiment same or similar partly mutually referring to.
Described above to the disclosed embodiments, makes professional and technical personnel in the field be capable of or uses the present invention.
Multiple modifications to these embodiments will be apparent from for those skilled in the art, as defined herein
General Principle can be realized without departing from the spirit or scope of the present invention in other embodiments.Therefore, this
Bright be not intended to be limited to the embodiments shown herein, and be to fit to and principles disclosed herein and features of novelty phase
Consistent scope the widest.
Claims (12)
1. it is characterised in that being applied to intelligence control system, described system includes a kind of intelligent control method:It is arranged at interior
At least one collection sensor of indoor environment state and at least one collection body state sensor, be arranged at outdoor
The sensor of at least one collection outdoor environment state, the processor being connected with multiple sensors of indoor and outdoor, methods described bag
Include:
Obtain multigroup sensing data of multiple sensor collections of described indoor and outdoor, the corresponding one group of sensing of one of sensor
Device data;
Described multigroup sensing data is inputted default neural network model, described default neural network model is in advance through at least
The training of one group of training sample, with the running status of air-conditioning for the model of output, wherein running status includes human comfort;
Export current running status after described default neural network model computing, empty by described current running status adjustment
Adjusting system;
When described human comfort includes at least two comfort level, by described current running status adjustment air-conditioning system
Including:
Obtain at least two comfort level in described current running status;
Multiple basic confidence levels corresponding with each comfort level, one of sample is obtained respectively in different sample spaces
Space corresponds to one group of sensing data;
Corresponding at least two comfort level multiple basic confidence levels are closed by Dempster-Shafer formula respectively
And, obtain at least two pooled functions, one of pooled function is corresponding with a comfort level;
When meeting pre-conditioned, obtain the functional value of at least two pooled functions, the maximum pooled function of functional value is corresponded to
Comfort level as current comfort level;
Judge whether described current comfort level is more than predetermined level;
If described current comfort level is more than described predetermined level, maintain the current running status of air-conditioning;
If described current comfort level is less than described predetermined level, by described current comfort level and described predetermined level
Gap size described air-conditioning system is carried out corresponding to adjustment.
2. the method for claim 1 is it is characterised in that judge that meeting pre-conditioned process includes:
Solve belief function and the plausibility function of each comfort level respectively, wherein said belief function represents comfort level etc.
Level result is genuine trusting degree, and described plausibility function represents the non-false trusting degree of comfort level result;
Using the difference between described plausibility function value and described belief function value as nondeterministic function value;
When described nondeterministic function value is less than preset value, and when the value of belief function is more than nondeterministic function value, judge to meet in advance
If condition.
3. the method for claim 1 is it is characterised in that the described running status of default neural network model also includes:
Optimum temperature value, optimal wet angle value and optimal wind-force value.
4. method as claimed in claim 3 is it is characterised in that included by described current running status adjustment air-conditioning system:
Regulate and control described air-conditioning system by the optimum temperature value in current running status, optimal wet angle value and optimal wind-force value.
5. method as claimed in claim 3 is it is characterised in that the building process of described default neural network model includes:
Set the neuronal quantity of input layer, hidden layer and output layer in neural network model, sensing data, described input
Initial weights between layer, described hidden layer and described output layer, judge initial threshold value and iterations at the end of training, its
In, the quantity of the input layer of described neural network model is consistent with the quantity of multiple sensors, output layer neuron
Quantity is four, respectively corresponding temperature value, humidity value, wind-force value and human comfort;
Obtain one group of training sample, described training sample includes the raw sample data of multiple sensors and in this original sample
The target operation state of air-conditioning under data, described target operation state includes optimum temperature value, optimal wet angle value, optimal wind-force value
With current human comfort under raw sample data;
By the raw sample data input neural network model of one group of training sample, through described input layer, described hidden layer and institute
Running status undetermined is exported, described running status undetermined includes temperature value undetermined, humidity undetermined after the weighted calculation stating output layer
Value, wind-force value undetermined and human comfort undetermined;
If running status undetermined is less than threshold value with the error of target operation state, sample training terminates, otherwise modification weights and threshold
Restart sample training, until the error of running status undetermined and target operation state is less than threshold value or reaches iteration time after value
Number;
The neural network model that current weights and threshold value are built is as described default neural network model.
6. the method for claim 1 is it is characterised in that the plurality of sensing data includes:
It is arranged at the warm and humid angle value of at least one Temperature Humidity Sensor collection, the collection of at least one infrared temperature-test sensor of interior
Temperature value;
It is arranged at the warm and humid angle value of at least one Temperature Humidity Sensor collection of outdoor and at least one intensity of illumination sensor is adopted
The illumination intensity value of collection.
7. it is characterised in that being applied to intelligence control system, described system includes a kind of intelligent controlling device:It is arranged at interior
At least one collection sensor of indoor environment state and at least one collection body state sensor, be arranged at outdoor
The sensor of at least one collection outdoor environment state, the processor being connected with multiple sensors of indoor and outdoor, described device bag
Include:
Obtain data cell, for obtaining multigroup sensing data of multiple sensor collections of described indoor and outdoor, one of
Sensor corresponds to one group of sensing data;
Input block, for inputting default neural network model, described default neutral net mould by described multigroup sensing data
Type is to train through least one set training sample in advance, with the running status of air-conditioning for the model of output;
Processing unit, for exporting current running status after described default neural network model computing, by described current
Running status adjusts air-conditioning system, and wherein said running status includes human comfort;
Described processing unit includes:Second processing unit, when including two comfort level for described running status, obtains institute
State at least two comfort level in running status;Obtain corresponding with each comfort level respectively in different sample spaces
Multiple basic confidence level, one of sample space corresponds to one group of sensing data;At least two comfort level are corresponded to
Multiple basic confidence level merge by Dempster-Shafer formula respectively, obtain at least two pooled functions, wherein one
Individual pooled function is corresponding with a comfort level;When meeting pre-conditioned, obtain the functional value of at least two pooled functions,
Using the maximum corresponding comfort level of pooled function of functional value as current comfort level;Judge described current comfort level etc.
Whether level is more than predetermined level;If described current comfort level is more than described predetermined level, maintain the current fortune of air-conditioning
Row state;If described current comfort level is less than described predetermined level, default with described by described current comfort level
The error size of grade carries out corresponding to adjustment to described air-conditioning system.
8. device as claimed in claim 7 is it is characterised in that described processing unit also includes:
3rd processing unit, for when running status also includes optimum temperature value, optimal wet angle value and optimal wind-force value, by working as
Optimum temperature value in front running status, optimal wet angle value and air-conditioning system described in optimal wind-force value control.
9. device as claimed in claim 8 is it is characterised in that second processing unit includes:Condition judgment unit, for respectively
Solve belief function and the plausibility function of each comfort level, wherein said belief function represents that comfort level result is
Genuine trusting degree, described plausibility function represents the non-false trusting degree of comfort level result;By described plausibility function
Difference between value and described belief function value is as nondeterministic function value;When described nondeterministic function value is less than preset value, and
When the value of belief function is more than nondeterministic function value, judgement meets pre-conditioned.
10. device as claimed in claim 8 is it is characterised in that also include:
Build the construction unit of default neural network model;Described construction unit includes:
Initialization unit, for setting the neuronal quantity of input layer in neural network model, hidden layer and output layer, sensor
Initial weights between data, described input layer, described hidden layer and described output layer, initial threshold value at the end of judging training
And iterations, wherein, the quantity of the input layer of described neural network model is consistent with the quantity of multiple sensors, defeated
The quantity going out layer neuron is four, respectively corresponding temperature value, humidity value, wind-force value and human comfort;
Obtain sample unit, for obtaining one group of training sample, described training sample includes the original sample number of multiple sensors
The target operation state of air-conditioning according to this and under this raw sample data, described target operation state includes optimum temperature value,
Good humidity value, optimal wind-force value and current human comfort under raw sample data;
Integrated unit, for by the raw sample data input neural network model of one group of training sample, through described input layer, institute
Running status undetermined is exported, described running status undetermined includes temperature undetermined after stating hidden layer and the weighted calculation of described output layer
Value, humidity value undetermined, wind-force value undetermined and human comfort undetermined;
Judging unit, if the error for running status undetermined and target operation state is less than threshold value, sample training terminates, no
Then restart sample training after modification weights and threshold value, until the error of running status undetermined and target operation state is less than threshold
Value or reach iterations;
Complete unit, for the neural network model that builds current weights and threshold value as described default neutral net mould
Type.
A kind of 11. intelligence control systems are it is characterised in that include:
Multiple sensors, the processor being connected with the plurality of sensor, the plurality of sensor includes being arranged at interior
The sensor of at least one collection indoor environment state and the sensor of at least one collection body state, are arranged at outdoor extremely
The sensor of a few collection outdoor environment state;
Described processor, for obtaining multigroup sensing data of multiple sensor collections of described indoor and outdoor, one of biography
Sensor corresponds to one group of sensing data;Described multigroup sensing data is inputted default neural network model, described default nerve
Network model is to train through least one set training sample in advance, with the running status of air-conditioning for the model of output;Through described default
Current running status is exported, by described current running status adjustment air-conditioning system, wherein institute after neural network model computing
State running status and include human comfort;When described human comfort includes at least two comfort level, by described current
Running status adjustment air-conditioning system include:
Obtain at least two comfort level in described current running status;
Multiple basic confidence levels corresponding with each comfort level, one of sample is obtained respectively in different sample spaces
Space corresponds to one group of sensing data;
Corresponding at least two comfort level multiple basic confidence levels are closed by Dempster-Shafer formula respectively
And, obtain at least two pooled functions, one of pooled function is corresponding with a comfort level;
When meeting pre-conditioned, obtain the functional value of at least two pooled functions, the maximum pooled function of functional value is corresponded to
Comfort level as current comfort level;
Judge whether described current comfort level is more than predetermined level;
If described current comfort level is more than described predetermined level, maintain the current running status of air-conditioning;
If described current comfort level is less than described predetermined level, by described current comfort level and described predetermined level
Gap size described air-conditioning system is carried out corresponding to adjustment.
12. systems as claimed in claim 11 it is characterised in that also including multiple sensor interfaces, multiple sensor interfaces
For connecting each sensor and processor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410691067.4A CN104374053B (en) | 2014-11-25 | 2014-11-25 | Intelligent control method, device and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201410691067.4A CN104374053B (en) | 2014-11-25 | 2014-11-25 | Intelligent control method, device and system |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104374053A CN104374053A (en) | 2015-02-25 |
CN104374053B true CN104374053B (en) | 2017-02-22 |
Family
ID=52553119
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201410691067.4A Active CN104374053B (en) | 2014-11-25 | 2014-11-25 | Intelligent control method, device and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104374053B (en) |
Families Citing this family (43)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104833063B (en) * | 2015-06-04 | 2017-12-01 | 安徽建筑大学 | Air conditioner control method and system |
CN105588274B (en) * | 2015-11-06 | 2018-09-25 | 青岛海信日立空调系统有限公司 | A kind of natural wind control method and device |
CN105757882A (en) * | 2016-03-01 | 2016-07-13 | 中国建筑科学研究院 | Comprehensive control method and device for thermal comfort indoor environment |
CN106322656B (en) * | 2016-08-23 | 2019-05-14 | 海信(山东)空调有限公司 | A kind of air conditioning control method and server and air-conditioning system |
CN106770989B (en) * | 2017-03-15 | 2019-05-10 | 江西佳信房地产资产评估有限公司 | A kind of effective indoor air quality assessment system |
CN108235722B (en) * | 2017-05-22 | 2020-10-09 | 深圳微自然创新科技有限公司 | Comfort zone determination method and device |
CN108139101B (en) * | 2017-05-22 | 2020-10-30 | 深圳微自然创新科技有限公司 | Method for determining indoor environment comfort zone and terminal equipment |
CN109073253B (en) * | 2017-05-22 | 2020-06-30 | 深圳微自然创新科技有限公司 | Environment adjusting method and server |
CN107477772A (en) * | 2017-07-21 | 2017-12-15 | 天津大学 | House VMC control method based on indoor monitoring data |
CN107390753B (en) * | 2017-08-29 | 2019-07-30 | 贵州省岚林阳环保能源科技有限责任公司 | Intelligent plant growth environment regulating device and method based on Internet of Things cloud platform |
CN107390754B (en) * | 2017-08-29 | 2019-07-30 | 贵州省岚林阳环保能源科技有限责任公司 | Intelligent plant growth environment adjustment system and method based on Internet of Things cloud platform |
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 |
CN107826027B (en) * | 2017-09-21 | 2019-09-03 | 山东大学 | Refrigerator car temprature control method and system based on big data analysis |
CN107833573B (en) * | 2017-10-23 | 2021-02-09 | 上海百芝龙网络科技有限公司 | Machine learning-based family scene semantic understanding auxiliary method |
CN108181837B (en) * | 2017-11-28 | 2020-11-27 | 珠海格力电器股份有限公司 | Control method and control device |
CN108036473B (en) * | 2017-11-30 | 2020-05-05 | 广东芬尼克兹节能设备有限公司 | Intelligent temperature and humidity control method and device |
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 |
CN108931027A (en) * | 2018-08-01 | 2018-12-04 | 北京创思工贸有限公司 | A kind of visual remote monitoring control maintenance air-conditioning system |
CN109028402A (en) * | 2018-08-04 | 2018-12-18 | 浙江国芝科技有限公司 | The water-supply control of radiation air-conditioner, device and system |
CN110822616A (en) * | 2018-08-13 | 2020-02-21 | 格力电器(武汉)有限公司 | Automatic air conditioner adjusting method and device |
CN110837229B (en) * | 2018-08-17 | 2021-06-29 | 珠海格力电器股份有限公司 | Control method and device for household appliance |
CN109140665A (en) * | 2018-08-31 | 2019-01-04 | 珠海格力电器股份有限公司 | Control the method, apparatus and air-conditioning device of air-conditioning |
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 |
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 |
CN109357352B (en) * | 2018-11-05 | 2021-01-26 | 四川长虹电器股份有限公司 | Air conditioner intelligent temperature control system and method based on decision tree classification algorithm |
CN109376795A (en) * | 2018-11-19 | 2019-02-22 | 四川长虹电器股份有限公司 | Air conditioner intelligent temperature control method based on Decision Tree Algorithm |
CN109489212B (en) * | 2018-11-21 | 2020-05-05 | 珠海格力电器股份有限公司 | Intelligent sleep control method, adjustment system and equipment for air conditioner |
CN109695943B (en) * | 2018-11-29 | 2021-04-06 | 中国汽车工业工程有限公司 | Temperature and humidity control system of coating air conditioner based on big data deep learning |
CN109695944B (en) * | 2018-11-29 | 2021-06-18 | 中国汽车工业工程有限公司 | Control method for coating fresh air conditioner based on multi-model deep learning |
CN110230866A (en) * | 2019-06-11 | 2019-09-13 | 陕西大疆电力科技有限公司 | One kind being used for indoor temperature and humidity intelligent control system |
TWI699637B (en) * | 2019-07-10 | 2020-07-21 | 中華電信股份有限公司 | System for controlling environmental comfort using deep reinforcement learning and method thereof |
CN110395286B (en) * | 2019-08-02 | 2020-08-07 | 中南大学 | Method and system for monitoring air quality and regulating ventilation in train |
CN110738120B (en) * | 2019-09-17 | 2024-02-09 | 平安科技(深圳)有限公司 | Environment data adjustment method and device, electronic equipment and medium |
CN110726220A (en) * | 2019-10-29 | 2020-01-24 | 珠海格力电器股份有限公司 | Control method and device of air conditioner |
CN110726214A (en) * | 2019-10-29 | 2020-01-24 | 珠海格力电器股份有限公司 | Method and device for controlling air conditioner |
CN112050448A (en) * | 2020-08-18 | 2020-12-08 | 吉安职业技术学院 | Indoor temperature and humidity adjusting system of low-voltage apparatus controlled by PLC circuit |
CN112484259B (en) * | 2020-12-04 | 2021-11-16 | 珠海格力电器股份有限公司 | Humidity control method and device, electronic equipment and storage medium |
CN112506059B (en) * | 2020-12-07 | 2022-09-16 | 常州常工电子科技股份有限公司 | Classroom self-adaptive control system based on energy-saving model of personal sensory comfort |
CN112947637B (en) * | 2021-01-29 | 2022-06-24 | 阜阳小欧智能科技有限公司 | Office environment intelligent regulation system |
CN113311892B (en) * | 2021-05-27 | 2022-05-10 | 中通服咨询设计研究院有限公司 | Optimal pre-cooling control method for energy efficiency of central air conditioner of exhibition venue |
CN113327219B (en) * | 2021-06-21 | 2022-01-28 | 易成功(厦门)信息科技有限公司 | Image processing method and system based on multi-source data fusion |
CN114370696B (en) * | 2021-12-27 | 2023-04-21 | 浙江英集动力科技有限公司 | D-S evidence theory-based central air conditioner cooling tower water outlet temperature control method |
CN114265458A (en) * | 2022-01-05 | 2022-04-01 | 信息产业电子第十一设计研究院科技工程股份有限公司 | Method for realizing accurate control of air pipe temperature and humidity and room temperature and humidity |
-
2014
- 2014-11-25 CN CN201410691067.4A patent/CN104374053B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN104374053A (en) | 2015-02-25 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104374053B (en) | Intelligent control method, device and system | |
CN106322656B (en) | A kind of air conditioning control method and server and air-conditioning system | |
Jazizadeh et al. | User-led decentralized thermal comfort driven HVAC operations for improved efficiency in office buildings | |
CN103912966B (en) | A kind of earth source heat pump refrigeration system optimal control method | |
KR102380397B1 (en) | METHOD FOR MANAGING SMART BUILDING USING IoT SENSOR AND ARTIFICIAL INTELIGENCE | |
Xikai et al. | Comparison of regression models for estimation of carbon emissions during building's lifecycle using designing factors: a case study of residential buildings in Tianjin, China | |
CN108413588B (en) | Personalized air conditioner control system and method based on thermal imaging and BP neural network | |
CN100501608C (en) | Posterior control and multi-sensor infusion method and central conditioner energy-saving control device | |
CN107085732B (en) | Cowshed environment ammonia intelligent monitoring system based on wireless sensor network | |
CN111639462B (en) | Building indoor thermal comfort prediction method in natural ventilation environment based on deep belief neural network | |
CN109405195A (en) | Air conditioner intelligent control system and method | |
CN105190191A (en) | Energy saving heating, ventilation, air conditioning control system | |
CN105698861A (en) | Indoor environment comfort level evaluation method | |
CN106482502A (en) | The intelligence that is recommended based on cloud platform big data dries long-range control method and system | |
CN105320184A (en) | Intelligent monitoring system of indoor environment of building | |
CN111103220A (en) | Method and system for predicting and regulating concentration of atmospheric pollutants | |
CN114662201B (en) | Optimizing method for intelligent regulation and control of natural ventilation | |
CN113240184B (en) | Building space unit cold load prediction method and system based on federal learning | |
CN111461413A (en) | Highway road surface performance detecting system | |
CN106499656A (en) | A kind of fan wind speed intelligent control method | |
CN111444947A (en) | Bridge settlement intelligent detection system | |
CN113418229A (en) | Heating management and control valve intelligent regulation and control system based on convolution-self-coding neural network | |
CN109435630A (en) | A kind of crew module's temprature control method based on artificial neural network algorithm | |
CN111625030A (en) | Greenhouse environment control method, device, equipment, system and storage medium | |
CN112594906A (en) | Intelligent identification comfortable air conditioning system with memory function according to different persons and operation method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |