A method and system for constructing a thermal comfort prediction model

By collecting and classifying natural and human-made factors of thermal comfort, combining KMC and moth to flame algorithm, a prediction model for thermal comfort affecting factors is constructed, which solves the problem of incomplete factor analysis in the existing technology and achieves thermal comfort Accurate prediction.

CN118643363BActive Publication Date: 2024-11-26TAIYUAN UNIVERSITY OF TECHNOLOGY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410716855.8
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-06-04
Publication Date
2024-11-26
Estimated Expiration
2044-06-04

AI Technical Summary

Technical Problem

The existing thermal comfort prediction model is difficult to fully consider natural and human factors, resulting in inaccurate prediction results and insufficient analysis in the case of large differences in indoor and outdoor environments.

Method used

By collecting natural and human-made influencing factors data of thermal comfort, using KMC algorithm for classification and clustering, building a comprehensive comfort level, combining the moth to the flame algorithm to optimize the clustering center, building a data prediction model for thermal comfort influencing factors to achieve the Accurate prediction of future influencing factors.

Benefits of technology

It improves the accuracy and precision of thermal comfort prediction, can accurately predict people's comfort conditions in different environments, and enhances the ability to predict future comfort.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118643363B_ABST
    Figure CN118643363B_ABST
Patent Text Reader

Abstract

The present invention discloses a method and system for constructing a thermal comfort prediction model, and relates to the field of data model construction. The present invention first collects historical data of natural influencing factors of thermal comfort and artificial influencing factors of thermal comfort, as well as people's comprehensive comfort levels when these data appear, and classifies the historical data of the influencing factors by adopting a KMC algorithm, and matches the classified categories with the comprehensive comfort levels, thereby constructing a corresponding relationship between the comprehensive comfort levels and the historical data of the influencing factors, and then predicts the data of the influencing factors that will appear in the future based on the collected historical data of the influencing factors, thereby obtaining predicted data; the predicted data is classified, and matched with the comprehensive comfort levels according to their categories, thereby obtaining people's comfort conditions when these influencing factor data appear in the future; thereby achieving the purpose of predicting people's comfort in the future.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the field of data model construction, and in particular, relates to a method and system for constructing a thermal comfort prediction model. Background Art

[0002] A thermal comfort prediction model is a mathematical model used to evaluate and predict the impact of indoor or vehicle thermal environments on human comfort. These models are usually based on human thermal physiology principles, environmental conditions, and human activity levels to simulate and predict people's comfort in a specific thermal environment. Thermal comfort prediction models can be simple empirical formulas or complex multivariable dynamic models, such as numerical simulations and artificial intelligence algorithms. These models help optimize the thermal environment during design and operation to meet people's comfort needs and energy conservation and emission reduction goals. They are widely used in fields such as building environmental engineering, vehicle engineering, and ergonomics.

[0003] Chinese patent CN117332308A discloses a method for predicting human thermal comfort based on deep learning. First, a data set is obtained. Each sample of the data set includes an environmental parameter sequence and an individual physiological parameter sequence. The sample labels include subjective thermal sensation and comfort. Then, the data set is preprocessed and the preprocessed data is sliced ​​to obtain a number of data slices. Then, a thermal sensation classification model is constructed to extract deep features reflecting human thermal sensation and prediction of thermal sensation. A thermal comfort classification model and a thermal comfort classification model are constructed. Finally, the trained thermal sensation classification model, thermal comfort classification model and thermal comfort classification model are reorganized to obtain a thermal comfort prediction model. According to the thermal sensation prediction results of the thermal sensation classification model, the deep features are input into the thermal comfort classification model or the thermal comfort classification model to predict comfort. This method follows the neural mechanism of human body temperature regulation, distinguishes the comfort under different sensations, and predicts thermal sensation and comfort respectively, which is more in line with the subjective real feelings of individuals.

[0004] Thermal comfort is related to many influencing factors, such as humidity, wind speed, light, etc., which will affect thermal comfort, and environmental comfort will have a certain impact on human emotions and physiology; if environmental comfort cannot be predicted, people will not be able to prepare in advance to deal with the emergence of various external influencing factors, which will affect and inconvenience people's activities. Summary of the invention

[0005] In view of the problems in the related art, the present invention proposes a method and system for constructing a thermal comfort prediction model to overcome the above-mentioned technical problems existing in the existing related art.

[0006] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0007] The present invention is a method for constructing a thermal comfort prediction model, comprising the following steps:

[0008] S1. Collect data on natural influencing factors of thermal comfort and data on human influencing factors of thermal comfort to obtain a first natural influencing factor data matrix, a second natural influencing factor data matrix, a first human influencing factor data matrix, and a second human influencing factor data matrix; then set a comprehensive comfort level set and collect comprehensive comfort level data;

[0009] S2. Construct a classification model of thermal comfort influencing factors, wherein the classification model of thermal comfort influencing factors uses a KMC algorithm to classify the first natural influencing factor data matrix, the second natural influencing factor data matrix, the first human influencing factor data matrix, and the second human influencing factor data matrix and calculate the cluster center of each classified data set to obtain a first final classification matrix and a second final classification matrix as well as a first final cluster center set and a second final cluster center set, and at the same time set a corresponding relationship between the comprehensive comfort level data and the initial classification to obtain a first initial classification and comfort level corresponding relationship set and a second initial classification and comfort level corresponding relationship set;

[0010] S3, reconstructing a data prediction model for thermal comfort influencing factors, wherein the data prediction model for thermal comfort influencing factors predicts the first natural influencing factor data matrix, the first human influencing factor data matrix, the second natural influencing factor data matrix, and the second human influencing factor data matrix to obtain prediction results, and calculates the minimum Euclidean distance and the maximum Euclidean distance from each data in the first final classification matrix to the first final cluster center set and from each data in the second final classification matrix to the second final cluster center set, to obtain a first maximum and minimum Euclidean distance interval set and a second maximum and minimum Euclidean distance interval set;

[0011] S4. Finally, a thermal comfort level determination model is constructed. The thermal comfort level determination model combines the prediction results of the thermal comfort influencing factor data prediction model to obtain a first prediction result data matrix and a second prediction result data matrix; the first prediction result data matrix and the second prediction result data matrix are classified to obtain a first prediction data classification result set and a second prediction data classification result set; and then a first prediction comfort level data set and a second prediction comfort level data set are obtained according to the first prediction data classification result set and the second prediction data classification result set;

[0012] Since thermal comfort is influenced by both natural and human factors, the scheme of the present invention collects data on natural and human factors affecting thermal comfort, analyzes the data on both natural and human factors, and makes the analysis more comprehensive and accurate. At the same time, a comprehensive comfort level is used as a criterion for judging thermal comfort. The comprehensive comfort level is preferably divided into seven levels, including very uncomfortable: -3; uncomfortable: -2; slightly uncomfortable: -1; moderate: 0; slightly comfortable: 1; comfortable: 2; very comfortable: 3. By adopting the comprehensive comfort level, people's comfort can be quantified, thereby facilitating subsequent prediction operations. First, by collecting data on natural factors affecting thermal comfort, the thermal comfort level is analyzed. As well as the historical data of artificial influencing factors of thermal comfort, and the comprehensive comfort level of people when these data appear, the historical data of the influencing factors are classified by using the KMC algorithm, and the classified categories are matched with the comprehensive comfort level. Therefore, in addition to constructing the correspondence between the comprehensive comfort level and the historical data of the influencing factors, the data of the influencing factors that will appear in the future are predicted based on the collected historical data of the influencing factors, so as to obtain the predicted data; the predicted data are classified and matched with the comprehensive comfort level according to their categories, so as to obtain the comfort of people when these influencing factor data appear in the future; the purpose of predicting people's comfort in the future is achieved.

[0013] Preferably, the S1 comprises the following steps:

[0014] S11. Set the natural influencing factors of thermal comfort a = {a1, a2, ..., a i ,...,a a′} and the set of human factors affecting thermal comfort a i represents the i-th natural influencing factor of thermal comfort, and a′ represents the total number of natural influencing factors of thermal comfort; represents the set i-th thermal comfort human influencing factor, Indicates the total number of artificial influencing factors of thermal comfort;

[0015] Then set the indoor space b1 and the outdoor space b2;

[0016] S12: Set a historical data collection cycle, divide the historical data collection cycle into time points, and obtain a historical data collection time point set. c i ′ represents the i-th time point divided from the historical data collection period, Indicates the total number of time points divided into the historical data collection period;

[0017] According to the historical data collection time point set and the set of natural factors affecting thermal comfort a={a1,a2,...,a i ,...,a a′} Collect the data of natural influencing factors of thermal comfort in indoor space b1 and outdoor space b2 respectively, and obtain the first natural influencing factor data matrix and the second natural influencing factor data matrix d; and then according to the historical data collection time point set and thermal comfort human influencing factors The thermal comfort human influence factor data are collected in the indoor space b1 and the outdoor space b2 respectively to obtain the first human influence factor data matrix d′ and the second human influence factor data matrix The first natural influencing factor data matrix The second natural influencing factor data matrix d, the first artificial influencing factor data matrix d′ and the second artificial influencing factor data matrix They are as follows:

[0018]

[0019]

[0020] in, represents the data collected from the indoor space b1 at the i-th time point on the j-th natural influencing factor of thermal comfort; d ij represents the data collected from the outdoor space b2 at the i-th time point on the j-th natural influencing factor of thermal comfort; d i ' j represents the data collected from the indoor space b1 at the i-th time point on the j-th thermal comfort artificial influencing factor; represents the data collected from the outdoor space b2 at the i-th time point on the j-th thermal comfort human influencing factor;

[0021] S13. Set comprehensive comfort level set b i ′ represents the set i-th comprehensive comfort level, Indicates the total number of set comprehensive comfort levels;

[0022] According to the comprehensive comfort level set Historical data collection time point set And the corresponding comprehensive comfort level data of indoor space b1 is collected to obtain the first comprehensive comfort level data set represents the comprehensive comfort level data collected from the indoor space b1 at the i-th time point;

[0023] According to the comprehensive comfort level set Historical data collection time point set And the corresponding comprehensive comfort level data of outdoor space b2 is collected to obtain the second comprehensive comfort level data set c i represents the comprehensive comfort level data collected from the outdoor space b2 at the i-th time point;

[0024] Natural factors affecting thermal comfort include indoor air temperature, relative humidity, air velocity and environmental radiation, etc.; human factors affecting thermal comfort include human activity level, clothing thermal resistance and the number of people, etc.; since the indoor and outdoor environments are quite different, this solution sets indoor and outdoor spaces to analyze the indoor and outdoor conditions separately, making the analysis more detailed and accurate; and setting a set of historical data collection time points provides a basis for the subsequent collection of influencing factor data.

[0025] Preferably, S2 comprises the following steps:

[0026] S21, constructing a classification model of thermal comfort influencing factors, wherein the classification model of thermal comfort influencing factors firstly classifies the first natural influencing factor data matrix and the first human influence factor data matrix d′ to obtain a first classification training data matrix; then the second natural influence factor data matrix d and the second human influence factor data matrix Merge to obtain the second classification training data matrix;

[0027] S22: The thermal comfort influencing factor classification model sets a first KMC algorithm model and a second KMC algorithm model, and uses the first KMC algorithm model to classify the first classification training data matrix to obtain a first initial classification matrix as follows,

[0028]

[0029] in, represents the jth influencing factor data set in the i-th category of the first initial classification set, Represents the total number of influencing factor data sets in the i-th category of the first initial classification set;

[0030] Then the second KMC algorithm model is used to classify the second classification training data matrix to obtain the second initial classification matrix f; as follows,

[0031]

[0032] Among them, f ijrepresents the jth influencing factor dataset in the i-th category of the second initial classification set, f i ′ represents the total number of influencing factor data sets in the i-th category of the second initial classification set;

[0033] Calculate the first initial classification matrix respectively The cluster centers of each classification data set in the second initial classification matrix f are optimized to obtain the first final cluster center set and the second final cluster center set

[0034] S23: The thermal comfort influencing factor classification model sets the first comprehensive comfort level data set With the first initial classification matrix The correspondence between the first initial classification and the comfort level is obtained. and Respectively represent the i-th category in the first initial classification set and the corresponding comprehensive comfort level data;

[0035] Then set the second comprehensive comfort level data set The corresponding relationship between the second initial classification matrix f and the comfort level is obtained. and Respectively represent the i-th category in the second initial classification set and the corresponding comprehensive comfort level data;

[0036] The KMC algorithm model is used to classify the collected influencing factor data, so that the corresponding comprehensive comfort level can be determined according to the classification results.

[0037] Preferably, in S22, using the first KMC algorithm model to classify the first classification training data matrix and using the second KMC algorithm model to classify the second classification training data matrix and calculate and optimize the cluster center of each classification specifically includes the following steps:

[0038] S221: construct a first initial cluster center set according to the first classification training data matrix and the second classification training data matrix And the second initial cluster center set g 1i represents the cluster center corresponding to the i-th comprehensive comfort level in the first initial cluster center set, g 2i represents the cluster center corresponding to the i-th comprehensive comfort level in the second initial cluster center set;

[0039] S222, respectively calculate the first initial cluster center set for each row of data in the first classification training data matrix The Euclidean distance of each cluster center in the second classification training data matrix and the distance from each row of data to the second initial cluster center set The Euclidean distance of each cluster center in is obtained to obtain the first Euclidean distance data matrix g1′ and the second Euclidean distance data matrix g′2;

[0040] The influencing factor classification training data corresponding to the minimum data in each row of the first Euclidean distance data matrix g1′ and the second Euclidean distance data matrix g′2 and the corresponding cluster center are taken as the same category to obtain a first initial classification matrix and a second initial classification matrix;

[0041] S223, using the moth-to-flame algorithm to and the second initial cluster center set Optimize and get the first final cluster center set and the second final cluster center set represents the cluster center corresponding to the i-th comprehensive comfort level in the first final cluster center set, represents the cluster center corresponding to the i-th comprehensive comfort level in the second initial cluster center set;

[0042] Firstly, the KMC algorithm model is used to classify the collected influencing factor data, and then the cluster centers of each category are calculated by the moth-to-flame algorithm, thus providing a comparative basis for the subsequent classification of the predicted data.

[0043] Preferably, the S221 includes the following steps:

[0044] S2211, setting a first initial cluster center set g1 and a second initial cluster center set g2;

[0045] S2212, randomly selecting a row of data from the first classification training data matrix and the second classification training data matrix and putting them into the first initial cluster center set g1 and the second initial cluster center set g2; at the same time, deleting the selected data from the first classification training data matrix and the second classification training data matrix to obtain a first updated first classification training data matrix and a first updated second classification training data matrix;

[0046] S2213, respectively calculating the Euclidean distances of each data in the first updated first classification training data matrix and the first updated second classification training data matrix to the data in the first initial cluster center set g1 and the second initial cluster center set g2, selecting the classification training data corresponding to the largest Euclidean distance and putting them into the first initial cluster center set g1 and the second initial cluster center set g2, and deleting them from the first updated first classification training data matrix and the first updated second classification training data matrix, to obtain the second updated first classification training data matrix and the second updated second classification training data matrix;

[0047] S2214, respectively calculating the Euclidean distances of each data in the second updated first classification training data matrix and the second updated second classification training data matrix to each data in the first initial cluster center set g1 and the second initial cluster center set g2, selecting the largest Euclidean distance and the smallest Euclidean distance and multiplying them, putting the classification training data matrix corresponding to the data with the largest multiplication result into the first initial cluster center set g1 and the second initial cluster center set g2, and deleting it from the second updated first classification training data matrix and the second updated second classification training data matrix, to obtain the third updated first classification training data matrix and the third updated second classification training data matrix;

[0048] S2215: When the number of data in the first initial cluster center set g1 and the second initial cluster center set g2 is greater than or equal to When , stop putting in; otherwise, repeat S2214;

[0049] By adopting the maximum and minimum distance product method, the randomness of the initial cluster center selection when using the KMC algorithm model can be avoided, thereby improving the clustering accuracy and further improving the accuracy of subsequent predictions of people's comfort in the future.

[0050] Preferably, the S223 comprises the following steps:

[0051] S2231, constructing a first moth population and a second moth population; setting the sizes of the first moth population and the second moth population to be and Then the first moth population and the second moth population are represented as and h 1i and h 2i denote the i-th moth in the first moth population and the i-th moth in the second moth population respectively; set the first maximum number of iterations h1′ and the second maximum number of iterations h2′; set the search space dimensions of the first moth population and the second moth population to be dimension;

[0052] S2232, randomly select a piece of data from each category of the first initial classification matrix and combine them as the initial position of the first moth in the first moth population; randomly select a piece of data from each category of the second initial classification matrix and combine them as the initial position of the first moth in the second moth population; repeat the above operation to obtain the first initial position set and the second initial position set Represents the j-th dimension coordinate data of the initial position of the i-th moth in the first moth population; Represents the j-th dimension coordinate data of the initial position of the i-th moth in the second moth population;

[0053] Set the fitness functions of the first moth population and the second moth population as follows:

[0054]

[0055] In the formula, and Respectively represent the fitness functions of the first moth population and the second moth population; l1 represents the sum of the Euclidean distances between each piece of data in the first initial classification matrix and the moth positions in the first moth population, and l2 represents the sum of the Euclidean distances between each piece of data in the second initial classification matrix and the moth positions in the second moth population;

[0056] S2233, start the iteration operation, set the current iteration number of the first moth population to l1′ and the current iteration number of the second moth population to l2′; calculate the global optimal fitness value of the first moth population and the second moth population and the corresponding global optimal moth individual according to the fitness function of the first moth population and the second moth population in each round of iteration; and update the position of each moth in the first moth population and the second moth population according to the global optimal fitness value and repeat S222;

[0057] When l1′≥h1′, the first moth population stops the iteration operation and outputs the first final cluster center set and the first final classification matrix;

[0058] When l2′≥h2′, the second moth population stops the iteration operation and outputs the second final cluster center set and the second final classification matrix;

[0059] The moth algorithm is used to calculate the cluster center of each category, which improves the convergence speed and optimization accuracy when optimizing the cluster center; the inverse of the maximum distance within the class is used as the fitness function to guide the optimization of the cluster center in the KMC iteration process and continuously improve the clustering accuracy.

[0060] Preferably, S3 comprises the following steps:

[0061] S31, reconstructing a thermal comfort influencing factor data prediction model, wherein the thermal comfort influencing factor data prediction model uses a BP neural network model to respectively calculate the first natural influencing factor data matrix The first artificial influencing factor data matrix d′ and the second natural influencing factor data matrix d, the second artificial influencing factor data matrix Make predictions and obtain the first prediction result natural data matrix The first prediction result is an artificial data matrix and the second predicted natural data matrix m and the second predicted artificial data matrix m′; respectively,

[0062]

[0063] in, represents the predicted data of the jth natural influencing factor of thermal comfort in the indoor space b1 at the i-th time point in the future; represents the predicted data of the jth thermal comfort human factors influencing the indoor space b1 at the i-th time point in the future; m ij represents the predicted data of the jth natural influencing factor of thermal comfort in the outdoor space b2 at the i-th time point in the future; m i ' j represents the predicted data of the jth thermal comfort human factors influencing the outdoor space b2 at the i-th time point in the future; Indicates the total number of predicted future time points;

[0064] S32: The thermal comfort influencing factor data prediction model calculates each data in the first final classification matrix to the first final cluster center set The minimum Euclidean distance and the maximum Euclidean distance of the first maximum and minimum Euclidean distance interval set are obtained. and Respectively represent the minimum Euclidean distance and the maximum Euclidean distance from the classified data in the ith classification in the first final classification matrix to the ith cluster center in the first final cluster center set;

[0065] Calculate each data in the second final classification matrix to the second final cluster center set The maximum Euclidean distance and the minimum Euclidean distance of the second maximum and minimum Euclidean distance interval set are obtained. n i min and n i maxRespectively represent the minimum Euclidean distance and the maximum Euclidean distance from the classified data in the ith category in the second final classification matrix to the ith cluster center in the second final cluster center set;

[0066] The BP neural network model is used to predict the future influencing factor data to obtain the predicted data. Then the maximum and minimum Euclidean distances from each data in each class divided by the historical influencing factor data in the previous step to the cluster center are calculated to determine the scope of each classification for subsequent comparison.

[0067] Preferably, S4 comprises the following steps:

[0068] S41, finally constructing a thermal comfort level determination model, wherein the thermal comfort level determination model is used to determine the first prediction result natural data matrix The first prediction result is an artificial data matrix Merge to obtain a first prediction result data matrix; then merge the second prediction natural data matrix m and the second prediction artificial data matrix m′ to obtain a second prediction result data matrix;

[0069] S42, the thermal comfort level determination model calculates each row of data in the first prediction result data matrix to the first final cluster center set The Euclidean distances of each cluster center in the second prediction result data matrix are calculated to obtain the first Euclidean distance data matrix n′; and then the distances of each row of data in the second prediction result data matrix to the second final cluster center set are calculated. The Euclidean distance of each cluster center in the second Euclidean distance data matrix is ​​obtained. The n′ and They are as follows:

[0070]

[0071] Among them, n i ' j represents the Euclidean distance from the i-th row of data in the first prediction result data matrix to the j-th cluster center in the first final cluster center set; represents the Euclidean distance from the i-th row of data in the second prediction result data matrix to the j-th cluster center in the second final cluster center set;

[0072] S43: The thermal comfort level determination model selects the first Euclidean distance data matrix n′ and the second Euclidean distance data matrix respectively. Each of the rows in the set lies in the first maximum and minimum Euclidean distance interval And the second largest and smallest Euclidean distance interval set The data in the interval at the corresponding position in is obtained to obtain the first predicted initial Euclidean distance data matrix and the second predicted initial Euclidean distance data matrix;

[0073] S44, the thermal comfort level determination model selects the minimum value of each row of data in the first initial Euclidean distance data matrix and the second initial Euclidean distance data matrix to obtain the first final Euclidean distance data set and the second final Euclidean distance data set; according to the influencing factor data corresponding to each Euclidean distance in the first final Euclidean distance data set and the second final Euclidean distance data set and the classification to which it belongs, the first prediction data classification result set is obtained and the second prediction data classification result set

[0074] and i Respectively represent the categories to which the influencing factor data at the i-th time point in the first prediction result data matrix and the second prediction result data matrix belong;

[0075] S45: the thermal comfort level determination model classifies the result set according to the first prediction data and the second prediction data classification result set And the corresponding relationship set between the first initial classification and the comfort level and the corresponding relationship set between the second initial classification and the comfort level Obtaining a first predicted comfort level data set and a second predicted comfort level data set;

[0076] By comparing the predicted data with the obtained Euclidean distance interval of each category, the category of the predicted data is obtained, and the comfort level corresponding to the predicted data is determined.

[0077] A system for constructing a thermal comfort prediction model, comprising a thermal comfort natural influencing factor data collection module, a thermal comfort human influencing factor data collection module, a comprehensive comfort level data collection module, a first classification module, a cluster center calculation module, an influencing factor data prediction module, a second classification module, and a predicted comfort level data determination module;

[0078] The thermal comfort natural influencing factor data collection module is used to collect thermal comfort natural influencing factor data;

[0079] The thermal comfort human influencing factor data collection module is used to collect thermal comfort human influencing factor data;

[0080] The first classification module is used to classify the first natural influencing factor data matrix, the second natural influencing factor data matrix, the first human influencing factor data matrix and the second human influencing factor data matrix using the KMC algorithm;

[0081] The cluster center calculation module is used to calculate the cluster center of each classification data set;

[0082] The influencing factor data prediction module is used to predict the first natural influencing factor data matrix, the first artificial influencing factor data matrix, the second natural influencing factor data matrix, and the second artificial influencing factor data matrix;

[0083] The second classification module is used to classify the first prediction result data matrix and the second prediction result data matrix;

[0084] The predicted comfort level data determination module is used to obtain a first predicted comfort level data set and a second predicted comfort level data set according to the first predicted data classification result set and the second predicted data classification result set.

[0085] The present invention has the following beneficial effects:

[0086] 1. The present invention sets a data collection module for natural influencing factors of thermal comfort, a data collection module for artificial influencing factors of thermal comfort, a comprehensive comfort level data collection module, a first classification module, a cluster center calculation module, an influencing factor data prediction module, a second classification module and a predicted comfort level data determination module; firstly, by collecting historical data of various natural influencing factors of thermal comfort and various artificial influencing factors of thermal comfort, the influence of various external factors and people's comprehensive comfort level when these data appear are fully considered, and the data of various external factors are classified and predicted, so that the prediction result is more accurate; at the same time, by setting the indoor space and indoor For outdoor space, the indoor and outdoor situations are analyzed separately, making the analysis more detailed and accurate; the historical data of influencing factors are classified by using the KMC algorithm, and the classified categories are matched with the comprehensive comfort level. Therefore, in addition to the correspondence between the comprehensive comfort level and the historical data of influencing factors, the corresponding relationship between the comprehensive comfort level and the historical data of influencing factors is constructed, and then the data of future influencing factors are predicted based on the collected historical data of influencing factors, so as to obtain predicted data; the predicted data are classified and matched with the comprehensive comfort level according to their categories, so as to obtain the comfort of people when these influencing factor data appear in the future; thus, the purpose of predicting people's comfort in the future is achieved.

[0087] 2. In the present invention, the collected influencing factor data are classified by adopting the KMC algorithm model, so as to facilitate the subsequent determination of the corresponding comprehensive comfort level according to the classification results, and then the cluster centers of each category are calculated by the moth-to-flame algorithm, thereby providing a comparison basis for the subsequent classification of the predicted data; among them, the maximum and minimum distance product method can be well avoided to avoid the randomness of the selection of the initial cluster center when adopting the KMC algorithm model, thereby improving the clustering accuracy, and then improving the accuracy of the subsequent prediction of people's comfort in the future.

[0088] 3. In the present invention, the moth algorithm is used to calculate the cluster center of each category, thereby improving the convergence speed and optimization accuracy when optimizing the cluster center; the inverse of the maximum distance within the class is used as the fitness function to guide the optimization of the cluster center in the KMC iteration process, thereby continuously improving the clustering accuracy.

[0089] Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described above at the same time. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] In order to more clearly illustrate the technical solutions of the embodiments of the invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the invention. For ordinary technicians in this field, they can also obtain drawings based on these drawings without paying creative work.

[0091] Figure 1 A schematic diagram of a flow chart of constructing a thermal comfort prediction model by a construction system of a thermal comfort prediction model according to the present invention. DETAILED DESCRIPTION

[0092] The following will be combined with the drawings in the embodiments of the invention to clearly and completely describe the technical solutions in the embodiments of the invention. Obviously, the described embodiments are only part of the embodiments of the invention, not all of the embodiments. Based on the embodiments in the invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the invention.

[0093] In the description of the present invention, it is necessary to understand that the terms "opening", "upper", "lower", "top", "middle", "inside" and the like indicating orientation or positional relationship are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the components or elements referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as limiting the invention.

[0094] The present invention is a method for constructing a thermal comfort prediction model, comprising the following steps:

[0095] S1. Collect data on natural influencing factors of thermal comfort and data on human influencing factors of thermal comfort to obtain a first natural influencing factor data matrix, a second natural influencing factor data matrix, a first human influencing factor data matrix, and a second human influencing factor data matrix; then set a comprehensive comfort level set and collect comprehensive comfort level data;

[0096] The S1 comprises the following steps:

[0097] S11. Set the natural influencing factors of thermal comfort a = {a1, a2, ..., a i ,...,a a′} and the set of human factors affecting thermal comfort a i represents the i-th natural influencing factor of thermal comfort, and a′ represents the total number of natural influencing factors of thermal comfort; represents the set i-th thermal comfort human influencing factor, Indicates the total number of artificial influencing factors of thermal comfort;

[0098] Then set the indoor space b1 and the outdoor space b2;

[0099] S12: Set a historical data collection cycle, divide the historical data collection cycle into time points, and obtain a historical data collection time point set. c i ′ represents the i-th time point divided from the historical data collection period, Indicates the total number of time points divided into the historical data collection period;

[0100] According to the historical data collection time point set and the set of natural factors affecting thermal comfort a={a1,a2,...,a i ,...,a a′} Collect the data of natural influencing factors of thermal comfort in indoor space b1 and outdoor space b2 respectively, and obtain the first natural influencing factor data matrix and the second natural influencing factor data matrix d; and then according to the historical data collection time point set and thermal comfort human influencing factors The thermal comfort human influence factor data are collected in the indoor space b1 and the outdoor space b2 respectively to obtain the first human influence factor data matrix d′ and the second human influence factor data matrix The first natural influencing factor data matrix The second natural influencing factor data matrix d, the first artificial influencing factor data matrix d′ and the second artificial influencing factor data matrix They are as follows:

[0101]

[0102] in, represents the data collected from the indoor space b1 at the i-th time point on the j-th natural influencing factor of thermal comfort; d ij represents the data collected from the outdoor space b2 at the i-th time point on the j-th natural influencing factor of thermal comfort; d i ' j represents the data collected from the indoor space b1 at the i-th time point on the j-th thermal comfort artificial influencing factor; represents the data collected from the outdoor space b2 at the i-th time point on the j-th thermal comfort human influencing factor;

[0103] S13. Set comprehensive comfort level set b i ′ represents the set i-th comprehensive comfort level, Indicates the total number of set comprehensive comfort levels;

[0104] According to the comprehensive comfort level set Historical data collection time point set And the corresponding comprehensive comfort level data of indoor space b1 is collected to obtain the first comprehensive comfort level data set represents the comprehensive comfort level data collected from the indoor space b1 at the i-th time point;

[0105] According to the comprehensive comfort level set Historical data collection time point set And the corresponding comprehensive comfort level data of outdoor space b2 is collected to obtain the second comprehensive comfort level data set c i represents the comprehensive comfort level data collected from the outdoor space b2 at the i-th time point;

[0106] S2. Construct a classification model of thermal comfort influencing factors, wherein the classification model of thermal comfort influencing factors uses a KMC algorithm to classify the first natural influencing factor data matrix, the second natural influencing factor data matrix, the first human influencing factor data matrix, and the second human influencing factor data matrix and calculate the cluster center of each classified data set to obtain a first final classification matrix and a second final classification matrix as well as a first final cluster center set and a second final cluster center set, and at the same time set a corresponding relationship between the comprehensive comfort level data and the initial classification to obtain a first initial classification and comfort level corresponding relationship set and a second initial classification and comfort level corresponding relationship set;

[0107] The S2 comprises the following steps:

[0108] S21, constructing a classification model of thermal comfort influencing factors, wherein the classification model of thermal comfort influencing factors firstly classifies the first natural influencing factor data matrix and the first human influence factor data matrix d′ to obtain a first classification training data matrix; then the second natural influence factor data matrix d and the second human influence factor data matrix Merge to obtain the second classification training data matrix;

[0109] S22: The thermal comfort influencing factor classification model sets a first KMC algorithm model and a second KMC algorithm model, and uses the first KMC algorithm model to classify the first classification training data matrix to obtain a first initial classification matrix as follows,

[0110]

[0111] in, represents the jth influencing factor data set in the i-th category of the first initial classification set, Represents the total number of influencing factor data sets in the i-th category of the first initial classification set;

[0112] Then the second KMC algorithm model is used to classify the second classification training data matrix to obtain the second initial classification matrix f; as follows,

[0113]

[0114] Among them, f ij represents the jth influencing factor dataset in the i-th category of the second initial classification set, f i ′ represents the total number of influencing factor data sets in the i-th category of the second initial classification set;

[0115] Calculate the first initial classification matrix respectively The cluster centers of each classification data set in the second initial classification matrix f are optimized to obtain the first final cluster center set and the second final cluster center set

[0116] In S22, using the first KMC algorithm model to classify the first classification training data matrix and using the second KMC algorithm model to classify the second classification training data matrix and calculate and optimize the cluster center of each classification specifically includes the following steps:

[0117] S221: construct a first initial cluster center set according to the first classification training data matrix and the second classification training data matrix And the second initial cluster center set g 1i represents the cluster center corresponding to the i-th comprehensive comfort level in the first initial cluster center set, g 2i represents the cluster center corresponding to the i-th comprehensive comfort level in the second initial cluster center set;

[0118] The S221 includes the following steps:

[0119] S2211, setting a first initial cluster center set g1 and a second initial cluster center set g2;

[0120] S2212, randomly selecting a row of data from the first classification training data matrix and the second classification training data matrix and putting them into the first initial cluster center set g1 and the second initial cluster center set g2; at the same time, deleting the selected data from the first classification training data matrix and the second classification training data matrix to obtain a first updated first classification training data matrix and a first updated second classification training data matrix;

[0121] S2213, respectively calculating the Euclidean distances of each data in the first updated first classification training data matrix and the first updated second classification training data matrix to the data in the first initial cluster center set g1 and the second initial cluster center set g2, selecting the classification training data corresponding to the largest Euclidean distance and putting them into the first initial cluster center set g1 and the second initial cluster center set g2, and deleting them from the first updated first classification training data matrix and the first updated second classification training data matrix, to obtain the second updated first classification training data matrix and the second updated second classification training data matrix;

[0122] S2214, respectively calculating the Euclidean distances of each data in the second updated first classification training data matrix and the second updated second classification training data matrix to each data in the first initial cluster center set g1 and the second initial cluster center set g2, selecting the largest Euclidean distance and the smallest Euclidean distance and multiplying them, putting the classification training data matrix corresponding to the data with the largest multiplication result into the first initial cluster center set g1 and the second initial cluster center set g2, and deleting it from the second updated first classification training data matrix and the second updated second classification training data matrix, to obtain the third updated first classification training data matrix and the third updated second classification training data matrix;

[0123] S2215: When the number of data in the first initial cluster center set g1 and the second initial cluster center set g2 is greater than or equal to When , stop putting in; otherwise, repeat S2214;

[0124] S222, respectively calculate the first initial cluster center set for each row of data in the first classification training data matrix The Euclidean distance of each cluster center in the second classification training data matrix and the distance from each row of data to the second initial cluster center set The Euclidean distance of each cluster center in is obtained to obtain the first Euclidean distance data matrix g1′ and the second Euclidean distance data matrix g′2;

[0125] The influencing factor classification training data corresponding to the minimum data in each row of the first Euclidean distance data matrix g1′ and the second Euclidean distance data matrix g′2 and the corresponding cluster center are taken as the same category to obtain a first initial classification matrix and a second initial classification matrix;

[0126] S223, using the moth-to-flame algorithm to and the second initial cluster center set Optimize and get the first final cluster center set and the second final cluster center set represents the cluster center corresponding to the i-th comprehensive comfort level in the first final cluster center set, represents the cluster center corresponding to the i-th comprehensive comfort level in the second initial cluster center set;

[0127] The S223 comprises the following steps:

[0128] S2231, constructing a first moth population and a second moth population; setting the sizes of the first moth population and the second moth population to be and Then the first moth population and the second moth population are represented as and h 1i and h 2i denote the i-th moth in the first moth population and the i-th moth in the second moth population respectively; set the first maximum number of iterations h1′ and the second maximum number of iterations h2′; set the search space dimensions of the first moth population and the second moth population to be dimension;

[0129] S2232, randomly select a piece of data from each category of the first initial classification matrix and combine them as the initial position of the first moth in the first moth population; randomly select a piece of data from each category of the second initial classification matrix and combine them as the initial position of the first moth in the second moth population; repeat the above operation to obtain the first initial position set and the second initial position set Represents the j-th dimension coordinate data of the initial position of the i-th moth in the first moth population; Represents the j-th dimension coordinate data of the initial position of the i-th moth in the second moth population;

[0130] Set the fitness functions of the first moth population and the second moth population as follows:

[0131]

[0132] In the formula, and Respectively represent the fitness functions of the first moth population and the second moth population; l1 represents the sum of the Euclidean distances between each piece of data in the first initial classification matrix and the moth positions in the first moth population, and l2 represents the sum of the Euclidean distances between each piece of data in the second initial classification matrix and the moth positions in the second moth population;

[0133] S2233, start the iteration operation, set the current iteration number of the first moth population to l1′ and the current iteration number of the second moth population to l2′; calculate the global optimal fitness value of the first moth population and the second moth population and the corresponding global optimal moth individual according to the fitness function of the first moth population and the second moth population in each round of iteration; and update the position of each moth in the first moth population and the second moth population according to the global optimal fitness value and repeat S222;

[0134] When l1′≥h1′, the first moth population stops the iteration operation and outputs the first final cluster center set and the first final classification matrix;

[0135] When l2′≥h2′, the second moth population stops the iteration operation and outputs the second final cluster center set and the second final classification matrix;

[0136] S23: The thermal comfort influencing factor classification model sets the first comprehensive comfort level data set With the first initial classification matrix The correspondence between the first initial classification and the comfort level is obtained. and Respectively represent the i-th category in the first initial classification set and the corresponding comprehensive comfort level data;

[0137] Then set the second comprehensive comfort level data set The corresponding relationship between the second initial classification matrix f and the comfort level is obtained. and Respectively represent the i-th category in the second initial classification set and the corresponding comprehensive comfort level data;

[0138] S3, reconstructing a data prediction model for thermal comfort influencing factors, wherein the data prediction model for thermal comfort influencing factors predicts the first natural influencing factor data matrix, the first human influencing factor data matrix, the second natural influencing factor data matrix, and the second human influencing factor data matrix to obtain prediction results, and calculates the minimum Euclidean distance and the maximum Euclidean distance from each data in the first final classification matrix to the first final cluster center set and from each data in the second final classification matrix to the second final cluster center set, to obtain a first maximum and minimum Euclidean distance interval set and a second maximum and minimum Euclidean distance interval set;

[0139] The S3 comprises the following steps:

[0140] S31, reconstructing a thermal comfort influencing factor data prediction model, wherein the thermal comfort influencing factor data prediction model uses a BP neural network model to respectively calculate the first natural influencing factor data matrix The first artificial influencing factor data matrix d′ and the second natural influencing factor data matrix d, the second artificial influencing factor data matrix Make predictions and obtain the first prediction result natural data matrix The first prediction result is an artificial data matrix and the second predicted natural data matrix m and the second predicted artificial data matrix m′; respectively,

[0141]

[0142] in, represents the predicted data of the jth natural influencing factor of thermal comfort in the indoor space b1 at the i-th time point in the future; represents the predicted data of the jth thermal comfort human factors influencing the indoor space b1 at the i-th time point in the future; m ij represents the predicted data of the jth natural influencing factor of thermal comfort in the outdoor space b2 at the i-th time point in the future; m i ' jrepresents the predicted data of the jth thermal comfort human factors influencing the outdoor space b2 at the i-th time point in the future; Indicates the total number of predicted future time points;

[0143] S32: The thermal comfort influencing factor data prediction model calculates each data in the first final classification matrix to the first final cluster center set The minimum Euclidean distance and the maximum Euclidean distance of the first maximum and minimum Euclidean distance interval set are obtained. and Respectively represent the minimum Euclidean distance and the maximum Euclidean distance from the classified data in the ith classification in the first final classification matrix to the ith cluster center in the first final cluster center set;

[0144] Calculate each data in the second final classification matrix to the second final cluster center set The maximum Euclidean distance and the minimum Euclidean distance of the second maximum and minimum Euclidean distance interval set are obtained. n i min and n i max Respectively represent the minimum Euclidean distance and the maximum Euclidean distance from the classified data in the ith category in the second final classification matrix to the ith cluster center in the second final cluster center set;

[0145] S4. Finally, a thermal comfort level determination model is constructed. The thermal comfort level determination model combines the prediction results of the thermal comfort influencing factor data prediction model to obtain a first prediction result data matrix and a second prediction result data matrix; the first prediction result data matrix and the second prediction result data matrix are classified to obtain a first prediction data classification result set and a second prediction data classification result set; and then a first prediction comfort level data set and a second prediction comfort level data set are obtained according to the first prediction data classification result set and the second prediction data classification result set;

[0146] The S4 comprises the following steps:

[0147] S41, finally constructing a thermal comfort level determination model, wherein the thermal comfort level determination model is used to determine the first prediction result natural data matrix The first prediction result is an artificial data matrix Merge to obtain a first prediction result data matrix; then merge the second prediction natural data matrix m and the second prediction artificial data matrix m′ to obtain a second prediction result data matrix;

[0148] S42, the thermal comfort level determination model calculates each row of data in the first prediction result data matrix to the first final cluster center set The Euclidean distances of each cluster center in the second prediction result data matrix are calculated to obtain the first Euclidean distance data matrix n′; and then the distances of each row of data in the second prediction result data matrix to the second final cluster center set are calculated. The Euclidean distance of each cluster center in the second Euclidean distance data matrix is ​​obtained. The n′ and They are as follows:

[0149]

[0150] Among them, n i ' j represents the Euclidean distance from the i-th row of data in the first prediction result data matrix to the j-th cluster center in the first final cluster center set; represents the Euclidean distance from the i-th row of data in the second prediction result data matrix to the j-th cluster center in the second final cluster center set;

[0151] S43: The thermal comfort level determination model selects the first Euclidean distance data matrix n′ and the second Euclidean distance data matrix respectively. Each of the rows in the set lies in the first maximum and minimum Euclidean distance interval And the second largest and smallest Euclidean distance interval set The data in the interval at the corresponding position in is obtained to obtain the first predicted initial Euclidean distance data matrix and the second predicted initial Euclidean distance data matrix;

[0152] S44, the thermal comfort level determination model selects the minimum value of each row of data in the first initial Euclidean distance data matrix and the second initial Euclidean distance data matrix to obtain the first final Euclidean distance data set and the second final Euclidean distance data set; according to the influencing factor data corresponding to each Euclidean distance in the first final Euclidean distance data set and the second final Euclidean distance data set and the classification to which it belongs, the first prediction data classification result set is obtained and the second prediction data classification result set

[0153] and i Respectively represent the categories to which the influencing factor data at the i-th time point in the first prediction result data matrix and the second prediction result data matrix belong;

[0154] S45: the thermal comfort level determination model classifies the result set according to the first prediction data and the second prediction data classification result set And the corresponding relationship set between the first initial classification and the comfort level and the corresponding relationship set between the second initial classification and the comfort level A first predicted comfort level data set and a second predicted comfort level data set are obtained.

[0155] A system for constructing a thermal comfort prediction model, comprising a thermal comfort natural influencing factor data collection module, a thermal comfort human influencing factor data collection module, a comprehensive comfort level data collection module, a first classification module, a cluster center calculation module, an influencing factor data prediction module, a second classification module, and a predicted comfort level data determination module;

[0156] The thermal comfort natural influencing factor data collection module is used to collect thermal comfort natural influencing factor data;

[0157] The thermal comfort human influencing factor data collection module is used to collect thermal comfort human influencing factor data;

[0158] The first classification module is used to classify the first natural influencing factor data matrix, the second natural influencing factor data matrix, the first human influencing factor data matrix and the second human influencing factor data matrix using the KMC algorithm;

[0159] The cluster center calculation module is used to calculate the cluster center of each classification data set;

[0160] The influencing factor data prediction module is used to predict the first natural influencing factor data matrix, the first artificial influencing factor data matrix, the second natural influencing factor data matrix, and the second artificial influencing factor data matrix;

[0161] The second classification module is used to classify the first prediction result data matrix and the second prediction result data matrix;

[0162] The predicted comfort level data determination module is used to obtain a first predicted comfort level data set and a second predicted comfort level data set according to the first predicted data classification result set and the second predicted data classification result set.

[0163] In the description of this specification, the description with reference to the terms "one embodiment", "example", "specific example", etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0164] The preferred embodiments of the invention disclosed above are only used to help explain the invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific implementation methods described. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the invention, so that those skilled in the art can understand and use the invention well.

Claims

1. A method for constructing a thermal comfort prediction model, characterized in that: The following steps are involved: S1. Collect data on natural influencing factors of thermal comfort and data on human influencing factors of thermal comfort to obtain a first natural influencing factor data matrix, a second natural influencing factor data matrix, a first human influencing factor data matrix and a second human influencing factor data matrix; the data on natural influencing factors of thermal comfort include indoor air temperature, relative humidity, air velocity and ambient radiation; the data on human influencing factors of thermal comfort include human activity level, clothing thermal resistance and number of people; then set a comprehensive comfort level set and collect comprehensive comfort level data; The S1 comprises the following steps: S11. Set the natural influencing factors of thermal comfort a = {a1, a2, ..., a i ,...,a a′ } and the set of human factors affecting thermal comfort a i represents the i-th natural influencing factor of thermal comfort, and a′ represents the total number of natural influencing factors of thermal comfort; represents the i-th artificial influencing factor of thermal comfort, and a~ represents the total number of artificial influencing factors of thermal comfort; Then set the indoor space b1 and the outdoor space b2; S12: Set a historical data collection cycle, divide the historical data collection cycle into time points, and obtain a historical data collection time point set. c i ′ represents the i-th time point divided from the historical data collection period, Indicates the total number of time points divided into the historical data collection period; According to the historical data collection time point set and the set of natural factors affecting thermal comfort a={a1,a2,...,a i ,...,a a′ } Collect the data of natural influencing factors of thermal comfort in indoor space b1 and outdoor space b2 respectively, and obtain the first natural influencing factor data matrix and the second natural influencing factor data matrix d; and then according to the historical data collection time point set and thermal comfort human influencing factors The thermal comfort human influence factor data are collected in the indoor space b1 and the outdoor space b2 respectively to obtain the first human influence factor data matrix d′ and the second human influence factor data matrix The first natural influencing factor data matrix The second natural influencing factor data matrix d, the first artificial influencing factor data matrix d′ and the second artificial influencing factor data matrix They are as follows: in, represents the data collected from the indoor space b1 at the i-th time point on the j-th natural influencing factor of thermal comfort; d ij represents the data collected from the outdoor space b2 at the i-th time point on the j-th natural influencing factor of thermal comfort; d i ' j represents the data collected from the indoor space b1 at the i-th time point on the j-th thermal comfort artificial influencing factor; represents the data collected from the outdoor space b2 at the i-th time point on the j-th thermal comfort human influencing factor; S13. Set comprehensive comfort level set b i ′ represents the set i-th comprehensive comfort level, Indicates the total number of set comprehensive comfort levels; According to the comprehensive comfort level set Historical data collection time point set And the corresponding comprehensive comfort level data of indoor space b1 is collected to obtain the first comprehensive comfort level data set represents the comprehensive comfort level data collected from the indoor space b1 at the i-th time point; According to the comprehensive comfort level set Historical data collection time point set And the corresponding comprehensive comfort level data of outdoor space b2 is collected to obtain the second comprehensive comfort level data set c i represents the comprehensive comfort level data collected from the outdoor space b2 at the i-th time point; S2. Construct a classification model for thermal comfort influencing factors, wherein the classification model for thermal comfort influencing factors classifies the first natural influencing factor data matrix and the first artificial influencing factor data matrix and calculates the cluster center of each classified data set to obtain a first final classification matrix and a first final cluster center set; the classification model for thermal comfort influencing factors further classifies the second natural influencing factor data matrix and the second artificial influencing factor data matrix and calculates the cluster center of each classified data set to obtain a second final classification matrix and a second final cluster center set; and at the same time, set a corresponding relationship between the comprehensive comfort level data and the initial classification to obtain a first initial classification and comfort level corresponding relationship set and a second initial classification and comfort level corresponding relationship set; S3. Reconstruct a data prediction model for thermal comfort influencing factors, wherein the data prediction model predicts the first natural influencing factor data matrix, the first human influencing factor data matrix, the second natural influencing factor data matrix, and the second human influencing factor data matrix to obtain prediction results; calculate the minimum Euclidean distance and the maximum Euclidean distance from each data in the first final classification matrix to the first final cluster center set according to the prediction results, and obtain a first maximum and minimum Euclidean distance interval set; then calculate the minimum Euclidean distance and the maximum Euclidean distance from each data in the second final classification matrix to the second final cluster center set according to the prediction results, and obtain a second maximum and minimum Euclidean distance interval set; S4. Finally, a thermal comfort level determination model is constructed. The thermal comfort level determination model combines the prediction results of the thermal comfort influencing factor data prediction model to obtain a first prediction result data matrix and a second prediction result data matrix; the first prediction result data matrix and the second prediction result data matrix are classified to obtain a first prediction data classification result set and a second prediction data classification result set; and then a first prediction comfort level data set and a second prediction comfort level data set are obtained according to the first prediction data classification result set and the second prediction data classification result set; The S4 comprises the following steps: S41, finally constructing a thermal comfort level determination model, wherein the thermal comfort level determination model is used to determine the first prediction result natural data matrix The first prediction result is an artificial data matrix Merge to obtain a first prediction result data matrix; then merge the second prediction natural data matrix m and the second prediction artificial data matrix m′ to obtain a second prediction result data matrix; S42, the thermal comfort level determination model calculates each row of data in the first prediction result data matrix to the first final cluster center set The Euclidean distances of each cluster center in the second prediction result data matrix are calculated to obtain the first Euclidean distance data matrix n′; and then the distances of each row of data in the second prediction result data matrix to the second final cluster center set are calculated. The Euclidean distance of each cluster center in the second Euclidean distance data matrix is ​​obtained. The n′ and They are as follows: Among them, n i ' j represents the Euclidean distance from the i-th row of data in the first prediction result data matrix to the j-th cluster center in the first final cluster center set; represents the Euclidean distance from the i-th row of data in the second prediction result data matrix to the j-th cluster center in the second final cluster center set; S43: The thermal comfort level determination model selects the first Euclidean distance data matrix n′ and the second Euclidean distance data matrix respectively. Each of the rows in the set lies in the first maximum and minimum Euclidean distance interval And the second largest and smallest Euclidean distance interval set The data in the interval at the corresponding position in is obtained to obtain the first predicted initial Euclidean distance data matrix and the second predicted initial Euclidean distance data matrix; S44, the thermal comfort level determination model selects the minimum value of each row of data in the first initial Euclidean distance data matrix and the second initial Euclidean distance data matrix to obtain the first final Euclidean distance data set and the second final Euclidean distance data set; according to the influencing factor data corresponding to each Euclidean distance in the first final Euclidean distance data set and the second final Euclidean distance data set and the classification to which it belongs, the first prediction data classification result set is obtained and the second prediction data classification result set and i Respectively represent the categories to which the influencing factor data at the i-th time point in the first prediction result data matrix and the second prediction result data matrix belong; S45: the thermal comfort level determination model classifies the result set according to the first prediction data and the second prediction data classification result set And the corresponding relationship set between the first initial classification and the comfort level and the corresponding relationship set between the second initial classification and the comfort level A first predicted comfort level data set and a second predicted comfort level data set are obtained.

2. The method for constructing a thermal comfort prediction model according to claim 1, characterized in that: The S2 comprises the following steps: S21, constructing a classification model of thermal comfort influencing factors, wherein the classification model of thermal comfort influencing factors firstly classifies the first natural influencing factor data matrix and the first human influence factor data matrix d′ to obtain a first classification training data matrix; then the second natural influence factor data matrix d and the second human influence factor data matrix Merge to obtain the second classification training data matrix; S22: The thermal comfort influencing factor classification model sets a first KMC algorithm model and a second KMC algorithm model, and uses the first KMC algorithm model to classify the first classification training data matrix to obtain a first initial classification matrix as follows, in, represents the jth influencing factor data set in the i-th category of the first initial classification set, Represents the total number of influencing factor data sets in the i-th category of the first initial classification set; Then the second KMC algorithm model is used to classify the second classification training data matrix to obtain the second initial classification matrix f; as follows, Among them, f ij represents the jth influencing factor dataset in the i-th category of the second initial classification set, f i ′ represents the total number of influencing factor data sets in the i-th category of the second initial classification set; Calculate the first initial classification matrix respectively The cluster centers of each classification data set in the second initial classification matrix f are optimized to obtain the first final cluster center set and the second final cluster center set S23: The thermal comfort influencing factor classification model sets the first comprehensive comfort level data set With the first initial classification matrix The correspondence between the first initial classification and the comfort level is obtained. and Respectively represent the i-th category in the first initial classification set and the corresponding comprehensive comfort level data; Then set the second comprehensive comfort level data set The corresponding relationship between the second initial classification matrix f and the comfort level is obtained. and They respectively represent the i-th category in the second initial classification set and the corresponding comprehensive comfort level data.

3. The method for constructing a thermal comfort prediction model according to claim 2, characterized in that: In S22, using the first KMC algorithm model to classify the first classification training data matrix and using the second KMC algorithm model to classify the second classification training data matrix and calculate and optimize the cluster center of each classification specifically includes the following steps: S221: construct a first initial cluster center set according to the first classification training data matrix and the second classification training data matrix And the second initial cluster center set g 1i represents the cluster center corresponding to the i-th comprehensive comfort level in the first initial cluster center set, g 2i represents the cluster center corresponding to the i-th comprehensive comfort level in the second initial cluster center set; S222, respectively calculate the first initial cluster center set for each row of data in the first classification training data matrix The Euclidean distance of each cluster center in the second classification training data matrix and the distance from each row of data to the second initial cluster center set The Euclidean distance of each cluster center in is obtained to obtain the first Euclidean distance data matrix g1′ and the second Euclidean distance data matrix g′2; The influencing factor classification training data corresponding to the minimum data in each row of the first Euclidean distance data matrix g1′ and the second Euclidean distance data matrix g′2 and the corresponding cluster center are taken as the same category to obtain a first initial classification matrix and a second initial classification matrix; S223, using the moth-to-flame algorithm to and the second initial cluster center set Optimize and get the first final cluster center set and the second final cluster center set represents the cluster center corresponding to the i-th comprehensive comfort level in the first final cluster center set, Represents the cluster center corresponding to the i-th comprehensive comfort level in the second initial cluster center set.

4. The method for constructing a thermal comfort prediction model according to claim 3, characterized in that: The S221 includes the following steps: S2211, setting a first initial cluster center set g1 and a second initial cluster center set g2; S2212, randomly selecting a row of data from the first classification training data matrix and the second classification training data matrix and putting them into the first initial cluster center set g1 and the second initial cluster center set g2; at the same time, deleting the selected data from the first classification training data matrix and the second classification training data matrix to obtain a first updated first classification training data matrix and a first updated second classification training data matrix; S2213, respectively calculating the Euclidean distances of each data in the first updated first classification training data matrix and the first updated second classification training data matrix to the data in the first initial cluster center set g1 and the second initial cluster center set g2, selecting the classification training data corresponding to the largest Euclidean distance and putting them into the first initial cluster center set g1 and the second initial cluster center set g2, and deleting them from the first updated first classification training data matrix and the first updated second classification training data matrix, to obtain the second updated first classification training data matrix and the second updated second classification training data matrix; S2214, respectively calculating the Euclidean distances of each data in the second updated first classification training data matrix and the second updated second classification training data matrix to each data in the first initial cluster center set g1 and the second initial cluster center set g2, selecting the largest Euclidean distance and the smallest Euclidean distance and multiplying them, putting the classification training data matrix corresponding to the data with the largest multiplication result into the first initial cluster center set g1 and the second initial cluster center set g2, and deleting it from the second updated first classification training data matrix and the second updated second classification training data matrix, to obtain the third updated first classification training data matrix and the third updated second classification training data matrix; S2215: When the number of data in the first initial cluster center set g1 and the second initial cluster center set g2 is greater than or equal to When , stop putting in; otherwise, repeat S2214.

5. The method for constructing a thermal comfort prediction model according to claim 3, characterized in that: The S223 comprises the following steps: S2231, constructing a first moth population and a second moth population; setting the sizes of the first moth population and the second moth population to be and Then the first moth population and the second moth population are represented as and h 1i and h 2i denote the i-th moth in the first moth population and the i-th moth in the second moth population respectively; set the first maximum number of iterations h1′ and the second maximum number of iterations h2′; set the search space dimensions of the first moth population and the second moth population to be dimension; S2232, randomly select a piece of data from each category of the first initial classification matrix and combine them as the initial position of the first moth in the first moth population; randomly select a piece of data from each category of the second initial classification matrix and combine them as the initial position of the first moth in the second moth population; repeat the above operation to obtain the first initial position set and the second initial position set Represents the j-th dimension coordinate data of the initial position of the i-th moth in the first moth population; represents the j-th dimension coordinate data of the initial position of the i-th moth in the second moth population; Set the fitness functions of the first moth population and the second moth population as follows: In the formula, and Respectively represent the fitness functions of the first moth population and the second moth population; l1 represents the sum of the Euclidean distances between each piece of data in the first initial classification matrix and the moth positions in the first moth population, and l2 represents the sum of the Euclidean distances between each piece of data in the second initial classification matrix and the moth positions in the second moth population; S2233, start the iteration operation, set the current iteration number of the first moth population to l1′ and the current iteration number of the second moth population to l2′; calculate the global optimal fitness value of the first moth population and the second moth population and the corresponding global optimal moth individual according to the fitness function of the first moth population and the second moth population in each round of iteration; and update the position of each moth in the first moth population and the second moth population according to the global optimal fitness value and repeat S222; When l1′≥h1′, the first moth population stops the iteration operation and outputs the first final cluster center set and the first final classification matrix; When l2′≥h2′, the second moth population stops the iteration operation and outputs the second final cluster center set and the second final classification matrix.

6. The method for constructing a thermal comfort prediction model according to claim 1, characterized in that: The S3 comprises the following steps: S31, reconstructing a thermal comfort influencing factor data prediction model, wherein the thermal comfort influencing factor data prediction model uses a BP neural network model to respectively calculate the first natural influencing factor data matrix The first artificial influencing factor data matrix d′ and the second natural influencing factor data matrix d, the second artificial influencing factor data matrix Make predictions and obtain the first prediction result natural data matrix The first prediction result is an artificial data matrix and the second predicted natural data matrix m and the second predicted artificial data matrix m′; respectively, in, represents the predicted data of the jth natural influencing factor of thermal comfort in the indoor space b1 at the i-th time point in the future; represents the predicted data of the jth thermal comfort human factors influencing the indoor space b1 at the i-th time point in the future; m ij represents the predicted data of the jth natural influencing factor of thermal comfort in the outdoor space b2 at the i-th time point in the future; m i ' j represents the predicted data of the jth thermal comfort human factors influencing the outdoor space b2 at the i-th time point in the future; Indicates the total number of predicted future time points; S32: The thermal comfort influencing factor data prediction model calculates each data in the first final classification matrix to the first final cluster center set The minimum Euclidean distance and the maximum Euclidean distance of the first maximum and minimum Euclidean distance interval set are obtained. and Respectively represent the minimum Euclidean distance and the maximum Euclidean distance from the classified data in the ith classification in the first final classification matrix to the ith cluster center in the first final cluster center set; Calculate each data in the second final classification matrix to the second final cluster center set The maximum Euclidean distance and the minimum Euclidean distance of the second maximum and minimum Euclidean distance interval set are obtained. n i min and n i max They respectively represent the minimum Euclidean distance and the maximum Euclidean distance from the classified data in the i-th category in the second final classification matrix to the i-th cluster center in the second final cluster center set.

7. A system for implementing the method for constructing a thermal comfort prediction model according to any one of claims 1 to 6, characterized in that: It includes a data collection module for natural influencing factors of thermal comfort, a data collection module for human influencing factors of thermal comfort, a data collection module for comprehensive comfort level data, a first classification module, a cluster center calculation module, an influencing factor data prediction module, a second classification module and a predicted comfort level data determination module; The thermal comfort natural influencing factor data collection module is used to collect thermal comfort natural influencing factor data; The thermal comfort human influencing factor data collection module is used to collect thermal comfort human influencing factor data; The first classification module is used to classify the first natural influencing factor data matrix, the second natural influencing factor data matrix, the first human influencing factor data matrix and the second human influencing factor data matrix using the KMC algorithm; The cluster center calculation module is used to calculate the cluster center of each classification data set; The influencing factor data prediction module is used to predict the first natural influencing factor data matrix, the first artificial influencing factor data matrix, the second natural influencing factor data matrix, and the second artificial influencing factor data matrix; The second classification module is used to classify the first prediction result data matrix and the second prediction result data matrix; The predicted comfort level data determination module is used to obtain a first predicted comfort level data set and a second predicted comfort level data set according to the first predicted data classification result set and the second predicted data classification result set.