AU2021105951A4 - Method and System for Adjusting Indoor Environment Comfort Based on Deep Learning - Google Patents
Method and System for Adjusting Indoor Environment Comfort Based on Deep Learning Download PDFInfo
- Publication number
- AU2021105951A4 AU2021105951A4 AU2021105951A AU2021105951A AU2021105951A4 AU 2021105951 A4 AU2021105951 A4 AU 2021105951A4 AU 2021105951 A AU2021105951 A AU 2021105951A AU 2021105951 A AU2021105951 A AU 2021105951A AU 2021105951 A4 AU2021105951 A4 AU 2021105951A4
- Authority
- AU
- Australia
- Prior art keywords
- data
- indoor environment
- indoor
- module
- environmental
- 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
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/0265—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric the criterion being a learning criterion
-
- 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
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/0205—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric not using a model or a simulator of the controlled system
- G05B13/026—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric not using a model or a simulator of the controlled system using a predictor
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L67/00—Network arrangements or protocols for supporting network services or applications
- H04L67/01—Protocols
- H04L67/12—Protocols specially adapted for proprietary or special-purpose networking environments, e.g. medical networks, sensor networks, networks in vehicles or remote metering networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/33—Services specially adapted for particular environments, situations or purposes for indoor environments, e.g. buildings
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/38—Services specially adapted for particular environments, situations or purposes for collecting sensor information
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F2110/00—Control inputs relating to air properties
- F24F2110/10—Temperature
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F2110/00—Control inputs relating to air properties
- F24F2110/20—Humidity
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F2110/00—Control inputs relating to air properties
- F24F2110/40—Pressure, e.g. wind pressure
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D23/00—Control of temperature
- G05D23/19—Control of temperature characterised by the use of electric means
- G05D23/1917—Control of temperature characterised by the use of electric means using digital means
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Abstract
The invention relates to a method and a system for adjusting indoor environmental
comfort degree based on deep learning, which comprise the following steps: Si, collecting
environmental data by using sensor equipment in a data collection module, and sending
environmental state data to an edge preprocessing and control module; S2, building a model
based on deep learning LSTM neural network to predict the change of indoor environment in
a certain period of time in the future; S3, the multi-source heterogeneous environment data
uploaded by the sensor at the current time is preprocessed and used as the input of the indoor
environment prediction model; S4, the indoor environment prediction model completes the
prediction of indoor multi-source heterogeneous environment data changes according to the
uploaded indoor multi-source heterogeneous environment data, and sends the results to the
thermal comfort control model based on different mode feature constraints in the next step; S5,
constructing a thermal comfort control model based on different mode feature constraints,
which is used for receiving the results of the indoor environment prediction model in the
previous step, setting the initial environmental parameters and the adjustment range of
environmental parameters according to different scene modes, and proposing an indoor
environment adjustment scheme for improving human comfort under different scene mode
constraints. According to the invention, the dynamic prediction of indoor environment data is
realized in the Internet of Things scene by adopting a deep learning algorithm, so that the
energy consumption is effectively reduced while providing a comfortable indoor environment
for users.
2/2
Indoor historical multi
dimensional
environmental dataset
Data cleaning and noise reduction
LSTM indoor environment
prediction model#
Comfort Setting of
Enviro e Data d e indoor Environme Edge Model of evrne
ntaldata cleaning Edge Harmoenvironen
collection and noise gateway, Room t change
reduction t Environment rangebased asedon
Mode on system
mode
characterist
Datacenter Related ics.
equipment.
Figure 2
Description
2/2
Indoor historical multi dimensional environmental dataset
Data cleaning and noise reduction
LSTM indoor environment prediction model#
Comfort Setting of Enviro e Data Environme d e Edge indoor Model of evrne ntaldata cleaning Edge Harmoenvironen collection and noise gateway, Room t change reduction t Environment rangebased asedon Mode on system mode characterist Datacenter Related ics. equipment.
Figure 2
Method and System for Adjusting Indoor Environment Comfort Based on
Deep Learning
The invention relates to the field of indoor environment control, in particular to
an indoor environment comfort degree adjusting method and system based on deep
learning.
At present, most families or office areas adopt a single constant temperature
control indoor temperature adjustment method, which only improves human comfort
by adjusting indoor temperature, and does not consider other controllable
environmental factors affecting indoor comfort; moreover, the temperature control is
accomplished through the setting of air conditioning system by people's subjective
consciousness. In addition, the temperature change in indoor environment is nonlinear
and non-real-time, so this control method can not meet people's real needs for indoor
environment comfort, and the repeated operation process will make people's indoor
environment experience worse. Therefore, it has realistic demand and application
value to establish a reasonable control strategy for indoor thermal comfort.
In order to avoid meaningless guidance and adjustment due to shortcomings of
indicators in indoor environment comfort control, this paper comprehensively
considers various environmental factors that affect human comfort, establishes an
environmental comfort prediction model based on deep learning, and further
constructs an environmental comfort control model according to the predicted values
of the model. With the help of ASHRAE (American Society of Heating, Refrigerating and Air-conditioning Engineers) standard, the indoor environmental comfort can be quantitatively evaluated, and it can be used as the control variable of the indoor comfort control system, so as to effectively adjust the indoor environmental comfort through indirect control of indicators. Meanwhile, considering that the indoor environment is affected by indoor working scenes and operating modes, the invention further proposes an environmental comfort control model limited by the indoor operating mode, so that the environmental comfort control is more in line with actual requirements.
The invention aims to provide a method and a system for adjusting indoor
environmental comfort based on deep learning. Considering the Predicted Mean Vote,
PMV) of human thermal response and saving energy consumption, we provide users
with a relatively comfortable indoor environment. By adjusting the indoor
environment, we avoid excessive changes and non-real-time adjustment of the indoor
environment, improve user comfort and save energy consumption of indoor
environment adjustment equipment.
The invention provides an indoor environment comfort adjustment system based
on deep learning. According to functional attributes, it is divided into four modules,
which are data collection module, edge preprocessing and control module, indoor
environment prediction module based on deep learning, and thermal comfort control
module based on pattern feature constraints from bottom to top.
The data collection module is mainly responsible for collecting multi-source
heterogeneous environment data in the current scene, including temperature, humidity
and wind speed, and integrates the multi-source heterogeneous environment data into the edge preprocessing and control module through this module. The edge preprocessing and control module uses noise reduction and outliers to improve data availability, ensure the accuracy of input data and improve the accuracy of prediction model. Indoor environment prediction module is based on deep learning neural network, which is responsible for predicting indoor environment changes in a certain period of time in the future. The thermal comfort control module based on mode feature constraints can realize the reasonable adjustment of indoor environment under different mode constraints, so as to improve the thermal comfort of indoor environment.
The invention provides an indoor environment comfort adjustment method based
on deep learning, which comprises the following steps:
Si. A data collection module collects indoor real-time multi-source
heterogeneous environment data and sends the environment data to an edge
preprocessing and control module;
S2. Carrying out data preprocessing on multi-source heterogeneous environment
data uploaded by sensors in an edge preprocessing and control module;
S3. Building a model based on deep learning LSTM neural network to predict
the change of indoor environment in a certain period of time in the future;
S4. Taking the multi-source heterogeneous environment data processed by the
edge preprocessing and control module as the input of the indoor environment
prediction model;
S5. The indoor environment prediction model completes the prediction of indoor
environment changes according to the uploaded indoor multi-source heterogeneous environment data, and sends the results to the thermal comfort control model based on different mode feature constraints in the next step;
S6. Constructing a thermal comfort control model based on different mode
feature constraints, and proposing an indoor environment adjustment scheme for
improving human comfort under different scene mode constraints;
S7. The data center is directly connected with the edge preprocessing and control
module. If the regulation result cannot meet the user's own needs, the application
program can adjust the instruction, and the application program sends the system
instruction to the edge preprocessing and control module, and then the edge
preprocessing and control module sends the instruction to the relevant environmental
control equipment.
In the step S2 described in the present invention, the pretreatment operation for
the multi-source heterogeneous environment program is included, including:
S21: Aiming at the multi-source heterogeneous environment data uploaded by
sensors affected by environmental fluctuations and faults in the venue, delete the
abnormal values or set the missing bits, and the missing values are processed by
interpolation or elimination;
S22: At the same time, multi-source and heterogeneous environmental data are
processed based on Kalman filter algorithm, and Kalman gain is calculated and
continuously corrected and updated to realize noise reduction of environmental data.
At the same time, the preprocessed data is transmitted to the indoor environment
prediction model.
Step S3 is described as the present invention, includes the construction and
training of the deep learning model, including:
S31: The preprocessed multi-source heterogeneous historical environment data
set including temperature, humidity and wind speed is normalized and divided into
training and verification data sets.
S32: Set the number of neural network layers, input dimension and time step of
input data, LSTM input data reading batch size, window length LSTM model
optimizer and learning rate, and model iteration times;
S33: Finally, continuously adjust the parameters, check the convergence degree
of the model by the model loss, and select the parameters with high convergence
degree to form an indoor environment prediction model based on LSTM;
In the step S6 described in the present invention, the indoor environment is
adjusted, including:
S62: Receive the result of the indoor environment prediction model in the
previous step, calculate the comfort level by PMV formula and obtain the target
temperature of the indoor environment for different modes by combining the indoor
environment parameter variation range set under the constraint of scene mode, and
generate the environmental control signal according to the target temperature.
S63: Finally, the adjustment result is fed back to the indoor environment
prediction model for training and updating the model to improve the system efficiency.
Compared with the prior art, the indoor environmental comfort adjustment
method and system based on deep learning provided by the invention analyze
environmental data uploaded by sensors through a prediction model, predict future indoor environmental changes, and feed them back to an indoor thermal comfort control model; the control model puts forward an optimization scheme of human thermal comfort based on current scene mode constraints, and feeds the results back to the prediction model, At the same time, the related equipment is controlled to adjust the indoor environment to a suitable state, which effectively avoids the repeated adjustment of the environment, reduces the energy consumption, and can put forward a reasonable scheme for indoor environment adjustment according to different scenarios and operation modes.
In order to explain the technical scheme of the embodiments of the present
invention more clearly, the figures used in the embodiments will be briefly introduced
below. Obviously, the drawings in the following description are only some
embodiments of the present invention, and for those of ordinary skill in the art, other
figures can be obtained according to these drawings without paying creative labor.
Fig. 1 is a work flow chart of an indoor environment comfort adjustment method
and system based on deep learning provided by the present invention.
Fig. 2 is a structural schematic diagram of the indoor environment comfort
adjustment system provided in embodiment 1 of the present invention.
In order to make the objectives, technical solutions and advantages of the
embodiments of the present invention clearer, the technical solutions in the
embodiments of the present invention will be described clearly and completely in
combination with the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, not all embodiments.
Based on the embodiments of the present invention, all other embodiments obtained
by ordinary technicians in the field without creative labor belong to the scope of
protection of the present invention.
Terms used in the embodiments of the present invention are for the purpose of
describing specific embodiments only, and are not intended to limit the present
invention. As used in the embodiments of the present invention and the appended
claims, the singular forms "a", "the" and "the" are also intended to include the plural
forms unless the context clearly indicates other meaning.
The following is a further detailed description of the present invention with
reference to the accompanying drawings:
Example 1
As shown in fig. 1, the present invention provides an indoor environment
comfort adjustment method and system based on deep learning. Fig. 2 is a structural
schematic diagram of the indoor environment comfort adjustment system provided in
embodiment 1 of the present invention, taking the indoor environment as an example,
and specifically comprising the following steps:
Si: Accquiring environmental data by using sensor equipment in a data
acquisition module, and sending environmental state data to an edge preprocessing
and control module;
S2: After collecting indoor environmental data, the edge preprocessing and
control module caches and preprocesses the data through the micro server, and then
sends the data to the data center through the gateway;
The edge preprocessing and control module is responsible for collecting data
from temperature, humidity, wind speed and other sensors, and sending the data to
indoor environment prediction model and data center after data cleaning and noise
reduction.
Among them, "Statistical Product and Service Solutions" software (SPSS) is
used for data cleaning to delete abnormal values or set missing bits, and the missing
values are processed by interpolation or elimination;
Noise reduction of environmental data based on the principle of Kalman filter;
according to the best estimate x at a certain time in the past, predict the state variable
y at the next time, observe the state at the same time, and get the observed variable z,
and then analyze between prediction and observation, or modify the predicted
quantity by observation, so as to get the best estimate at the next time.
S3: Storing the environmental state data in the edge preprocessing and control
module, and inputting the real-time environmental data into the deep learning LSTM
prediction model trained in advance to obtain the environmental changes in the future;
The core of LSTM lies in the control unit state C, and the control includes
forgetting gate ft, input gate it and output gate ot. At the current time t, the forgetting
gate ft is responsible for controlling how much ct-i at the previous time is saved to ct at
the current time; the input gate it is responsible for controlling how much of the
immediate state at the current time is input to the current unit state ct; output gate ot is
responsible for controlling how much of the current cell state ct is the hidden layer
output ht at the current time. The calculation formulas are as follows:
(1) ft =G(wf•[xt,ht-1]+bf)
(2) it= a(wi•[xt,ht-i]+bi)
(3) ot = a(w•[xt,ht-1]+bo)
wf, wi, wo are the weight matrices of forgetting gate, input gate and output
gate respectively, bf bi, bo are the bias terms of forgetting gate, input gate and
output gate respectively, and Y is sigmoid function.
The LSTM input includes: the unit state ct-i at the previous time, the LSTM
hidden layer output value ht-i at the previous time, and the network input value xt at
the current time t; outputs of LSTM include the current cell state ct and the hidden
layer output value ht of LSTM.
Among them, the current input unit state is determined by the network input xt at
the current time t and the LSTM hidden layer output value ht-i at the previous time,
and its calculation formula is: 61-tanh (wc.[xt,ht-1]+bc)
Where we is the weight matrix of the input cell state, be is the bias term of the
input cell state, and tanh is a hyperbolic tangent function
The current cell state ct is determined by the forgetting gate ft, the last cell state
ct-i, the input gate it and the current input cell state, and its calculation formula is
&t-ftL0 et- i+it L0 & = ft E ct - i + it L0 tanh (wo.-[ xt, h t -1I] + bc)
Among them, the symbol ) indicates multiplication by element. The hidden
layer output value ht of LSTM at the current time is jointly determined by the output gate ot and the current cell state ct, and its calculation formula is: hj=ot F tanh(ct)
The LSTM neural network output Xt+1 = is determined by the hidden layer
output layer ht of LSTM, the weight matrix w of the output layer and the bias term b
of the output layer, and its calculation formula is:Xt+=(wLht+b
The mean square error (MSE) is used as the loss function, and its calculation
I N loss- (X - x')2 formula is as follows: N t
In which n is the number of samples, x is the observed value and x' is the
predicted value.
In the data preprocessing stage of the model, when the environmental data is
predicted by using the series data composed of multiple variable sequences, the
dimensions of different variables are different, and the numerical values are also very
different. Considering the input and output range of nonlinear activation function in
the model, in order to avoid neuron saturation, it is necessary to normalize the
variable time series.
The prediction model is based on Keras deep learning framework, using LSTM
network related modules in Keras framework, by setting the input dimension of
LSTM and the time step of input data; LSTM input data reads batch size and window
length LSTM model optimizer and learning rate; number of hidden ganglion points;
Times of model iteration; constantly adjust the parameters, check the convergence degree of the model by the model loss, and select the parameters with high convergence degree to form the indoor environment prediction model based on LSTM.
S4: Based on the results of the indoor environment prediction model in the
previous step, the thermal comfort control model under the constraint of different
modes is used to receive the results of the indoor environment prediction model in the
previous step, and combined with the indoor environment parameter variation range
set under the constraint of scene mode and PMV thermal sensation scale (as shown in
Table 1), PMV formula (as shown below) is used to calculate comfort and fuzzy
algorithm to obtain the target temperature of indoor environment for different modes,
and environmental control signals are generated according to the target temperature.
PMV thermal cold cool Light Moderate Light warm hot sensing scale cool warm
PMV value <-2.5 -2.5 to - -1.5 to - -0.5 to 0.5 to 1.5 1.5 to 2.5 >3 range 1.5 0.5 0.5
Table 1
PMV=[0.303e-0.036M+0.028]{M-W-3.0*1a-3[5.733-6.99(M-W)-pa][0039]
0.42[(M-W)-58.15]-1.7*10- 5 M(5867-pa)-0.0014M(34-ta)-[0040]3.96*10 8 fci[(tci+273) 4 -(ts+273)4 -fcihc(tci-ta)]}
In which M is the metabolic rate, and the unit is W/m2; W is the power of
human body, and the unit is W/s; ta is the indoor air temperature in °C; ts is the
average radiation temperature, in °C. tci is the average temperature of the outer
surface of the dressed human body, in °C: pa is the partial pressure of water vapor in ambient air, and its unit is Pa, and its specific calculation formula is as follows: pa = 6107.8x RH x exp[ta / (ta + 238.2)x 17.2694] fi is the ratio of the surface area of the dressed human body to the naked body, and its specific calculation formula is as follows; fICl {1.1.05+0.6451cl,Icl>0.078 00+1.290IclIcl<0.078 he is the convective heat exchange coefficient in W/(s . m2 . °C), and its specific calculation formula is as follows: h =12.1x Fi
There are 8 variables in the formula: M (metabolic rate), W (human body
working power), pa (partial pressure of water vapor in ambient air), ta (indoor air
temperature), fi (ratio of surface area of dressed human body to naked human body),
ts (average radiant temperature), tci (average temperature of outer surface of dressed
human body), he (convective heat exchange coefficient). Actually, M (metabolic rate),
pa (partial pressure of water vapor in ambient air), ta (indoor air temperature), fi (ratio
of dressed human body to naked surface area), ts (average radiant temperature) and tci
(average temperature of dressed human body outer surface) can be replaced by the
average value of statistics in a period of time, and W (human body power) is
considered as 0 here.
From the above formula, it can be seen that there are six factors that affect
human comfort: air temperature, air velocity (wind speed), relative humidity, average
radiation temperature, human metabolic rate and clothing thermal resistance. For human metabolic rate and clothing thermal resistance, we take fixed values here, and adjust them according to different seasons and weather. Secondly, through the analysis of PMV formula, we know that the relative humidity has little influence on the comfort level, so in the comfort level adjustment model, we mainly adjust the indoor comfort level by adjusting two environmental variables: temperature and air velocity (wind speed).
Fuzzy control algorithm is used to adjust thermal comfort, Tup is the upper
temperature limit of target temperature domain, Tdown is the lower temperature limit;
Wup is the upper temperature limit, Wdown is the lower temperature limit, Tup is the
upper temperature limit and Tdown is the lower temperature limit. Firstly, the indoor
temperature valueT and wind speed value W at the next moment are obtained from
the indoor environment prediction model. According to the difference of PMV
thermal sensing scale corresponding to indoor thermal comfort, the target temperature
is given by using different adjustment ranges according to fuzzy control algorithm.
The specific fuzzy control algorithm is shown in Table 2:
Update of target parameters
Hot Tup= T;Tdown= T-3x;T =(Tup+Tdown)/2 Wup =W+3y;Wwn= W;W =(Wup+Wdown)/2 Warm Tup= T;Tdown= Tup-2x;T =(Tup+Tdown)/2 Wup =W+2y;Wown= W;W =(Wup+Wdown)/2 Light warm Tup = T;Tdown= Tup-x;T= (Tup+Tdown)/2 Wup =W+y;Wdown= W;W =(Wup+Wdown)/2 Moderate No update required Light cool Tup= T+x;Tdown= T;T =(Tup+Tdown)/2 Wup=W;Wdown= W-y;W =(Wup+Wdown)/2 Cool Tup= T+2x;Tdown= T;T =(Tup+Tdown)/2 Wup=W;Wdown= W-2y;W =(Wup+Wdown)/2 Cold Tup= T+3x;Tdown= T; T =(Tup+Tdown)/2
Wup=W;Wdown= W-3y;W=(Wup+Wown)/2 Table 2
x represents the preset indoor temperature control return difference, which can
be set independently, and usually takes 0.5°C or 1 C. y represents the preset indoor
wind speed control return difference, which is usually 0.05m/s
. At the same time, the fuzzy algorithm is limited according to the range set by the
system mode parameters in the thermal comfort adjustment module. When the
adjustment range of a certain parameter exceeds the range set by the system mode, the
current adjustment operation of this environmental parameter is terminated and the
previous value is kept unchanged. And the final adjustment results are fed back to the
indoor prediction model, and the effect of the model is optimized to adjust the indoor
environment.
In the embodiment provided by the present invention, it should be understood
that the disclosed system, device and method can be realized in other ways. For
example, the system embodiment described above is only schematic. For example, the
division of the unit is only a logical function division. In actual implementation, there
may be another division mode. For example, multiple units or components can be
combined or integrated into another system, or some features can be ignored or not
executed. On the other hand, the mutual coupling, direct coupling or communication
connection shown or discussed may be indirect coupling or communication
connection through some interfaces, devices or units, and may be in electrical,
mechanical or other forms.
The above integrated units realized in the form of software functional units can
be stored in a computer readable storage medium. The above-mentioned software functional units are stored in a storage medium, and include several instructions to make a computer device (which may be a personal computer, a server, or a network device, etc.) or a Processor execute part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage media include: USB flash drive, mobile hard disk, Read-Only Memory (ROM), Random
Access Memory (RAM), magnetic disk or optical disk, and other media that can store
program codes.
The above embodiments are only used to illustrate the technical scheme of the
present invention, and are not used to limit the present invention. For those skilled in
the art, the present invention can be modified and varied. Any modification,
equivalent substitution, improvement, etc. made within the spirit and principle of the
present invention shall be included in the protection scope of the present invention.
Claims (6)
1. An indoor environment comfort adjustment system based on deep learning,
comprising: According to its functional attributes, the indoor environment comfort
adjustment system is divided into four modules, which are data acquisition module,
edge preprocessing and control module, indoor environment prediction module and
thermal comfort adjustment module based on pattern feature constraints from bottom
to top; the data acquisition module is mainly responsible for environmental data
acquisition, including temperature and humidity sensors, wind speed sensors and
equipment controllers; the edge preprocessing and control module mainly
preprocesses the multi-source heterogeneous environment data uploaded by sensors;
the data preprocessing module uses noise reduction, outlier removal and other means
to improve the data availability, ensure the accuracy of the input data of the prediction
model and improve the accuracy of the prediction model, and upload the preprocessed
data to the edge preprocessing and control module. Indoor environment prediction
module is the core layer of indoor environment comfort adjustment system; based on
deep learning neural network, indoor environment prediction module is responsible
for predicting the changes of venue environment in a certain period of time in the
future; the thermal comfort adjustment module based on mode feature constraints is
responsible for the interaction between indoor environment control system and users,
and adjusts indoor environment equipment through algorithm analysis, so as to realize
reasonable adjustment of venue environment under different mode constraints, thus
achieving the purpose of improving indoor environment thermal comfort.
2. The indoor environment comfort adjustment method based on deep learning is
characterized by comprising the following steps:
Si: Acquiring environmental data by using sensor equipment in a data
acquisition module, and sending environmental state data to an edge preprocessing
and control module;
S2: After the edge preprocessing and control module collects the environmental
state data, the data is preprocessed by the micro server and then sent to the indoor
environment prediction module through the gateway;
S3: Storing the environmental state data in the indoor environment prediction
module, and inputting the real-time environmental state data into the deep learning
LSTM model trained in advance to obtain the change of the indoor environment at the
next moment;
S4: The indoor environment prediction module sends the signal of indoor
environment change at the next moment output by the prediction model to the thermal
comfort adjustment module, and the adjustment module sends the regulation
information of the system to the edge preprocessing and control module through an
algorithm, and the edge preprocessing and control module sends instructions to
relevant environmental control equipment;
S5: The data center is directly connected with the edge preprocessing and control
module; if the regulation result of the thermal comfort adjustment module is not
enough to meet the user's own needs, the application program can adjust the
instruction, and the application program sends the system instruction to the edge
preprocessing and control module, and then the edge preprocessing and control
module sends the instruction to the relevant environmental control equipment.
3. The step S according to claim 2 is characterized by comprising:
Sensors such as temperature, humidity and wind speed are deployed to detect
current indoor environmental conditions such as temperature, humidity and wind
speed, and the devices such as temperature, humidity and wind speed communicate
with the edge preprocessing and control module through Zigbee or Echonet Lite.
4. The step S2 according to claim 2 is characterized by comprising:
The edge preprocessing and control module is responsible for collecting data
from sensors and sending the data to the indoor environment prediction module
through MQTT protocol (a famous Internet of Things protocol for data collection).
5. The step S3 according to claim 2 is characterized by comprising:
S31: In the data preprocessing stage of the model, when the environmental data
is predicted by using the sequence data composed of multiple variable sequences, the
dimensions of different variables are different, and the numerical values are also very
different. Considering the input and output range of nonlinear activation function in
the model, in order to avoid neuron saturation, it is necessary to normalize the
variable time series;
S32: The prediction model is based on the Keras deep learning framework, and
the LSTM network related module in the Keras framework is used to set the input
dimension of LSTM and the time step of input data;
S33: LSTM input data reads the batch size and window length LSTM model
optimizer and learning rate;
S34: The number of hidden ganglion points; Times of model iteration;
Constantly adjust the parameters, check the convergence degree of the model by the model loss, and select the parameters with high convergence degree to form the indoor environment prediction model based on LSTM.
6. The step S4 according to claim 2 is characterized by comprising:
Fuzzy control algorithm is used to adjust thermal comfort, Tup is the upper
temperature limit of target temperature domain, Tdown is the lower temperature limit;
Wup is the upper temperature limit, Wdown is the lower temperature limit, Tup is the
upper temperature limit and Tdown is the lower temperature limit. Firstly, the indoor
temperature value t and wind speed value w at the next moment are obtained from the
indoor environment prediction model; according to the difference of PMV thermal
sensing scale corresponding to indoor thermal comfort, the target temperature is given
by using different adjustment ranges according to fuzzy control algorithm.
FIGURES 1/2
Figure 1
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| AU2021105951A AU2021105951A4 (en) | 2021-08-19 | 2021-08-19 | Method and System for Adjusting Indoor Environment Comfort Based on Deep Learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| AU2021105951A AU2021105951A4 (en) | 2021-08-19 | 2021-08-19 | Method and System for Adjusting Indoor Environment Comfort Based on Deep Learning |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| AU2021105951A4 true AU2021105951A4 (en) | 2021-10-28 |
Family
ID=78179587
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| AU2021105951A Active AU2021105951A4 (en) | 2021-08-19 | 2021-08-19 | Method and System for Adjusting Indoor Environment Comfort Based on Deep Learning |
Country Status (1)
| Country | Link |
|---|---|
| AU (1) | AU2021105951A4 (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114355767A (en) * | 2022-03-21 | 2022-04-15 | 青岛理工大学 | Q learning-based model-free control method for indoor thermal environment of endowment building |
| CN116624976A (en) * | 2023-07-07 | 2023-08-22 | 圣辉工程科技有限公司 | Central air conditioner remote control system and method |
-
2021
- 2021-08-19 AU AU2021105951A patent/AU2021105951A4/en active Active
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114355767A (en) * | 2022-03-21 | 2022-04-15 | 青岛理工大学 | Q learning-based model-free control method for indoor thermal environment of endowment building |
| CN114355767B (en) * | 2022-03-21 | 2022-06-24 | 青岛理工大学 | Q learning-based model-free control method for indoor thermal environment of endowment building |
| CN116624976A (en) * | 2023-07-07 | 2023-08-22 | 圣辉工程科技有限公司 | Central air conditioner remote control system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113485498B (en) | Indoor environment comfort level adjusting method and system based on deep learning | |
| AU2021105951A4 (en) | Method and System for Adjusting Indoor Environment Comfort Based on Deep Learning | |
| EP3029389B1 (en) | Controlling system for environmental comfort degree and controlling method of the controlling system | |
| CN111609534B (en) | Temperature control method and device and central temperature control system | |
| CN108413588B (en) | Personalized air conditioner control system and method based on thermal imaging and BP neural network | |
| CN105571048B (en) | Group dynamic environment control | |
| CN105717960A (en) | Environmental comfort level control system and method | |
| CN108304965A (en) | The space in building is distributed based on comfortable model | |
| Turhan et al. | Development of a personalized thermal comfort driven controller for HVAC systems | |
| CN103282841A (en) | Building automation system | |
| CN114484557B (en) | Building group heat supply load regulation and control method based on target energy consumption management and control | |
| KR20190104926A (en) | Air conditioning system and controlling method thereof | |
| Li et al. | Toward intelligent multizone thermal control with multiagent deep reinforcement learning | |
| CN110986314A (en) | Intelligent air supply adjusting method of air conditioner and air conditioner | |
| CN110377084A (en) | A kind of Building Indoor Environment regulation method based on wisdom control strategy | |
| CN110726209B (en) | Air conditioner control method and device, storage medium and processor | |
| CN109857177B (en) | Building electrical energy-saving monitoring method | |
| CN114838470A (en) | Control method and system for heating, ventilating and air conditioning | |
| TWI746087B (en) | Air conditioning system control method | |
| EP3771957A1 (en) | Method and system for controlling of heating, ventilation and air conditioning | |
| Eftekhari et al. | Design and performance of a rule-based controller in a naturally ventilated room | |
| CN113418286B (en) | Self-adaptive thermal sensing robot and air conditioner temperature adjusting method | |
| KR101799754B1 (en) | An intelligent homeostasis maintaining system and method for making a comfortable environment | |
| US20210033299A1 (en) | Method and system for controlling heating, ventilation and air conditioning | |
| CN113760022B (en) | Public space thermal environment air conditioner control device and method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| FGI | Letters patent sealed or granted (innovation patent) |