Thermal comfort prediction method and system based on body movement thermal discomfort behavior index

Abstract

The application provides a thermal comfort prediction method and system based on a thermal discomfort behavior index of limb action, acquires current environmental parameters, limb action voting data sets corresponding to thermal sensation descriptions of personnel thermal discomfort limb action; obtains the probability that the thermal sensation of personnel is j under the premise that the i-th posture appears in the limb action voting data set, assigns a weight to the probability, sums, and obtains the thermal sensation degree represented by the limb action i, that is, the thermal discomfort behavior index TDI; taking the current environmental parameters and the thermal discomfort behavior index TDI as inputs, adopting a support vector machine algorithm to perform thermal comfort prediction, and obtaining a thermal comfort quantitative evaluation result. The application quantifies the representation degree of thermal discomfort limb action on thermal sensation by using the thermal discomfort behavior index, and predicts thermal comfort data by using environmental parameters and the thermal discomfort behavior index. The effectiveness of the representation of thermal comfort by the limb action is verified through experiments, and a new method is provided for contactless thermal comfort prediction.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of thermal environment regulation, and particularly relates to a thermal comfort prediction method and system based on a thermal discomfort behavior index of limb action. BACKGROUND

[0002] A room temperature control scheme oriented to thermal comfort can realize energy saving and consumption reduction under the premise of meeting the thermal comfort of human bodies, and is a mainstream trend of thermal environment regulation at present. Thermal comfort is a subjective state of consciousness under the joint action of physiological and psychological reactions, and exists in individual differences, so real-time and accurate measurement of individual thermal comfort states is a prerequisite for realizing energy saving control.

[0003] Real-time voting and thermal comfort prediction based on physiological parameters are common means for obtaining the thermal comfort state of personnel, but real-time voting will interfere with the normal life of users, and the contact / semi-contact collection mode of physiological parameters will also bring inconvenience to users and increase cost investment; it is more feasible and accurate to evaluate the subjective comfort degree of individuals to a thermal environment by using thermal discomfort limb action, but the limb action is greatly subjective due to the influence of physiological, psychological and various other factors, and the relationship between human behavior and thermal comfort and how to realize quantitative prediction of thermal comfort through limb action have not been fully discussed in the current research. SUMMARY

[0004] In order to solve the problems in the prior art, the application provides a thermal comfort prediction method and system based on a thermal discomfort behavior index of limb action, which quantifies the representation degree of thermal discomfort limb action to thermal sensation by using the thermal discomfort behavior index, and performs thermal comfort evaluation and prediction by using environmental parameters and the thermal discomfort behavior index, so that the thermal comfort state of users can be more intuitively and accurately reflected, and a new method for contactless thermal comfort prediction is provided.

[0005] To achieve the above object, the application provides the following technical scheme: a thermal comfort prediction method based on a thermal discomfort behavior index of limb action, and the specific steps are as follows:

[0006] S1 obtains current environmental parameters and a limb action voting data set corresponding to the description of thermal sensation of personnel thermal discomfort limb action;

[0007] S2 obtains the probability that the personnel represent thermal sensation j under the premise of the appearance of the ith posture in the limb action voting data set, and assigns a weight to the probability, sums up, and obtains the thermal sensation degree represented by the limb action i, that is, the thermal discomfort behavior index TDI;

[0008] S3 takes the current environmental parameters and the thermal discomfort behavior index TDI as inputs, adopts a support vector machine algorithm to perform thermal comfort prediction, and obtains a thermal comfort quantitative evaluation result.

[0009] Further, in S1, the limb action voting data set of the personnel thermal discomfort limb action corresponding to the thermal sensation description is obtained by means of a questionnaire.

[0010] Further, in S1, the question of the questionnaire is: "When you feel the thermal sensation description as X, what will you do in the indoor space?", wherein the ith posture in each question is the determined personnel thermal discomfort limb action; the thermal sensation description X is an option, and the option includes the ASHRAE 7-point thermal sensation scale and the options of "never do" and "the action is meaningless and cannot reflect your cold and heat".

[0011] Further, in S2, the formula of the thermal discomfort behavior index TDI is as follows:

[0012]

[0013] wherein j represents the ASHRAE 7-point thermal sensation scale, j ∈ [-3, -2, -1, 0, 1, 2, 3], i represents the limb action type, P(A i ) represents the probability of the ith limb action, P(A i C j ) represents the probability of the ith limb action and the thermal sensation being j; P(C j |A i ) represents the probability of the thermal sensation of the personnel being j under the premise of the ith posture, wherein P(A i ) = 1-P(n i ), and P(n i ) is never doing the ith action.

[0014] Further, the thermal discomfort limb action is 20, i takes values [1, 2,..., 20], and the thermal discomfort limb action includes crossing arms, rubbing hands, wearing a hat, stamping feet, blowing hands, hugging shoulders, sneezing, putting on clothes, taking off a hat, putting hands in a pocket, wiping sweat, fanning, rolling up sleeves, brushing hair, taking off clothes, pulling a collar, shaking clothes in front of the chest, and putting hands on the neck.

[0015] Further, in S2, the value range of TDI is selected to be consistent with the ASHRAE 7-point thermal sensation scale, that is, [-3, 3], wherein -3 represents cold, 3 represents heat, and 0 represents neither cold nor hot.

[0016] Further, in S3, the environmental variables include the air temperature, the relative humidity, the radiant temperature, and the air flow rate of the space where the personnel is located.

[0017] Further, in S3, the thermal comfort quantitative evaluation result is represented by the ASHRAE 3-point thermal comfort scale.

[0018] The application further provides a thermal comfort prediction system based on a thermal discomfort behavior index of limb action, comprising:

[0019] A data acquisition module is configured to acquire a current environmental parameter and a limb action voting data set of a thermal sensation description corresponding to a thermal discomfort limb action of a person;

[0020] A thermal discomfort behavior index calculation module is configured to obtain a probability that a person's thermal sensation is j before the appearance of the ith posture, and assign a weight to the probability, sum up, and obtain a thermal sensation degree represented by the limb action i, i.e., a thermal discomfort behavior index TDI.

[0021] A thermal comfort prediction module is configured to use a support vector machine algorithm to perform thermal comfort prediction with the current environmental parameter and the thermal discomfort behavior index TDI as inputs, and obtain a thermal comfort quantitative evaluation result.

[0022] The application further provides a terminal device, which comprises a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the thermal comfort prediction method according to any one of claims 1-8 when executing the computer program, or the processor implements the functions of the modules in the system according to claim 9 when executing the computer program.

[0023] Compared with the prior art, the application has at least the following beneficial effects:

[0024] The application provides a thermal comfort prediction method based on a thermal discomfort behavior index of limb action, which quantifies the representation degree of thermal discomfort limb action on thermal sensation through the thermal discomfort behavior index TDI, and uses the thermal discomfort behavior index TDI and environmental parameters as inputs to select a support vector machine algorithm for thermal comfort evaluation and quantitative prediction.

[0025] The thermal comfort quantitative evaluation result predicted by the application can be used as an input for air conditioning regulation and control to adjust the room temperature, thereby meeting the thermal comfort needs of indoor personnel. BRIEF DESCRIPTION OF DRAWINGS

[0026] Figure 1 is a whole structure diagram;

[0027] Figure 2 is a subject basic information distribution (height, weight, age, gender) diagram;

[0028] Figure 3 is a laboratory plan;

[0029] Figure 4 is a room temperature and humidity change graph;

[0030] Figure 5 is an experimental time arrangement graph;

[0031] Figure 6 is a subjective questionnaire voting measure graph;

[0032] Figure 7 is a TCV and TDI distribution scatter plot; DETAILED DESCRIPTION

[0033] The present application will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0034] As shown in the drawings, the present application provides a thermal comfort prediction method based on a body movement thermal discomfort behavior index, and the specific steps are as follows: Figure 1

[0035] S1 quantifies thermal discomfort body movement, and calculates a thermal discomfort behavior index TDI:

[0036] 1) According to the questionnaire, obtain a body movement voting data set of 20 existing thermal discomfort body movements corresponding to 9 thermal sensation descriptions (including 7 thermal sensations, and additionally including “never do” and “the action is meaningless and cannot reflect your cold and heat” two options), and the thermal sensation is expressed by an ASHRAE 7-point thermal sensation scale, which includes cold-3, cool-2, slightly cool-1, neutral 0, slightly warm 1, warm 2, and hot 3.

[0037] The 20 thermal discomfort body movements include: crossing arms; rubbing hands; wearing a hat; stamping feet; blowing hands; hunching shoulders; sneezing; putting on clothes; taking off a hat; putting hands in pockets; wiping sweat; fanning; rolling up sleeves; pulling hair; taking off clothes; pulling a collar; shaking clothes in front of the chest; and putting hands on the neck.

[0038] 2) Let j represent the thermal sensation, j∈[-3,-2,-1,0,1,2,3], i represent the body movement type, i∈[1,2,…,20], obtain the probability of the occurrence of the ith body movement and the probability of the occurrence of the ith body movement and the thermal sensation of j in the body movement voting data set, obtain the probability of the thermal sensation of j represented by the personnel under the premise of the occurrence of the ith posture, and obtain the probability of the occurrence of the thermal sensation of j corresponding to each thermal sensation, and then sum up to represent the probability of the occurrence of the thermal sensation of j corresponding to the body movement i in the body movement voting data set, and obtain the thermal sensation degree represented by the body movement i, i.e., obtain the thermal discomfort behavior index TDI, as shown in formula (1):

[0039]

[0040] ​Wherein, P(Ai), i∈[1,2,…,20] represents the probability of the occurrence of the i-th limb action, P(AiC j ), i∈[1,2,…,20], j∈[-3,-2,-1,0,1,2,3] represents the probability of the occurrence of the i-th limb action and the thermal sensation is j; P(C j |A i ) represents the probability of the thermal sensation of the person's representation being j when the i-th posture appears, wherein P(A i )=1-P(n i ), P(n i ) is never the i-th action.

[0041] Preferably, the value range of TDI is selected to be consistent with the ASHRAE 7 thermal sensation scale, that is, [-3, 3], wherein -3 represents cold, 3 represents heat, and 0 represents neither cold nor hot. The more the TDI value approaches -3, the deeper the cold discomfort degree represented by the limb behavior, and vice versa. The more the TDI value approaches 3, the deeper the heat discomfort degree represented by the limb behavior.

[0042] S2 obtains the current environmental parameters, and constructs a thermal comfort prediction network based on a support vector machine (SVM) algorithm in combination with the thermal discomfort behavior index TDI, as shown in formula (2);

[0043] K(x, y)=x·y (2)

[0044] Wherein x and y are vectors of input data.

[0045] S3 finds the best hyperparameters of the support vector machine (SVM) algorithm by using a grid search method, and obtains a thermal comfort prediction model;

[0046] In use, the environmental variables and the behavior variables (thermal discomfort behavior index TDI) are input as inputs into the thermal comfort prediction model to obtain a predicted thermal comfort quantitative evaluation result.

[0047] Preferably, the environmental variables include the air temperature, the relative humidity, the radiant temperature, and the air flow rate of the space in which the person is located.

[0048] Preferably, the room temperature can be controlled by the thermal comfort data predicted by the present application.

[0049] The present application provides a thermal comfort prediction system based on a limb action thermal discomfort behavior index, comprising:

[0050] A data acquisition module is configured to acquire current environmental parameters and a limb action voting data set corresponding to a thermal sensation description of a person's thermal discomfort limb action.

[0051] a thermal discomfort behavior index calculation module, configured to obtain a probability of a thermal sensation of a person being j before the i-th posture appears in the limb motion voting data set, assign a weight to the probability, sum the weights, and obtain a thermal sensation degree represented by the i-th limb motion, i.e., a thermal discomfort behavior index TDI;

[0052] a thermal comfort prediction module, configured to take the current environmental parameters and the thermal discomfort behavior index TDI as inputs, perform thermal comfort prediction by using a support vector machine algorithm, and obtain a thermal comfort quantitative evaluation result.

[0053] The present application provides a terminal device, comprising a processor, a memory, and a computer program stored in the memory and executable on the processor. The processor implements the steps in the above method when executing the computer program. Alternatively, the processor implements the functions of the modules / units in the above system when executing the computer program.

[0054] The above computer program can be divided into one or more modules / units, which are stored in the above memory and executed by the processor to complete the present application.

[0055] The terminal device can be a desktop computer, a notebook computer, a palm computer, a cloud server, or other computing devices.

[0056] The terminal device can include, but is not limited to, a processor and a memory.

[0057] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0058] The memory can be used to store the above computer program and / or modules. The processor realizes the various functions of the above terminal device by running or executing the computer program and / or modules stored in the memory, and calling the data stored in the memory.

[0059] The modules / units integrated in the terminal device, if realized in the form of software function units and sold or used as independent products, can be stored in a computer readable storage medium.

[0060] Based on such understanding, the present application implements all or part of the above-mentioned method, which can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer readable storage medium, and the computer program can implement the above-mentioned method steps when executed by a processor. The computer program includes computer program codes, which can be in the form of source code, object code, executable files or some intermediate forms, etc.

[0061] The above-mentioned computer readable medium can include any entity or device, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc. that can carry the above-mentioned computer program code. It should be noted that the content contained in the above-mentioned computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electric carrier signals and telecommunication signals.

[0062] Embodiments

[0063] 1. Calculate thermal discomfort behavior index TDI

[0064] 1) Due to the complex influence of multiple factors such as physiology and psychology on limb movement, in order to apply it to the research of thermal comfort prediction, it is necessary to quantify the representation of limb movement to thermal comfort. The existing research does not consider the individual difference of limb movement, which leads to imperfect research results. Therefore, extensive social investigation is carried out, 20 kinds of limb movements are defined based on previous research, and 1200 questionnaires are issued, 972 valid questionnaires are recovered, including 273 males and 699 females. The questionnaire content includes personal information such as gender, age, height, weight and limb movement investigation. For example, "when you feel __, you will cross your arms indoors?", There are 9 options for the answer, which are "cold", "cool", "slightly cool", "neutral", "slightly warm", "warm", "hot", "never do", "the action has no meaning and cannot reflect your cold and heat", as shown in Table 2. In order to ensure the effectiveness of the questionnaire, the subjects of the questionnaire involve different genders, ages, weights, heights and regions. The basic information of the subjects is shown in Table 1. Figure 2

[0065] Table 1: Limb movement survey of questionnaire

[0066]

[0067] ​

[0068] The results of the survey obtained the probability of each body movement representing different thermal sensation description. The results are shown in Table 2. It can be observed from the table that the cold and hot sensation of most body movements is quite clear, such as wiping sweat and fanning. However, the cold and hot sensation of a small number of body movements is quite ambiguous, such as placing hands on the neck and crossing legs. In the thermal sensation measurement with a large difference, the distribution is balanced, and it is not possible to directly and clearly determine the cold and hot degree represented by the body movement. Therefore, in the research of thermal comfort prediction using body movements, it is obviously not accurate to divide the thermal sensation degree according to the survey results. It is necessary to further analyze the thermal sensation state represented by the body movement and establish the corresponding quantitative mapping relationship between the body movement and the thermal sensation, which is particularly important for the application of body movement in thermal environment prediction.

[0069] Table 2 Probability distribution of body movement voting

[0070]

[0071] 2) As can be seen from the above, the voting probability of thermal sensation corresponding to each body movement is different, and the cold and hot difference between the seven measurements of thermal sensation is large. It is not reasonable to unify the cold and hot degree of all body movements. Therefore, a thermal discomfort state evaluation method based on body movement, thermal discomfort behavior index TDI, is proposed to represent the thermal sensation degree represented by each thermal discomfort body movement, as shown in formula (1).

[0072] 3) According to Bernoulli's law of large numbers, a large number of repeated experiments of random events will show almost certain rules. When the number of social survey samples is sufficient, the voting probability distribution of thermal sensation corresponding to each body movement will tend to a fixed value. Therefore, the thermal discomfort behavior index TDI can reflect the cold and hot degree represented by different body movements. The TDI values of 20 body movements can be obtained from formula (1), as shown in Table 3. The body behavior represented by the TDI value closer to -3 represents a deeper cold discomfort degree, and vice versa. The body behavior represented by the TDI value closer to 3 represents a deeper thermal discomfort degree.

[0073] Table 3 TDI values of thermal discomfort body movements

[0074]

[0075] 2. Laboratory and experimental equipment

[0076] A closed laboratory in a certain university experimental building was selected for the experiment. The laboratory plan is as follows Figure 3The laboratory is located on the 6th floor, with a space size of 9.0 m*5.6 m*3.0 m (length* width* height). The laboratory is oriented north-south, with a double-leaf opening and closing door on the south side, and two 1.2 m*0.8 m windows on the east and north walls. During the experiment, the doors and windows were closed to maintain a constant laboratory temperature. There are two central air conditioning units in the laboratory to control the indoor temperature and humidity, located on the north and west sides of the room near the wall. There are 35 workstations, with transparent glass partitions between the front-facing workstations. The seats are office chairs with backrests designed to fit the human sitting position.

[0077] To ensure accurate control, the accuracy and test range of the instruments used to monitor the physical environment parameters in the laboratory during the experiment meet the requirements of international standard ISO7726-1998 "Measurement of Physical Quantities of Human Biophysics in Thermal Environment". The clothing thermal resistance is referenced to international standard "Clothing Thermal Resistance Test Method Warm Body Dummy Method GB / T18398-2001", and manually recorded by researchers according to the subject's clothing situation in the camera device. Since the subjects were sitting and working during the experiment, and were at a certain distance from the air conditioning equipment, the metabolic rate was 1.2 Met, and the wind speed was 0.02 m / s.

[0078] Table 4 Measurement equipment and accuracy

[0079]

[0080] 3、Experimenters

[0081] The research object is the indoor thermal comfort of small office, and a total of 30 volunteers were recruited, including 15 males and 15 females. All subjects have lived in the test area for more than a year, which can avoid the influence of environmental adaptation. All subjects are healthy, with stable psychological condition, no strenuous exercise before the experiment, and no discomfort symptoms. Before the experiment, the basic information of the subjects was investigated for subsequent experimental data analysis, as shown in Table 5, including gender, age, weight, height, and body mass index (BMI), where BMI is obtained by dividing weight (kg) by the square of height (m).

[0082] Table 5 Basic information of subjects

[0083]

[0084] 4、Experimental steps

[0085] The experiment was conducted in May 2022 to avoid experimental bias and repeated for ten consecutive days. The daily time was from 13:00 to 15:30, and the average temperature and humidity distribution was as shown in Figure 4 The average outdoor temperature was between 24℃ and 26℃, and the relative average humidity was between 36% and 40%. The experimental steps were as followsFigure 5 The air conditioning set point temperature was gradually increased from 18℃-30℃, increasing by 2℃ every 20 minutes. Environmental and physiological parameters were recorded every 10 minutes, and subjective questionnaires were filled out. Thermal discomfort behaviors were recorded by a camera. Each experimental subject collected 14 sets of data each time, for a total of 4200 sets of effective data.

[0086] ASHRAE Standard 55-2013 defines thermal sensation as the conscious, subjective description of a thermal environment as "cool" or "warm"; thermal comfort is defined as the satisfaction of most people with an objective thermal environment in terms of both physiological and psychological aspects. The subjective questionnaire in this experiment included thermal comfort and thermal sensation. The questionnaire voting scale was as follows: Figure 6 As shown in the table: (1) Thermal sensation, using the ASHRAE 7-point thermal sensation scale, i.e. cool, cool, slightly cool, neutral, slightly warm, warm, and hot; (2) Thermal comfort, using the ASHRAE 3-point thermal comfort scale, i.e. cold discomfort, comfort, and heat discomfort; Table 2 was used to assign values to thermal discomfort behaviors. If no or 20 other body movements were observed, the thermal discomfort index TDI was 0. When multiple body movements occurred within the data collection interval, the thermal discomfort index TDI was linearly superimposed according to the number and type of thermal discomfort behaviors to calculate the actual thermal discomfort index.

[0087] During the 24-hour experiment, subjects were not allowed to stay up late or drink alcohol. During the experiment, subjects sat quietly at their desks and were only allowed to perform light activities. They were not allowed to adjust the air conditioning controller or the state of the doors and windows, and they were not allowed to discuss experiment-related content with each other to avoid interference with the results. During the experiment, subjects could adjust their clothing according to their own needs and fill out the subjective questionnaire with the actual clothing situation.

[0088] 5. Model construction

[0089] 1) Machine learning model selection

[0090] A machine learning algorithm is used to establish a thermal comfort prediction model, which makes up for the lack of self-correction ability of traditional models. The experimental data collected are used for model classification training, which is completed in the python 3.10.4 environment. Through experimental analysis, the classification algorithm support vector machine (SVM) with less requirement for data dimension and easy-to-adjust hyperparameters is selected as the best one. The input and output variable combinations are shown in Table 6. The input parameters are divided into four groups, namely environmental variables, physiological variables, individual variables and behavioral variables. Among them, the environmental variables are air temperature, radiant temperature, relative humidity and air flow rate. The physiological variables are wrist skin temperature, fingertip heart rate, clothing thermal resistance and metabolic rate. The individual variables are gender and BMI value, and the behavioral variables are thermal discomfort behavior index TDI. The four groups of input parameters are randomly combined into 8 groups of input parameter combination modes, and the output parameter is the thermal comfort voting data. The grid search method is used to find the best hyperparameters of the algorithm during the experiment.

[0091] Table 6 Input and output variables of thermal discomfort prediction model

[0092]

[0093] 2) Statistical analysis method

[0094] In this experiment, 80% of the data collected by oneself (including various input variables) is used as the training set, and 20% is used as the test set. The accuracy Accuracy is used as the evaluation index for model performance evaluation. Accuracy is one of the most commonly used performance indicators for thermal comfort prediction, which represents the number of correct predictions divided by the total number of predictions, as shown in equation (3). TP is the number of correctly predicted positive samples, TN is the number of correctly predicted negative samples, P is the number of positive samples, and N is the number of negative samples. To avoid overfitting and better reflect the prediction ability of the model, the k-fold cross-validation method is used. This is a resampling technique used to estimate the performance of the model on unknown data in the final stage of hyperparameter adjustment. In the model training process of this experiment, 5-fold cross-validation is used.

[0095] Accuracy = (TP + TN) / (P + N) (3)

[0096] 6. Experimental result analysis

[0097] The distribution scatter plot of thermal comfort TCV and thermal discomfort behavior index TDI with temperature change is as follows Figure 7(a) shown. With the increase of room temperature, both the thermal comfort and thermal discomfort behavior index distribution showed an increasing trend. When the indoor temperature was between 18-24℃, most of the subjects' subjective thermal comfort vote was cold discomfort, and the thermal discomfort behavior index was concentrated around -2, the probability of cold discomfort behavior was high, and most of the subjects had strong cold discomfort. At this time, the minimum value of thermal discomfort behavior index TDI reached -4.3, indicating that the subjects simultaneously appeared multiple cold discomfort behaviors. When the indoor temperature was between 24-27℃, most of the subjects' subjective thermal comfort vote was neutral, and the thermal discomfort behavior index was concentrated around 0, at this time, there were fewer body movements and higher thermal comfort, and most of the subjects were in a comfortable state. When the indoor temperature was between 27-30℃, most of the subjects' subjective thermal comfort vote was hot discomfort, and the thermal discomfort behavior index was concentrated around 2, at this time, there were more thermal discomfort behaviors and stronger thermal discomfort. Overall, in the low temperature state, it is easier to adjust the discomfort through body movements, and the probability of body movement is higher, and there are multiple actions at the same time. In the high temperature state, the distribution of thermal discomfort behavior index TDI is relatively dispersed, and the probability of body movement is significantly lower than that in the low temperature state.

[0098] For more intuitive analysis, the thermal discomfort behavior index TDI was rounded and normalized, as shown in Figure 7 (b) (c) shown. The distribution of thermal discomfort behavior index TDI and thermal comfort TCV with temperature change showed significant linear correlation, and the distribution trend was almost completely coincided after normalization. When the thermal discomfort behavior index was negative, the thermal comfort TCV vote result was cold discomfort, and when the thermal discomfort behavior index was positive, the thermal comfort TCV vote result was hot discomfort. Combined analysis can know that body movement has a greater correlation with the subjects' subjective thermal comfort vote, and the thermal discomfort behavior index TDI has a greater potential to predict thermal comfort.

[0099] Pearson correlation coefficient r is used to measure the linear relationship between variables, which is an important indicator of correlation analysis. The expression is shown in equation (4). The closer r is to 1, the stronger the correlation, and the closer r is to 0, the weaker the correlation. Pearson coefficient was used to analyze the correlation between thermal discomfort behavior index TDI and thermal comfort TCV by SPSSStatistics26.0, and the data used was the data collected in the experimental steps. The thermal discomfort behavior index TDI, thermal comfort TCV, and correlation coefficient and significance P value are shown in Table 7. When P<0.1, the correlation is significant. The results show that the thermal discomfort behavior index TDI and the thermal comfort TCV are significantly positively correlated, which is consistent with the observation of experimental results, and it can be tried to use the thermal discomfort behavior index TDI to establish the prediction model of thermal comfort TCV.

[0100]

[0101] Pearson correlation analysis of TDI and thermal indices

[0102]

[0103] Thermal comfort prediction

[0104] The thermal comfort prediction results of the machine learning model algorithm under 8 input combinations are shown in Table 8. Compared with input combination 1 and input combination 3, when only the environmental variables are used as input variables, the prediction accuracy is about 68%, and after adding physiological parameters, the prediction accuracy is significantly improved. Compared with input combination 3 and 5, the accuracy does not change much after adding individual factors, which shows that the individual factors have little effect on thermal comfort. When only the thermal discomfort behavior index TDI is used as an input variable, the prediction accuracy reaches 79%, which is higher than using environmental variables and physiological parameters as input variables. Compared with input combination 3 and 4 and input combination 5 and 6, compared with physiological parameters and TDI, it is obvious that the prediction accuracy is significantly higher when TDI is used as an input variable than when physiological parameters are used as an input variable. Compared with input combinations 4, 6, 7, and 8, when environmental parameters and TDI are used as input variables, and then individual variables and physiological variables are added, the accuracy is not greatly improved, and individual variables have little effect on thermal comfort.

[0105] Therefore, we can find that since thermal comfort is a subjective state of consciousness affected by both physiology and psychology, it cannot be well described by environmental parameters and physiological parameters. Using body movements can more intuitively reflect the thermal comfort state of the subjects. Input combination 4 uses only environmental parameters (air temperature, radiant temperature, relative humidity, and air flow rate) and the thermal discomfort behavior index for thermal comfort prediction, which can achieve a high accuracy, and adding input variables will actually cause the accuracy to decrease. The best combination for recognition is to use environmental variables and the thermal discomfort behavior index to establish a thermal comfort prediction model through the SVM algorithm.

[0106] Table 8 Comparison of thermal comfort prediction accuracy of machine learning model

[0107]

[0108] In summary, using environmental parameters and the thermal discomfort behavior index as input variables can achieve good prediction results, and the prediction accuracy is higher than using environmental parameters and physiological parameters as input variables. In the development of actual prediction models, the thermal discomfort behavior index TDI can be used instead of physiological indexes to realize non-contact thermal discomfort state prediction, and the thermal comfort quantitative evaluation result is represented by the ASHRAE 3-point thermal comfort scale.

Claims

1. A thermal comfort prediction method based on a body movement thermal discomfort behavior index, characterized by, The specific steps are as follows: S1, obtaining a current environment parameter and a body action voting data set of a body action corresponding to a thermal sensation description of a thermal discomfort action of a person; S2 obtains the probability of the thermal sensation of the person representation before the appearance of the gesture in the limb action voting data set i j The probability of the thermal sensation of the person representation before the appearance of the gesture in the limb action voting data set i The probability of the thermal sensation of the person representation before the appearance of the gesture in the limb action voting data set​ S3, inputting the current environment parameter and a thermal discomfort behavior index TDI, and performing thermal comfort prediction by using a support vector machine algorithm to obtain a thermal comfort quantitative evaluation result; In S2, the formula of the thermal discomfort behavior index TDI is as follows: in, j This indicates the ASHRAE 7-point thermal sensation. , i Indicates the type of body movement. Indicates the occurrence of the first i The probability of a certain type of physical action. Indicates the occurrence of the first i Such limb movements and heat sensation j The probability of; Indicates when the first occurrence i Under this posture, the thermal sensation represented by the personnel is j The probability of, where , For never doing the first i A type of action.

2. The method of claim 1, wherein the thermal comfort prediction method based on the body movement thermal discomfort behavior index is characterized by, In S1, the body action voting data set of the body action corresponding to the thermal sensation description of the thermal discomfort action of the person is obtained by means of a questionnaire.

3. The method of claim 2, wherein the thermal comfort prediction method is based on a body movement thermal discomfort behavior index. In S1, the questionnaire question is: "Indoors, when you feel heat, describe it as..." X When will do the first i "What kind of posture?", where in each question the first... i This posture is a defined limb movement indicating thermal discomfort in a person; description of thermal sensation. X The options include the ASHRAE7 thermal sensation measurement option and the options "Never do" and "This action is meaningless and does not reflect your temperature".

4. The method of claim 1 or 3, wherein the thermal comfort prediction method is based on a body movement thermal discomfort behavior index. In S1, the thermal discomfort limb movement is 20 kinds, i The value is The thermal discomfort limb movement includes crossing arms; rubbing hands; putting on a hat; stamping feet; blowing hands; shrugging shoulders; sneezing; putting on clothes; taking off a hat; putting hands in pockets; wiping sweat; fanning; rolling up sleeves; brushing hair; taking off clothes; pulling a collar; shaking a chest; putting hands on a neck.

5. The method of claim 1, wherein the thermal comfort prediction method based on the body movement thermal discomfort behavior index is characterized by, In S2, the value range of TDI is selected to be consistent with the ASHRAE 7-point thermal sensation scale, that is, [-3, 3], wherein -3 represents cold, 3 represents heat, and 0 represents neither cold nor hot.

6. The method of predicting thermal comfort based on a body movement thermal discomfort index according to claim 1, wherein, In S3, the environment parameter includes air temperature, relative humidity, radiant temperature and air flow rate of a space where the person is located.

7. The method of claim 1, wherein the thermal comfort prediction method is based on a body movement thermal discomfort behavior index. In S3, the thermal comfort quantitative evaluation result is represented by using an ASHRAE 3-point thermal comfort scale.

8. A thermal comfort prediction system based on a body movement thermal discomfort behavior index, characterized by, It comprises: a data acquisition module, configured to acquire a current environment parameter and a body action voting data set of a body action corresponding to a thermal sensation description of a thermal discomfort action of a person; The thermal discomfort behavior index calculation module is used to obtain the thermal discomfort behavior index in the body movement voting dataset. i Under this posture, the thermal sensation represented by the personnel is j The probability of the movement is calculated, weighted, and summed to obtain the limb movement. i The degree of thermal sensation is represented by the Thermal Discomfort Behavior Index (TDI). a thermal comfort prediction module, configured to input the current environment parameter and a thermal discomfort behavior index TDI, and perform thermal comfort prediction by using a support vector machine algorithm to obtain a thermal comfort quantitative evaluation result; In S2, the formula of the thermal discomfort behavior index TDI is as follows: wherein, j represents the ASHRAE 7 thermal sensation scale, , i represents the type of body movement, represents the probability of the occurrence of the i th type of body movement, represents the probability of the occurrence of the i th type of body movement and the thermal sensation is j ; represents the probability of the thermal sensation of the person representation being i when the j th posture occurs, wherein , is not doing the i th movement.

9. A terminal device, comprising: It comprises: a processor, a memory, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps in the thermal comfort prediction method according to any one of claims 1-7 when executing the computer program, or the processor implements the functions of each module in the system according to claim 8 when executing the computer program.

Images