CN110991478A - Method for establishing thermal comfort model and method and system for setting user preference temperature - Google Patents

Method for establishing thermal comfort model and method and system for setting user preference temperature Download PDF

Info

Publication number
CN110991478A
CN110991478A CN201911037626.9A CN201911037626A CN110991478A CN 110991478 A CN110991478 A CN 110991478A CN 201911037626 A CN201911037626 A CN 201911037626A CN 110991478 A CN110991478 A CN 110991478A
Authority
CN
China
Prior art keywords
thermal
thermal comfort
user
model
temperature
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201911037626.9A
Other languages
Chinese (zh)
Other versions
CN110991478B (en
Inventor
白燕
冯壮壮
万陶成
张玮
武璐璐
王秀东
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Xian University of Architecture and Technology
Original Assignee
Xian University of Architecture and Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Xian University of Architecture and Technology filed Critical Xian University of Architecture and Technology
Priority to CN201911037626.9A priority Critical patent/CN110991478B/en
Publication of CN110991478A publication Critical patent/CN110991478A/en
Application granted granted Critical
Publication of CN110991478B publication Critical patent/CN110991478B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/043—Architecture, e.g. interconnection topology based on fuzzy logic, fuzzy membership or fuzzy inference, e.g. adaptive neuro-fuzzy inference systems [ANFIS]
    • 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models
    • G06N5/048—Fuzzy inferencing

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Biomedical Technology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Automation & Control Theory (AREA)
  • Fuzzy Systems (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Air Conditioning Control Device (AREA)

Abstract

The invention belongs to the field of building thermal environments and discloses a thermal comfort model establishing method and a user preference temperature setting method and system. The thermal comfort model building method comprises the steps of collecting a thermal comfort data set, and carrying out data cleaning on the thermal comfort data set to obtain a sample data set; then carrying out ECM clustering on the sample data set to obtain a cluster set; establishing fuzzy rules, wherein the fuzzy front part of each fuzzy rule is a cluster set, the fuzzy back part is a coefficient of a target function calculated based on a first-order Takagi-Sugeno model, the coefficient of the target function is subjected to iterative optimization, and a thermal comfort model is obtained according to the optimized fuzzy rules. The setting method of the user preference temperature comprises the steps of collecting the thermal sensation index TPI and the air temperature T of the current user, and outputting the preference temperature of the current user by using a thermal comfort model. The invention breaks through the traditional static modeling of personal thermal comfort, and uses the dynamic evolution neural fuzzy inference system to model and research, thereby better outputting the comfortable preference temperature range of the user.

Description

Method for establishing thermal comfort model and method and system for setting user preference temperature
Technical Field
The invention belongs to the field of building thermal environments, and particularly relates to a thermal comfort model establishing method and a user preference temperature setting method and system.
Background
People spend about 90% of the time indoors, and the comfortable indoor environment is the premise of healthy life and efficient work of people, so the scientific control on the thermal environment not only relates to the thermal comfort of people, but also influences the sustainable development of building energy consumption and resources. PMV (predicted Mean volume) thermal comfort evaluation index is provided from the factory, and thermal comfort is widely researched in the fields of environmental thermal comfort evaluation, indoor thermal environment control, air-conditioning system energy-saving optimization and the like. However, thermal comfort is not only influenced by various factors such as local climate, living habits, customs, thermal experiences, thermal expectations, thermal environmental responses and personal constitutions, but also has time-varying characteristics that vary with different activities, clothes, diet and mood. The thermal comfort dynamic difference causes the comfortable preference temperature deviation among individuals to be more than 2.6 ℃, so that nearly 50% of temperature set values in office buildings are not in the range of summer cooling comfortable temperature (23.5-27 ℃) recommended by American society of heating, refrigeration and air-conditioning engineers (ASHRAE) PMV-PPD standard, and the thermal environment control in the range can only meet the comfort of most people in a statistical sense and is difficult to meet the requirement of thermal comfort diversity of a certain individual or group, which does not belong to the 'comfort' and 'energy conservation' in a real sense.
Disclosure of Invention
The invention aims to provide a thermal comfort model establishing method and a user preference temperature setting method and system, which are used for solving the problem that the requirement of a certain individual or group on thermal comfort diversity is difficult to meet in an individualized way in the prior art.
In order to realize the task, the invention adopts the following technical scheme:
a thermal comfort model building method comprises the following steps:
step 1: collecting a thermal comfort data set, wherein the thermal comfort data set comprises user basic information, thermal environment parameters and thermal voting parameters, and performing data cleaning on the thermal comfort data set to obtain a sample data set, wherein the sample data set comprises a thermal sensation index and an air temperature;
step 2: carrying out ECM clustering on the sample data set to obtain M cluster sets, wherein M is more than or equal to 1;
and step 3: establishing M fuzzy rules, wherein a fuzzy front part of each fuzzy rule is a cluster set, a fuzzy back part of each fuzzy rule is an objective function based on a first-order Takagi-Sugeno model, calculating coefficients of the objective function according to the sample data set obtained in the step 1 and the cluster set obtained in the step 2, performing iterative optimization on the coefficients of the objective function by using a least square method, and obtaining a thermal comfort model according to the optimized fuzzy rules.
Further, the user basic information in step 1 includes age, gender, activity status and clothing thermal resistance, the thermal environment parameters include air temperature, air flow rate and relative humidity, and the thermal voting parameters include a thermal sensation indicator, a thermal preference and thermal acceptability.
Further, the objective function based on the first-order Takagi-Sugeno model is shown in formula I:
f(TPI,T)＝β0+β1×(TPIi,Ti) I-1, 2, …, p is of formula I
Wherein (TPI)i,Ti) I-1, 2, …, p denotes p sets of sample data in the sample data set, TPIiA heat sensation index T representing the ith sample dataiAir temperature representing the ith sample data, β0And β1Satisfy β ═ β0β1]Tβ denotes a coefficient matrix and β ═ a (AWA)-1ATWY, A is the user heat sensation index input matrix and
Figure BDA0002251966820000021
y is an air temperature input matrix and Y ═ T1T2T3… Tp]TW is a cluster set influence factor matrix and
Figure BDA0002251966820000031
element W on the diagonal of W1Is the distance, w, from the 1 st sample data to the cluster center to which the sample belongspThe distance from the p sample data to the cluster center of the sample.
A setting method of user preference temperature comprises the following steps:
step a: after collecting user basic information, thermal environment parameters and thermal voting parameters of a current user, cleaning data to obtain a thermal sensation index TPI and an air temperature T of the current user;
step b: obtaining M cluster sets in the step 2 according to any one of the hot comfort model building methods, and selecting M cluster sets to which a hot feeling index TPI and an air temperature T of a current user belong, wherein M is more than or equal to 1 and less than or equal to M;
step c: and c, obtaining the thermal comfort model in the step 3 according to any one of the thermal comfort model establishing methods, wherein the thermal comfort model comprises m fuzzy rules corresponding to the m cluster sets selected in the step b, inputting the thermal sensation index TPI and the air temperature T of the current user into the thermal comfort model, and outputting the preference temperature of the current user.
Further, in step c, the preferred temperature of the current user is output by using formula ii:
Figure BDA0002251966820000032
wherein, yoIndicates the preferred temperature, omega, of the current userqRepresents the weight of the current user belonging to the q fuzzy rule and 0 < omegaq＜1。
Further, ω isiCalculated by a trigonometric membership function.
A setting system of user preference temperature comprises a data acquisition module, a data cleaning module, a classification module, a thermal comfort model and a preference temperature setting module;
the data acquisition module is used for acquiring user basic information, thermal environment parameters and thermal voting parameters of a current user and transmitting the acquired information to the data cleaning module;
the data cleaning module is used for cleaning the data acquired by the data acquisition module to obtain the thermal sensation index TPI and the air temperature T of the current user;
the classification module is used for selecting M cluster sets to which the thermal sensation index TPI and the air temperature T of the current user belong, wherein M is more than or equal to 1 and less than or equal to M, and the M cluster sets are obtained through the step 2 of any one of the thermal comfort model building methods;
the thermal comfort model is used for inputting a thermal sensation index TPI and an air temperature T of a current user into the thermal comfort model and outputting a preference temperature of the current user, the thermal comfort model is obtained according to any one of the thermal comfort model establishing methods, and the thermal comfort model comprises m fuzzy rules corresponding to the m cluster sets selected in the step b;
the preference temperature setting module is used for setting the indoor temperature according to the preference temperature of the current user output by the thermal comfort model.
Further, the thermal comfort model outputs the current user's preferred temperature using equation ii:
Figure BDA0002251966820000041
wherein, yoIndicates the preferred temperature, omega, of the current userqRepresents the weight of the current user belonging to the q fuzzy rule and 0 < omegaq＜1。
Further, ω isiCalculated by a trigonometric membership function.
Compared with the prior art, the invention has the following technical characteristics:
(1) the method for establishing the thermal comfort model and the method and the system for setting the user preference temperature solve the problem of complexity in the process of counting the user thermal information in the traditional questionnaire form by virtue of the advantages of low investment, easiness in maintenance, convenience in interaction, information visualization and the like, and improve the user participation degree.
(2) According to the effectiveness of the adaptive model in thermal comfort environment control, the temperature range is defined as a thermal comfort zone, the invention simplifies the user thermal comfort model, simplifies the user thermal comfort zone model according to the user comfort temperature range, and outputs the user comfort preference temperature range by using the dynamic evolution neural fuzzy inference system for modeling research. The simplified model research not only reduces the detection of thermal comfort related variable parameters, but also shortens the modeling period.
(3) The method is realized based on a dynamic evolutionary neural fuzzy inference system algorithm, is suitable for theory and application research of a dynamic system, combines an evolutionary clustering algorithm (ECM) with a fuzzy membership function, constructs a fuzzy set in an online clustering mode, extracts a fuzzy rule and determines a fuzzy rule antecedent; and constructing a fuzzy rule back part by adopting a Takagi-Sugeno inference model according to the local linear fuzzy membership degree. The algorithm can dynamically update the model parameters along with the increase of the sample data volume, revise the user thermal comfort area, can perform dynamic modeling, and breaks through the traditional static modeling of personal thermal comfort, so that the algorithm is the optimal model for simulating the user thermal comfort online learning process with time-varying characteristics.
Drawings
FIG. 1 is a block diagram of an inventive intelligent interactive system;
FIG. 2 is a graph of youth user data and clustering results from a simulation experiment of the present invention;
FIG. 3 is a graph of the middle-aged user data and clustering results of a simulation experiment implementation of the present invention;
FIG. 4 is a graph of senior user data and clustering results from a simulation experiment of the present invention;
FIG. 5 is a diagram of the thermal comfort reasoning results of young users implemented in a simulation experiment of the invention;
FIG. 6 is a diagram of the results of a reasoning of thermal comfort for a middle-aged user performed in a simulation experiment of the invention;
FIG. 7 is a diagram of the inference results of thermal comfort of an elderly user performed by a simulation experiment of the present invention;
FIG. 8 is a graph of the number of rules for different Dthr for a simulation experiment implementation of the invention;
FIG. 9 is a graph of the impact of Dthr on the inference prediction error of young users as performed by a simulation experiment of the present invention;
FIG. 10 is a graph of the impact of Dthr on the inference prediction error of a middle-aged user as performed by a simulation experiment of the present invention;
FIG. 11 is an illustration of the impact of Dthr on the inference prediction error of an elderly user as performed by a simulation experiment of the present invention;
FIG. 12 is a chart of user 1 thermal comfort reasoning results for one field experimental implementation of the invention;
FIG. 13 is a chart of user 2 thermal comfort reasoning results for one field experimental implementation of the invention;
FIG. 14 is a chart of user 3 thermal comfort reasoning results for one field experimental implementation of the invention;
FIG. 15 is a diagram of group user thermal comfort reasoning results for a multi-user experiment implementation of the invention.
Detailed Description
The following embodiments and examples are provided to demonstrate the present invention, and it should be noted that the present invention is not limited to the following embodiments, and all equivalent changes based on the technical solutions of the present application fall within the protection scope of the present invention.
The definitions and connotations related to the present invention will be explained first:
ecm (evolution clustering method) clustering algorithm: the algorithm dynamically increases the number of clusters or changes the cluster centers and the cluster radii in real time along with the increase of input samples, and is a dynamic and online clustering algorithm which is restricted by a certain maximum distance Dthr. The clustering process starts from an empty cluster set, the cluster number is dynamically increased or the cluster center and the cluster radius are updated in real time along with the increase of new sample data, and the updating is stopped when the radius reaches a cluster radius threshold Dthr. The algorithm can dynamically update the clustering number along with the increase of the sample data volume so as to meet the time-varying characteristic of thermal comfort of a user.
TPI heat sensation index: the real thermal sensation of the user is defined as real thermal sensation of real-time feedback, a range of [ -3, +3] is taken by referring to a thermal comfort zone established by an ASHRAE standard, interaction is realized on a client APP in a sliding bar mode, a positive value represents thermal sensation, a negative value represents cold sensation, and the larger an absolute value is, the more obvious the thermal sensation of the user is.
Thermal preference: the user can feed back the expectation of the current temperature at the current temperature, including the expectation that the current temperature is high, the expectation that the current temperature is constant and the expectation that the current temperature is low.
Thermal acceptability: whether the current environment representation is acceptable mainly comprises acceptable and unacceptable.
Selection of (2).
Optimum preferred temperature: according to the American society of heating, refrigeration and air-conditioning Engineers (ASHRAE) PMV-PPD standard, the PMV value 0 is recommended to be used as an optimal comfort point, the temperature point corresponding to the thermal sensation index TPI of 0 is defined as the optimal preference temperature of the user, and the personal comfort range of the user is defined by the TPI of-0.5.
Fuzzy rules: the essence is a binary fuzzy relation R defined at X Y. The fuzzy rule is of the form: if x is Athen y is B.
In this embodiment, a thermal comfort model building method is disclosed, which includes the following steps:
step 1: the method comprises the steps of collecting a thermal comfort data set, wherein the thermal comfort data set comprises user basic information, thermal environment parameters and thermal voting parameters, and performing data cleaning on the thermal comfort data set to obtain a sample data set, wherein the sample data set comprises a thermal sensation index TPI and an air temperature T;
step 2: carrying out ECM clustering on the sample data set to obtain M cluster sets;
and step 3: establishing M fuzzy rules, wherein a fuzzy front part of each fuzzy rule is a cluster set, a fuzzy back part of each fuzzy rule is an objective function based on a first-order Takagi-Sugeno model, calculating coefficients of the objective function according to the sample data set obtained in the step 1 and the cluster set obtained in the step 2, performing iterative optimization on the coefficients of the objective function by using a least square method, and obtaining a thermal comfort model according to the optimized fuzzy rules.
The method is mainly realized based on a dynamic evolution neural fuzzy inference system algorithm, and the algorithm is suitable for theory and application research of a dynamic system by the characteristics of univariate modeling, high prediction precision, strong reliability, quick learning and the like. In the thermal comfort research, PMV index related parameters such as heat resistance, metabolic rate, air flow rate and the like are used in the traditional thermal comfort modeling, and the factors such as the difficulty in accurate measurement of the parameters, complex modeling and the like cause time waste and errors for model development; and the thermal comfort has time-varying characteristics which change along with climate, living habits and thermal experiences, so that the long-time modeling process lags behind the thermal comfort of a real user, therefore, the method is the optimal model for simulating the thermal comfort on-line learning process of the user.
Specifically, the user basic information in step 1 includes age, gender, activity status and clothing thermal resistance, the thermal environment parameters include air temperature, air flow rate and relative humidity, and the thermal voting parameters include a thermal sensation indicator, a thermal preference and thermal acceptability.
The activity states in the basic information comprise specific office states, such as states of relying, sitting still, standing, walking, physical work and the like, the states are used for setting different metabolism, and the thermal resistance of clothes is selected by dressing; then calculating PMV according to the air temperature, the air flow rate and the relative humidity parameters, wherein the value of PMV is-3, (-3 represents coldest, 0 represents no cold or no hot, 3 represents hottest, and the larger the value is, the hotter the value is); using PMV, three parameters of thermal preference and thermal acceptability are selected as effective learning samples, i.e. thermal sensation index TPI and temperature T.
Specifically, the data cleaning is to keep the data with basically consistent information expressed by the data of the thermal sensation index, the thermal preference and the thermal acceptability, and delete the inconsistent data, for example, if the TPI sliding value of the current data is 0, the data means that the individual is comfortable with the current temperature and does not need to increase or decrease the temperature, but if the thermal preference is selected at the moment, the current temperature is expected to be higher or lower, which is mutually contradictory with the current TPI, so that the data of the group needs cleaning.
Specifically, the fuzzy rule is expressed as follows:
If(TPI,T)is R1，Then y is ω1f(TPI,T)
If(TPI,T)is R2，Then y is ω2f(TPI,T)
…
If(TPI,T)is Rm，Then y isωmf(TPI,T)
where R isj J 1,2,3 …, m, a first order linear function ω, for different sets of clusters as the blur predecessorif (TPI, T) represents the fuzzy back-piece part;
preferably, the objective function based on the first order Takagi-Sugeno model is shown in formula I:
f(TPI,T)＝β0+β1×(TPIi,Ti) I-1, 2, …, p is of formula I
Wherein (TPI)i,Ti) I-1, 2, …, p denotes p sets of sample data in the sample data set, TPIiA heat sensation index T representing the ith sample dataiAir temperature representing the ith sample data, β0And β1Satisfy β ═ β0β1]Tβ denotes a coefficient matrix and β ═ a (AWA)-1ATWY, A is the user heat sensation index input matrix and
Figure BDA0002251966820000091
y is an air temperature input matrix and Y ═ T1T2T3… Tp]TW is a cluster set influence factor matrix and
Figure BDA0002251966820000092
element W on the diagonal of W1Is the distance, w, from the 1 st sample data to the cluster center to which the sample belongspThe distance from the p sample data to the cluster center of the sample.
Specifically, the sub-steps of the ECM algorithm in step 2 are as follows:
step 2.1, create the first class C1Inputting the first sample X1(TPI1,T1) As a clustering center C of the classc1Initialization of the clustering radius Ru1＝0。
Step 2.2, if all the sample data are processed, finishing clustering; otherwise, calculating the current input sample Xi and the clustering center C by using the formula (1)cjJ is the euclidean distance Dij of 1,2,3 …, n.
Figure BDA0002251966820000093
Step 2.3, if there is one DijSatisfy Dij≤RujThen, it indicates that the sample Xi belongs to the existing mth cluster CmI.e. Dim≤RumAt this time, the clustering does not need to be updated, and the step 1.2 is returned; otherwise, step 2.4 is performed.
Step 2.4, if sample X is calculatediThe distance from all cluster centers is greater than the cluster radius RujThen calculate Sij＝Dij+RujAnd taking the minimum value min (S)ij) Find class CaSo that Sia＝min(Sij)。
Step 2.5, if Sia>2 × Dthr, the existing class, return to step 1; otherwise, step 2.6 is performed.
Step 2.6, if SiaLess than or equal to 2 XDthr, updating Ca-like clustering center CcaAnd a cluster radius Ruj＝SiaAnd/2, returning to the step 2.2.
Through an ECM algorithm, data sample points of different users can obtain different clusters in a limited number of cluster sets, the sample points are gradually increased along with the interaction of the users, the cluster sets generated by the clustering are gradually changed, and the cluster number is increased in real time or the cluster center and the cluster radius are updated. The clustering process dynamically expresses the personalized data pattern of the user and represents the time-varying characteristic of the thermal comfort of the user.
A setting method of user preference temperature comprises the following steps:
step a: after collecting user basic information, thermal environment parameters and thermal voting parameters of a current user, cleaning data to obtain a thermal sensation index TPI and an air temperature T of the current user;
step b: obtaining M clusters of step 2 according to the thermal comfort modeling method of claim 1, selecting M clusters to which the thermal comfort index TPI and the air temperature T of the current user belong, wherein 1. ltoreq. m.ltoreq.M;
step c: the thermal comfort model building method according to claim 1, obtaining the thermal comfort model of step 3, the thermal comfort model comprising m fuzzy rules corresponding to the m clusters selected in step b, inputting thermal sensation index TPI and air temperature T of the current user into the thermal comfort model, and outputting the preferred temperature of the current user.
Preferably, the current user's preferred temperature is output in step c using formula ii:
Figure BDA0002251966820000101
wherein, yoIndicates the preferred temperature, omega, of the current userqRepresents the weight of the current user belonging to the q fuzzy rule and 0 < omegaq＜1。
Preferably, said ω isiCalculated by the trigonometric membership function μ (TPI), ωi＝μ(TPIi)：
Figure BDA0002251966820000111
Wherein b is the clustering center of the current input TPI, a is b-d × Dthr, c is b + d × Dthr, and d is 1.2-2.
Specifically, when a new sample point is input into the system, the ECM clustering algorithm is clustered again, and the inference system creates a new fuzzy rule or updates the existing rule; the Takagi-Sugeno model corrects the system output function coefficient, updates the user personal thermal comfort model on line, dynamically adjusts the user thermal comfort area, and meets the user thermal comfort time-varying characteristic requirement.
A setting system of user preference temperature comprises a data acquisition module, a data cleaning module, a classification module, a thermal comfort model and a preference temperature setting module;
the data acquisition module is used for acquiring user basic information, thermal environment parameters and thermal voting parameters of a current user and transmitting the acquired information to the data cleaning module;
the data cleaning module is used for cleaning the data acquired by the data acquisition module to obtain the thermal sensation index TPI and the air temperature T of the current user;
the classification module is used for selecting M cluster sets to which the thermal sensation index TPI and the air temperature T of the current user belong, wherein M is more than or equal to 1 and less than or equal to M, and the M cluster sets are obtained through the step 2 of any one of the thermal comfort model building methods;
the thermal comfort model is used for inputting a thermal sensation index TPI and an air temperature T of a current user into the thermal comfort model and outputting a preference temperature of the current user, the thermal comfort model is obtained according to any one of the thermal comfort model establishing methods, and the thermal comfort model comprises m fuzzy rules corresponding to the m cluster sets selected in the step b;
the preference temperature setting module is used for setting the indoor temperature according to the preference temperature of the current user output by the thermal comfort model.
Specifically, the thermal comfort model outputs the preferred temperature of the current user using formula ii:
Figure BDA0002251966820000121
wherein, yoIndicates the preferred temperature, omega, of the current useriRepresents the weight of the current user belonging to the ith fuzzy rule and 0 < omegai＜1。
Implementation of simulation experiment
Because the existence of multiple factors such as objective environment, physiology, psychology and the like all influences the individual thermal comfort of the user, an accurate model of the individual thermal comfort of the user cannot be calculated, and an effective simulation experiment needs to be carried out to depict an individual thermal comfort reference model of the user. Since the behavior habits (dressing and activities) of the users in the office building environment are regular, namely the thermal resistance and the metabolic rate of the individual clothes in the office building environment are relatively fixed, parameters such as the metabolic rate, the thermal resistance of the clothes, the air flow rate and the relative humidity are set according to the characteristics of three typical users, namely young, middle-aged and elderly, respectively, and relevant parameters are shown in table 1.
TABLE 1 PMV calculation-related parameter settings for three types of people
Figure BDA0002251966820000122
Based on the research result of the professor Fanger on the relation between the indoor air temperature and the PMV value, namely that the air flow rate, the relative humidity, the radiation temperature, the metabolism and the clothing thermal resistance have a linear relation under the condition that the air flow rate, the relative humidity, the radiation temperature, the metabolism and the clothing thermal resistance parameters are the same, a formula 4 is adopted, and an equation T ═ f (PMV) is constructed to serve as a user reference model; considering influence of personal emotion, psychological condition, environmental interference, perception error and other factors on user feedback information in the field interaction process, adding N (0, sigma)2) Random noise to simulate live user interaction data. And selecting variance of 2 ℃ according to experience, respectively generating 3 groups of typical user thermal comfort field interaction data, and respectively finishing clustering of samples through an ECM (information control model) algorithm, as shown in figures 2,3 and 4. And further learning and reasoning a predicted value of the personal thermal comfort model according to the clustering result, and completing construction of the personal thermal comfort model of the user.
As can be seen from fig. 2,3 and 4, the ECM clustering algorithm clusters the data of young, middle and old users into 5 cluster sets, but the cluster sets have different spatial structures, i.e., different cluster centers and cluster radii, and represent different user thermal comfort personalized data patterns. According to the ASHRAE standard, the recommended PMV is-0.5 as a comfortable range, the PMV is 0 as an optimal comfortable point, the personal comfortable range of a user is-0.5 of TPI, and the optimal comfortable point of the user is defined by taking TPI as 0. The reasoning results of the three typical user characteristic crowd models are shown in figures 5, 6 and 7, the optimal thermal comfort temperature of young users is 21.25 ℃, and the comfort temperature range is 19.12-23.42 ℃; the optimal heat comfortable temperature of the middle-aged user is 22.99 ℃, and the comfortable temperature range is 20.43-25.46 ℃; the optimal thermal comfort temperature of the elderly user is 26.32 ℃, and the comfort temperature range is 24.87-27.77 ℃. From simulation experiment results, the maximum calculation temperature difference of the comfortable temperatures of young users and old users can reach 8.65 ℃, personal comfortable temperatures of users with different characteristics are proved to have larger difference, and the necessity of setting the existing temperature in a comfortable area of the users is learned.
To verify the accuracy of the model inference prediction, the accuracy P is defined as the ratio of the number of all correct predictions divided by the total number of predictions, as shown in equation (7). And the mean square deviation (RESD) of the predicted value, the reference model value and the field data is analyzed and inferred, as shown in the formula (8). And evaluating the learning effect of the body thermal comfort model of young, middle-aged and old people.
Figure BDA0002251966820000131
In the formula: TP is true positive, TN is true negative, FP is false positive, FN is false negative.
Figure BDA0002251966820000141
In the formula: y isjRepresenting the jth sample temperature value of the reference model and the field data; y isiRepresenting the predicted temperature value at the jth user thermal comfort point; m represents the number of test samples.
The error results of model prediction are shown in table 2, and the errors of the inference prediction value and the reference model of the thermal comfort model of young, middle and old users are respectively 0.247 ℃, 0.286 ℃ and 0.396 ℃. The model prediction accuracy rates of the prediction on the comfortable temperature ranges of young, middle and old users are respectively 90.5%, 81.0% and 85.7% under the condition that the error between the model prediction and the user reference model is less than 0.5 ℃, and the prediction accuracy rate is far higher than 56% of the prediction accuracy rate of the traditional ASHRAE PMV-PPD standard on the comfortable temperature of the users. Errors of the predicted values of the thermal comfort models and the field interaction data of young, middle and old users are respectively within 1-1.6 ℃. The error is caused by the fact that when an individual feeds back a thermal sensation index, the self thermal sensation has certain ambiguity, and subjective feedback thermal comfort has uncertainty. Therefore, the DENFIS algorithm can effectively construct a user learning model from data containing errors, and errors caused by interference factors in an interaction process are reduced.
TABLE 2 errors between inferential predictions and reference model or field data values
Figure BDA0002251966820000142
In addition, the selection of the clustering radius threshold Dthr may affect the number of fuzzy sets, thereby affecting the fuzzy rule extraction, as can be seen from fig. 8, as Dthr increases, the rule number gradually decreases, and the change rate of the extraction rule number gradually decreases. Because of the increase of the cluster radius threshold Dthr, the cluster collection range is increased, and the number of sample points can be effectively covered, so that the number of extraction rules is rapidly reduced along with the increase of the Dthr, and the change rate of the number of the extraction rules is gradually reduced.
As can be seen from fig. 9, 10, and 11, when Dthr is less than 0.2, the error between the inference output and the field data is relatively stable (as can be seen from fig. 5, the number of fuzzy rules is greater than 2); when Dthr is more than 0.03 and less than 0.13, the error between the inference output and the reference model is relatively stable (the number of fuzzy rules is 4-11), wherein when Dthr is about 0.11, the error between the inference output and the reference model is the minimum, and the number of fuzzy rules is about 5 at the moment, the main reason is that the reference model adopts 5 fuzzy set standards of a PMV calculation model, so the fuzzy set with the minimum inference prediction result error in the text is consistent with the fuzzy set number in PMV calculation of professor Fanger, and the effectiveness of the DENFIS algorithm in the application of thermal comfort learning is verified.
Implementation of the field experiment
In order to research the practicability of the DENFIS algorithm on field application and the effectiveness of simulation experiments, the mobile terminal intelligent interaction and data acquisition system developed by the research is utilized to carry out hot comfortable field experiments on users in office environments, and based on a Linux platform, an intelligent interaction system is built on a hardware platform Raspberry PI 3B + by adopting the technology of the Internet of things. The system mainly comprises a thermal interaction module and a thermal environment monitoring module. The thermal induction interaction module takes a mobile phone mobile terminal APP as a user interface; the thermal environment monitoring module is integrated with various sensors by a Raspberry Pi 3B + environment collector. Selecting a working period of a typical weather day in summer, from 15 days 6 and 6 months in 2019 to 23 days 6 and 6 months in 2019, 9 am: 00-12: 00, afternoon 2: 00-5: 00. the uniform office clothing is adopted, the upper body is a short shirt, and the lower body is trousers. The method is carried out in a conference room of an office building in the city of Western Ann, and the office has two outer walls, three windows and one door. Three healthy college students were selected as experimental subjects, and the basic information thereof is shown in table 3.
TABLE 3 basic information of the subjects
Figure BDA0002251966820000151
Figure BDA0002251966820000161
The experimental content mainly comprises two parts of indoor thermal environment parameter measurement and experimental object individual subjective interaction information acquisition. The station positions were arranged according to the "station deployment at 20 cm in front of the chest of the subject" as specified in ASHRAE55-2013 "human resident thermal environment conditions". After collected field experiment data are cleaned, a DENFIS algorithm is adopted to construct a personal thermal comfort model for air temperature and thermal sensation index parameters, three experimental object comfort temperature ranges are output, and the results are shown in figures 12, 13 and 14.
From the user thermal comfort model reasoning results in fig. 12, 13 and 14, it is seen that the optimal thermal comfort temperature of the user 1 is 24.12 ℃, and the comfort temperature range is 22.58-26.84 ℃; the optimal thermal comfort temperature of the user 2 is 25.40 ℃, and the comfort temperature range is 23.83-26.98 ℃; the optimal thermal comfort temperature of the user 3 is 26.40 ℃, and the comfort temperature range is 24.43-28.42 ℃. The model reasoning result shows that the individual comfortable temperature difference can reach 5.84 ℃. The inference model prediction values of the three users and the RESD of the field experiment interaction data are respectively 0.87 ℃, 1.17 ℃ and 0.88 ℃. Error results are all smaller than simulation error results, the simulation experiment conclusion is met, and the effectiveness of the simulation experiment on the user personal thermal comfort model learning is proved. And the error result is less than 2.6 ℃ of deviation of comfortable preference temperature among individuals, which shows that the DENFIS algorithm can effectively predict the thermal comfort of individuals and control the error within an acceptable range in field application.
Multi-user experiment implementation
To further investigate the applicability of the DENFIS algorithm to group Thermal Comfort modeling, the relationship between air temperature and Thermal sensation index in a group user Thermal Comfort data set was analyzed herein in accordance with ASHRAE Global Thermal Comfort Database ii (the ASHRAE Global Thermal Comfort Database ii). Firstly, a data set of a 24-year-old female user in a summer office building is selected, then data cleaning is carried out according to information such as heat preference, heat acceptability and PMV value in the data set, data inconsistent with a heat sensation index and air temperature are removed, and 190 groups of data are obtained in total. The DENFIS algorithm is used to build a learning model for the thermal comfort of the population of users, as shown in FIG. 15.
The model is modeled by thermal sensation indexes TPI and indoor air temperature and is expressed in a form of a comfort zone, and the main reason is that the group users have different thermal comfort, namely the same thermal sensation index feedback corresponds to different temperatures. The best thermal comfort zone of the group user model is defined by the temperature range mapped by TPI being 0. As can be seen from FIG. 9, the model reasoning results show that the optimal comfortable temperature range of the users in the group is 22.92-24.50 ℃, and the mean square deviation between the model prediction value and the real interaction data is about 1.06 ℃. The error generation mainly comes from personal subjective feedback heat comfort uncertainty in group data, heat resistance of clothes among individuals, metabolism difference and other factors, but the error is about 1 ℃ and is consistent with the field research result, the error result is less than the deviation of comfortable preference temperature among individuals by 2.6 ℃, and the conclusion shows that the model can set the group heat comfort optimum preference temperature with reference value.

Claims (9)

1. The method for establishing the thermal comfort model is characterized by comprising the following steps of:
step 1: collecting a thermal comfort data set, wherein the thermal comfort data set comprises user basic information, thermal environment parameters and thermal voting parameters, and performing data cleaning on the thermal comfort data set to obtain a sample data set, wherein the sample data set comprises a thermal sensation index and an air temperature;
step 2: carrying out ECM clustering on the sample data set to obtain M cluster sets, wherein M is more than or equal to 1;
and step 3: establishing M fuzzy rules, wherein a fuzzy front part of each fuzzy rule is a cluster set, a fuzzy back part of each fuzzy rule is an objective function based on a first-order Takagi-Sugeno model, calculating coefficients of the objective function according to the sample data set obtained in the step 1 and the cluster set obtained in the step 2, performing iterative optimization on the coefficients of the objective function by using a least square method, and obtaining a thermal comfort model according to the optimized fuzzy rules.
2. The thermal comfort modeling method according to claim 1, wherein the user basic information in step 1 includes age, gender, activity status, and clothing thermal resistance, the thermal environment parameters include air temperature, air flow rate, and relative humidity, and the thermal voting parameters include a thermal sensation indicator, a thermal preference, and thermal acceptability.
3. The thermal comfort modeling method according to claim 1, wherein the objective function based on the first order Takagi-Sugeno model is represented by formula i:
f(TPI,T)＝β0+β1×(TPIi,Ti) I-1, 2, …, p is of formula I
Wherein (TPI)i,Ti) I is 1,2, …, p represents all sample data in the sample data set, p is the total number of sample data, TPIiA heat sensation index T representing the ith sample dataiAir temperature representing the ith sample data, β0And β1Satisfy β ═ β0β1]Tβ denotes a coefficient matrix and β ═ a (AWA)-1ATWY, A is the user heat sensation index input matrix and
Figure FDA0002251966810000021
y is an air temperature input matrix and Y ═ T1T2T3… Tp]TW is the cluster set influence factor momentAn array and
Figure FDA0002251966810000022
element W on the diagonal of W1Is the distance, w, from the 1 st sample data to the cluster center to which the sample belongspThe distance from the p sample data to the cluster center of the sample.
4. The setting method of the user preference temperature is characterized by comprising the following steps:
step a: after collecting user basic information, thermal environment parameters and thermal voting parameters of a current user, cleaning data to obtain a thermal sensation index TPI and an air temperature T of the current user;
step b: the thermal comfort modeling method according to any one of claims 1 to 3, obtaining M clusters of step 2, selecting M clusters to which a thermal sensation index TPI and an air temperature T of a current user belong, wherein M is greater than or equal to 1 and less than or equal to M;
step c: the thermal comfort model building method according to any one of claims 1 to 3, obtaining the thermal comfort model of step 3, the thermal comfort model including m fuzzy rules corresponding to the m clusters selected in step b, inputting the thermal sensation index TPI and the air temperature T of the current user into the thermal comfort model, and outputting the preferred temperature of the current user.
5. The method of claim 4, wherein the current user's preferred temperature is output in step c using formula ii:
Figure FDA0002251966810000023
wherein, yoIndicates the preferred temperature, omega, of the current userqRepresents the weight of the current user belonging to the q fuzzy rule and 0 < omegaq＜1。
6. User preferred temperature setting according to claim 5A method of determining ω, wherein ω isqCalculated by a trigonometric membership function.
7. The setting system of the user preference temperature is characterized by comprising a data acquisition module, a data cleaning module, a classification module, a thermal comfort model and a preference temperature setting module;
the data acquisition module is used for acquiring user basic information, thermal environment parameters and thermal voting parameters of a current user and transmitting the acquired information to the data cleaning module;
the data cleaning module is used for cleaning the data acquired by the data acquisition module to obtain the thermal sensation index TPI and the air temperature T of the current user;
the classification module is used for selecting M clusters to which the thermal sensation index TPI and the air temperature T of the current user belong, wherein M is more than or equal to 1 and less than or equal to M, and the M clusters are obtained by the step 2 of the thermal comfort feeling model building method according to any one of claims 1 to 3;
the thermal comfort model is used for inputting a thermal sensation index TPI and an air temperature T of a current user into the thermal comfort model and outputting a preference temperature of the current user, the thermal comfort model is obtained according to the thermal comfort model building method of any one of claims 1 to 3, and the thermal comfort model comprises m fuzzy rules corresponding to the m cluster sets selected in the step b;
the preference temperature setting module is used for setting the indoor temperature according to the preference temperature of the current user output by the thermal comfort model.
8. The method of claim 7, wherein the thermal comfort model outputs the current user's preferred temperature using formula ii:
Figure FDA0002251966810000031
wherein, yoIndicates the preferred temperature, omega, of the current userqIndicating the current userWeights subordinate to the qth fuzzy rule and 0 < omegaq＜1。
9. The setting method of user's preferred temperature according to claim 8, wherein ω is said ωiCalculated by a trigonometric membership function.
CN201911037626.9A 2019-10-29 2019-10-29 Thermal comfort model establishment method and user preference temperature setting method and system Active CN110991478B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201911037626.9A CN110991478B (en) 2019-10-29 2019-10-29 Thermal comfort model establishment method and user preference temperature setting method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201911037626.9A CN110991478B (en) 2019-10-29 2019-10-29 Thermal comfort model establishment method and user preference temperature setting method and system

Publications (2)

Publication Number Publication Date
CN110991478A true CN110991478A (en) 2020-04-10
CN110991478B CN110991478B (en) 2023-06-09

Family

ID=70082718

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201911037626.9A Active CN110991478B (en) 2019-10-29 2019-10-29 Thermal comfort model establishment method and user preference temperature setting method and system

Country Status (1)

Country Link
CN (1) CN110991478B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111898705A (en) * 2020-08-18 2020-11-06 山东省计算中心（国家超级计算济南中心） Fault feature parameter selection method based on fuzzy preference relation and adaptive hierarchical clustering
CN113276622A (en) * 2021-05-26 2021-08-20 上海三一重机股份有限公司 Air volume control method and device for operation machinery cockpit
CN116910482A (en) * 2023-08-08 2023-10-20 雅莹集团股份有限公司 Method and system for predicting cold-proof temperature of down jacket

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2005048185A1 (en) * 2003-11-17 2005-05-26 Auckland University Of Technology Transductive neuro fuzzy inference method for personalised modelling
CN107120782A (en) * 2017-02-28 2017-09-01 上海交通大学 A kind of HVAC system control method based on multi-user's hot comfort data
WO2019063079A1 (en) * 2017-09-28 2019-04-04 Siemens Aktiengesellschaft System, device and method for energy and comfort optimization in a building automation environment

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2005048185A1 (en) * 2003-11-17 2005-05-26 Auckland University Of Technology Transductive neuro fuzzy inference method for personalised modelling
CN107120782A (en) * 2017-02-28 2017-09-01 上海交通大学 A kind of HVAC system control method based on multi-user's hot comfort data
WO2019063079A1 (en) * 2017-09-28 2019-04-04 Siemens Aktiengesellschaft System, device and method for energy and comfort optimization in a building automation environment

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
徐巍等: "基于模糊C-均值聚类与支持向量机的PMV指标预测系统", 《系统工程理论与实践》 *
杨娜: "用户偏好的室内环境舒适度智能控制方法仿真研究", 《科学技术与工程》 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111898705A (en) * 2020-08-18 2020-11-06 山东省计算中心（国家超级计算济南中心） Fault feature parameter selection method based on fuzzy preference relation and adaptive hierarchical clustering
CN111898705B (en) * 2020-08-18 2023-04-25 山东省计算中心（国家超级计算济南中心） Fault feature parameter selection method based on fuzzy preference relation and self-adaptive hierarchical clustering
CN113276622A (en) * 2021-05-26 2021-08-20 上海三一重机股份有限公司 Air volume control method and device for operation machinery cockpit
CN116910482A (en) * 2023-08-08 2023-10-20 雅莹集团股份有限公司 Method and system for predicting cold-proof temperature of down jacket
CN116910482B (en) * 2023-08-08 2024-05-28 雅莹集团股份有限公司 Method and system for predicting cold-proof temperature of down jacket

Also Published As

Publication number Publication date
CN110991478B (en) 2023-06-09

Similar Documents

Publication Publication Date Title
Fard et al. Application of machine learning in thermal comfort studies: A review of methods, performance and challenges
Lu et al. Data-driven simulation of a thermal comfort-based temperature set-point control with ASHRAE RP884
Deng et al. Artificial neural network models using thermal sensations and occupants’ behavior for predicting thermal comfort
Langevin et al. Simulating the human-building interaction: Development and validation of an agent-based model of office occupant behaviors
Lee et al. A Bayesian approach for probabilistic classification and inference of occupant thermal preferences in office buildings
Wu et al. Using an ensemble machine learning methodology-Bagging to predict occupants’ thermal comfort in buildings
Jung et al. Occupant-centered real-time control of indoor temperature using deep learning algorithms
Brik et al. An IoT-based deep learning approach to analyse indoor thermal comfort of disabled people
Pazhoohesh et al. A satisfaction-range approach for achieving thermal comfort level in a shared office
CN110991478B (en) Thermal comfort model establishment method and user preference temperature setting method and system
CN112613232A (en) Indoor human body thermal comfort prediction and evaluation method under winter heating condition
Cakir et al. A bayesian deep neural network approach to seven-point thermal sensation perception
Brik et al. ThermCont: A machine Learning enabled Thermal Comfort Control Tool in a real time
Sun et al. Energy consumption optimization of building air conditioning system via combining the parallel temporal convolutional neural network and adaptive opposition-learning chimp algorithm
Li et al. Data-driven modeling and optimization of thermal comfort and energy consumption using type-2 fuzzy method
Fay et al. Gaussian Process models for ubiquitous user comfort preference sampling; global priors, active sampling and outlier rejection
Martins et al. Personal thermal comfort models: A deep learning approach for predicting older people’s thermal preference
Yuan et al. Modelling method for air-conditioning usage behavior in multi-occupant office space based on group decision-making strategy
Khalil et al. An IoT environment for estimating occupants’ thermal comfort
Kanthila et al. Markov chain-based algorithms for building occupancy modeling: A review
Yan et al. Occupant behavior impact in buildings and the artificial intelligence-based techniques and data-driven approach solutions
Suman et al. Toward personalization of user preferences in partially observable smart home environments
Lian Revisiting thermal comfort and thermal sensation
Chan et al. Wearing comfort of two construction work uniforms
Zhou et al. A model-driven learning approach for predicting the personalized dynamic thermal comfort in ordinary office environment

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant