CN114139937A - Indoor thermal comfort data generation method, system, equipment and medium - Google Patents
Indoor thermal comfort data generation method, system, equipment and medium Download PDFInfo
- Publication number
- CN114139937A CN114139937A CN202111436736.XA CN202111436736A CN114139937A CN 114139937 A CN114139937 A CN 114139937A CN 202111436736 A CN202111436736 A CN 202111436736A CN 114139937 A CN114139937 A CN 114139937A
- Authority
- CN
- China
- Prior art keywords
- indoor
- data
- generator
- temperature
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
Abstract
The invention belongs to the field of building thermal comfort, and particularly relates to a method, a system, equipment and a medium for generating indoor thermal comfort data, wherein the method comprises the following steps: collecting a plurality of groups of real data in a period of time, wherein each group of real data comprises indoor temperature, indoor relative humidity and indoor black ball temperature; inputting the current indoor temperature, the indoor relative humidity and the indoor black ball temperature into the generator after training is finished, and outputting the predicted indoor temperature, the indoor relative humidity and the indoor black ball temperature by the generator; and acquiring the clothing thermal resistance, the human body metabolism rate and the indoor wind speed, and predicting the indoor temperature, the indoor relative humidity and the indoor black ball temperature, and matching corresponding indoor thermal comfort data. The PMV is accurately predicted by measuring three factors of temperature, relative humidity and average radiation temperature, so that the time consumption level of a training stage is reduced, the stability is improved, and the calculation complexity of PMV indexes is reduced.
Description
Technical Field
The invention belongs to the field of building thermal comfort, and particularly relates to an indoor thermal comfort data generation method, system, equipment and medium.
Background
The thermal comfort index PMV is an evaluation index for characterizing the thermal response of a human body, which was taught by Fanger, a scientist in denmark, in the 70's 20 th century. Meanwhile, the PMV index is also a thermal comfort evaluation index with the widest application range and the highest recognition degree in the world at present, has a complex nonlinear relation with various environmental variables, human body parameters and the like, and cannot be directly measured. In the actual working process, under the multiple influences of the complexity of the HVAC system, certain coupling existing among the four variables and the complex nonlinear iterative relationship of the four variables in the PMV index calculation process, it is technically extremely difficult to simultaneously use the four variables as control variables.
The BP (Back propagation) neural network is proposed by a group of scientists including Rumelhart and McCelland in 1986, is a multi-layer feedforward network taking an error inverse propagation algorithm as a training algorithm, is one of the most mature and widely applied neural networks at present, and can learn and represent a large number of input-output mapping relations without the need of defining a mathematical equation of the mapping relation in advance.
The learning process of the BP neural network is a process of repeatedly training the network by using training samples and reducing errors between actual output and expected output by continuously changing the weight and the threshold of the network. When the BP neural network is used for generating indoor thermal comfort data, the method is found to have low convergence speed and be easy to fall into local minimum; meanwhile, the method is sensitive to network initial values, learning efficiency and the like, and the training result is unstable.
Disclosure of Invention
The invention provides a method, a system, equipment and a medium for generating indoor thermal comfort data, aiming at the problem of low convergence speed when a BP neural network is used for generating indoor thermal comfort data in the prior art.
In order to achieve the purpose, the technical scheme of the invention is as follows:
in a first aspect: an indoor thermal comfort data generation method, comprising the steps of:
s1: collecting a plurality of groups of real data in a period of time, wherein each group of real data comprises indoor temperature, indoor relative humidity and indoor black ball temperature;
s2: dividing the plurality of groups of real data into training data and discriminating data according to time, and increasing redundancy to obtain an original random noise point limiting condition;
s3: inputting training data and original random noise point limiting conditions into a generator, and outputting generated data by the generator;
s4: inputting the original random noise point limiting condition, the discrimination data and the generated data into a discriminator, and outputting a discrimination result by the discriminator;
s5: updating parameters of the discriminator according to the discrimination result, reversely propagating errors to the generator by the discriminator, and updating parameters of the generator;
s6: when the judgment result shows that the error reaches the preset precision or the learning frequency is more than the preset training iteration frequency, terminating the algorithm and finishing the generator training;
s7: inputting the current indoor temperature, the indoor relative humidity and the indoor black ball temperature into the generator after training is finished, and outputting the predicted indoor temperature, the indoor relative humidity and the indoor black ball temperature by the generator;
s8: and acquiring the thermal resistance of the clothes, the human body metabolism rate and the indoor wind speed, and the indoor temperature, the indoor relative humidity and the indoor black ball temperature predicted in the S7, and matching corresponding indoor thermal comfort data.
Further, the original random noise point limiting condition is a data interval obtained by adding 20% redundancy to each of the real data.
Further, the number of training iterations is 10000.
Further, the preset precision is error close 10-3。
Further, the hidden layer and output layer activation functions of the generator and the arbiter are f (x) max (0, x), and when x is 1, the obtained result is 1; when x is 0, the result is 0.
Further, in step S5, the corresponding parameters are updated according to the loss functions of the discriminator D and the generator G.
Further, the loss function is:
wherein z is random noise following a gaussian distribution; x is real data; g represents a generator; d represents a discriminator; pdata (x) represents the probability distribution of the real data; pz (x) represents the probability distribution of random noise; x to Pdata represent random extraction of x from the distribution of real data; z to Pz represent extraction of noise z from gaussian-distributed random noise; x | y represents x obtained under the constraint y, and z | y represents z obtained under the constraint y; d (x | y) represents the vector output by the discriminator D after receiving the input in parentheses; g (z | y) represents the vector that the generator G outputs after receiving the input in brackets.
In a second aspect, an indoor thermal comfort data generation system, comprises:
a data acquisition unit: the system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring a plurality of groups of real data in a period of time, and each group of real data comprises indoor temperature, indoor relative humidity and indoor black ball temperature;
a data preprocessing unit: the system is used for dividing the plurality of groups of real data into training data and judging data according to time, and increasing redundancy to obtain an original random noise point limiting condition;
a generation unit: the random noise limiting condition generator is used for inputting training data and an original random noise limiting condition into the generator and outputting generated data by the generator;
a determination unit: the device is used for inputting the original random noise point limiting condition, the discrimination data and the generated data into a discriminator and outputting a discrimination result by the discriminator;
a counter propagation unit: the device is used for updating the parameters of the discriminator according to the discrimination result, and the discriminator reversely transmits errors to the generator and updates the parameters of the generator;
a termination unit: the generator training device is used for terminating the algorithm and finishing generator training when the display error of the judgment result reaches the preset precision or the learning frequency is more than the preset training iteration frequency;
a prediction unit: inputting the current indoor temperature, the indoor relative humidity and the indoor black ball temperature into the generator after training is finished, and outputting the predicted indoor temperature, the indoor relative humidity and the indoor black ball temperature by the generator;
thermal comfort unit: the method is used for obtaining clothing thermal resistance, human body metabolism rate and indoor wind speed, and indoor temperature, indoor relative humidity and indoor black ball temperature predicted in S7, and matching corresponding indoor thermal comfort data.
In a third aspect, a computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements any one of the indoor thermal comfort data generation methods when executing the computer program.
In a fourth aspect, a computer readable storage medium storing a computer program, wherein the computer program when executed by a processor implements any one of the indoor thermal comfort data generation methods.
Compared with the prior art, the invention has the following beneficial effects:
firstly, the PMV is accurately predicted by measuring three factors of temperature, relative humidity and average radiation temperature, so that the time consumption level of a training stage is reduced, the stability is improved, and the calculation complexity of PMV indexes is reduced; according to a large number of field test results in rural areas in deep customs, two human subjective factors influencing PMV indexes, the metabolic rate and the human clothing thermal resistance are set to be fixed values, the convergence rate is higher, the stability is higher, and the implementation is more convenient; and a limit condition is added to the generation process of the random noise signal, and 20% of redundancy is added to the fluctuation range of the actual PMV index to serve as an additional limit condition, so that the convergence speed of the network is accelerated.
Secondly, the invention reduces the sensitivity of the network to the initial parameters, and does not need the cost of hardware equipment which is larger than that of the GAN network.
Thirdly, the training process of the invention has higher similarity with the way that people learn to grow, and a brand-new way for leading people to guide AI to complete high-level tasks is expanded.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the invention and together with the description serve to explain the invention and not to limit the invention.
In the drawings:
FIG. 1 is a schematic flow chart of a method for generating indoor thermal comfort data according to the present invention;
FIG. 2 is a block diagram of a training model of an indoor thermal comfort data generation method according to the present invention;
FIG. 3 is a comparison of iteration times of different prediction models of the indoor thermal comfort data generation method of the present invention;
FIG. 4 is a network structure diagram of a training model of an indoor thermal comfort data generation method according to the present invention;
FIG. 5 is a diagram of PMV generation effect based on BP neural network;
FIG. 6 is a diagram of PMV generation effect based on GAN network;
fig. 7 is a PMV generation effect diagram of an indoor thermal comfort data generation method according to the present invention.
Detailed Description
The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings. It should be noted that the embodiments and features of the embodiments may be combined with each other without conflict.
The following detailed description is exemplary in nature and is intended to provide further details of the invention. Unless otherwise defined, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention.
Generation of confrontation network gan (genetic adaptive networks) is a new framework proposed by Ian j. Nash equilibrium, well known in game theory, is a core source of thought for generating a countermeasure network. The GAN sets two participants of the game as a Generator (Generator) and a Discriminator (Discriminator), respectively, the game goal of the Generator is to learn the distribution of the real data as much as possible, the Discriminator is to discriminate whether the input data comes from the real data or the Generator as accurately as possible, the two game participants need to continuously perform self optimization to improve self generating capability or discriminating capability in order to finally win the game, and the learning optimization process is to find a nash balance between the two game participants.
Generation of a countermeasure network (GAN) to simulate real data distribution requires sampling, so that too high degree of freedom of sampling causes a GAN convergence process to have certain uncontrollable property when the data volume is very large. And the GAN has a training mode with very low dependency on prior knowledge (initially sampled samples), one of the drawbacks brought by this mode is that the sensitivity of the network to initial parameters is increased, and although the dependency is very low, the dependency still exists. Finally, although the GAN has strong performance, the performance of the GAN model has strong positive correlation with the performance of hardware equipment adopted in the training process, and a huge model needs to use an industry-oriented GPU, so that cost is needed to make GAN training strong.
Example 1
The invention provides a GAN-based method for generating indoor thermal comfort data of rural residences in Guanzhong province, which does not need excessive prior knowledge, can be effectively suitable for non-professional operators and can obviously improve the authenticity of generated data.
A method for generating indoor thermal comfort data of a rural-customs residence based on GAN comprises the following steps,
step 1, collecting real data.
The collected data includes a plurality of sets of indoor temperatures, indoor relative humidities and indoor black ball temperatures over a period of time. The collected data includes training data and discrimination data.
Among six variables affecting PMV thermal comfort, both clothing thermal resistance and human body metabolic rate belong to human subjective factors. The two factors have different values when the human body is in different time, different places and different activity states, and are difficult to determine.
In indoor thermal comfort control in general, it is often the case that appropriate values are directly determined based on a priori knowledge and the specific circumstances to assign these two subjective factors without using them as control variables. And because the wind speed can be defaulted to be the static wind speed indoors, the collected data only comprises three types, and after the real PMV data is obtained, the collected data is divided into training data and judgment data. And the additional information captured in the prior knowledge is used as a limiting condition, and is respectively transmitted to the generation model and the discriminant model to be used as a part of the input layer.
The improvement point is as follows: compared with a BP network, the time consumption level of a training stage is reduced, and the stability of the training stage can be improved; because the network structure of the BP neural network is related to the time consumption level of the training phase and the stability thereof (the network structure becomes more complicated, which not only increases the time consumption level of the training phase, but also can affect the stability thereof), there is a contradiction (i.e., the lower the error precision, the more complicated the network structure)
And step 2, setting limiting conditions.
Taking data intervals obtained by adding 20% of redundancy above and below the collected three types of data ranges as original random noise point limiting conditions;
this original random noise limit condition is used as input to the generator along with the training data.
The generator outputs corresponding generated data, the original random noise limiting condition and the discrimination data together, and a discrimination result is made under the action of the discriminator.
The improvement point is as follows: in a standard GAN network architecture, test data is input to the raw generator in a standard GAN network, and discrimination data and generated data are input to the raw discriminator. Although the training data may be not numerous, the input signal is random, resulting in a wide and random generator, which has the consequence that it takes a long time to train the network to converge. After the restriction condition is added, the training data can be narrowed down once during input, and the convergence time is further shortened.
And step 3, setting a GAN hyper-parameter.
The hyper-parameters of the GAN comprise the number of original random noise points, the limiting conditions of the original random noise points and the training iteration times. Setting the three factors collected in the step 1 as original random noise points, taking the data collected in the step 2 as original noise point limiting conditions, wherein the original random noise points are three of indoor temperature, indoor relative humidity and indoor black ball temperature; the original random noise point limiting conditions are the intervals obtained by adding 20% redundancy respectively above and below the actually measured indoor temperature t, indoor relative humidity h and indoor black ball temperature tmr.
The number of times of iteration of GAN training is related to accuracy, and the process of performing parameter updating on the model by using one time of data by the GAN is one iteration.
The improvement point is as follows: the GAN network has no original limit condition, so that the generator has too free input data, so that the generated data is more and random, which causes the judger to be quite time-consuming, and the invention aims to output PMV comfort which accords with the intelligence in a research room and is suffered by the intelligence more quickly and accurately.
And 4, constructing a generator and a discriminator of the GAN, and determining the activation functions of the hidden layer and the output layer.
The discriminator aims to accomplish the task of distinguishing the input data, i.e. to distinguish whether it is real data or fake data. The generator is directed to optimizing the representation of spurious data on the discriminators and reducing the difference between the representation of generated data on the discriminators and the representation of real data on the discriminators.
The improvement point is as follows: the generator and the discriminator are given additional information (limiting conditions) captured by the priori knowledge, and the additional information is respectively transmitted to the generator and the discriminator to be used as a part of an input layer, so that the GAN fusing the priori knowledge is realized, and the convergence controllability and the convergence speed of the GAN network are improved.
And 5, calculating loss functions of the generator and the discriminator, reversely propagating errors, and training the GAN network.
The discriminator and the generator continuously carry out the countermeasure, and the performances of the discriminator and the generator are continuously upgraded in the countermeasure process. Finally, even if the performance of the discriminator is quite high, the sources of the data samples are difficult to distinguish correctly, and at the moment, the decision generator completely masters the distribution rule of the real data samples. And judging whether the network error meets the condition, and terminating the algorithm when the error reaches the preset precision or the learning times is more than the preset maximum times. Otherwise, selecting the next learning sample and the corresponding expected output to carry out the next round of learning.
The GAN network usually changes its weight and threshold by means of error back-propagation in the process of obtaining the minimum sum of squared errors. The training sample is composed of two parts of input and expected output corresponding to the input. The GAN network uses training samples to train the network repeatedly, which reduces the error between its output and the expected output by continuously changing the network weight and threshold.
The improvement point is as follows: due to the addition of the constraint, the formula of the calculation probability in the formula of the generation countermeasure network (GAN) becomes the form of the conditional probability.
And 6, comparing the PMV data output by the generator in the model with the actual measured value, and judging the performance.
The invention provides a GAN-based method for generating indoor thermal comfort data of a rural residential quarter in the department, starting from modifying the limit conditions generated by random samples according to the specific situation of the indoor thermal comfort prediction problem. The method integrates the physical parameter change characteristics of the building indoor environment in the rural areas in the Guanzhong province captured by the field test in the early stage, increases the limiting conditions, and controls the relation between the output and the input so as to obtain the generated data under the strict limiting conditions.
Example 2
In this embodiment, the practical data of indoor thermal comfort measurement in winter of a certain rural residence in 160 gateways is selected, and the input characteristics include:
indoor air temperature, indoor relative humidity, indoor black ball temperature;
the output characteristics are as follows: indoor PMV.
The first 100 groups of data were used as training data, and the last 60 groups of data were used as discrimination data. The example mainly tests the generation effect of the method, and simultaneously compares the generation effect with the effect of a data generation method based on a BP neural network.
Step 1, collecting real data.
The collected data includes 160 sets of room temperature t, room relative humidity h, and room black-bulb temperature tmr over a period of time.
And 2, taking data intervals obtained by adding 20% of redundancy above and below the acquired data range as original random noise limit conditions.
And 3, setting the hyperparameters of the GAN, wherein the number of the original random noise points is 3, the limiting condition of the original random noise points is set, and the training iteration number is 10000. According to the step 1, the number of the original random noise points is 3, the limit condition of the original random noise points is set according to the step 2, and when the number of times of GAN training iteration is 10000, the accuracy of the verification set can be increased to 60%.
And 4, constructing a generator and a discriminator of the GAN, wherein the activation functions of the hidden layer and the output layer are ReLU.
The relu (rectified Linear unit) function is a very popular activation function in recent years, and its calculation formula is very simple, and f (x) is max (0, x). In the process of back propagation error, when x is 1, the obtained result is 1; when x is 0, the result is 0.
The process of back propagation is: when false samples are generated from the original test data array, the labels of the false samples are all set to 1, i.e., the false samples are considered to be true samples at the time of training of the generator. Since the error is generated by the discriminator at this time, and the purpose of the error feedback is to make the false samples generated by the generator gradually approximate to the true samples. When the false sample is not true and the label is 1, the error given by the discriminator is very large, so that the generator is forced to carry out very large adjustment; on the contrary, when the false sample is true enough and the label is 1, the error given by the discriminator is reduced, thus completing the process that the false sample gradually approaches to the true sample and playing the purpose of confusing the discriminator.
The parameters of the model can be updated using a back-propagation algorithm based on the discriminant model and the loss function of the generative model. The parameters of the discriminant model are updated first, and then the parameters of the generator are updated through the noise data obtained by re-sampling.
The loss function of the generator and the discriminator is calculated and the error is propagated backwards according to:
wherein z is random noise following a gaussian distribution; x is discrimination data; g represents a generator; d represents a discriminator; pdata(x) A probability distribution representing the discrimination data; pz(x) A probability distribution representing random noise; x to PdataMeans for randomly extracting x from the distribution of the discrimination data; z to PzRepresenting the extraction of noise z from gaussian distributed random noise; x | y denotes that x is obtained under the constraint y, and z | y denotes that z is obtained under the constraint y, then the probability expressions are changed from the state in the original GAN network to the present conditional probability formula. D (x | y) and G (z | y) each represent the vector that the arbiter and generator outputs after receiving the input in brackets. For generator G, which expects the sample it generates to fool discriminant D as much as possible, with random noise z as input, it is desirable to maximize the discriminant probability D (G (z | y)), so its objective function is to minimize log (1-D (G (z | y))). For the discriminator D, it is desirable to maximize the discrimination probability D (x | y) while minimizing the discrimination probability D (G (z | y)) in order to distinguish the discrimination data from the spurious generated data as much as possible.
In the generation countermeasure network fusing prior knowledge, probability calculation expression will evolve into conditional probability, the input of the generator is changed into a limiting condition which is added with the prior knowledge and fused with the random noise, and the input of the discriminator is changed into a limiting condition added with the prior knowledge and the generated data and the discriminating data.
The training set of the prior knowledge type GAN network is fused into: { real data, matched constraint }, { real data, unmatched constraint }, { generated data, matched constraint }; where only the first data pair, the discriminator D should decide as correct.
And 6, inputting data to be predicted into the trained generator, outputting the generated thermal comfort data by the generator, and comparing the generated thermal comfort data with an actual measured value to judge the performance.
The experimental result shows that the time for the thermal comfort prediction model based on the fusion prior knowledge type GAN to reach the convergence state is only about one third of that of the thermal comfort prediction model based on the original GAN, and the prediction error of the thermal comfort prediction model based on the fusion prior knowledge type GAN is the minimum among the three. In summary, it can be seen that the prior knowledge-fused GAN network is closer to the actual thermal comfort value PMV than the original GAN network to train the network, and the time consumed for training to converge is shorter than the original GAN network and the BP network.
The method mainly solves the problem of acquiring a plurality of factors influencing the PMV to more quickly and accurately predict the PMV of the real situation. The method can be applied to actually and accurately predicting the PMV sensed by the human body in the research room, so that some real-time measures can be taken to further enable the intelligent agent to reach the comfort level.
Example 3
An indoor thermal comfort data generation system, comprising:
a data acquisition unit: the system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring a plurality of groups of real data in a period of time, and each group of real data comprises indoor temperature, indoor relative humidity and indoor black ball temperature;
a data preprocessing unit: the system is used for dividing the plurality of groups of real data into training data and judging data according to time, and increasing redundancy to obtain an original random noise point limiting condition;
a generation unit: the random noise limiting condition generator is used for inputting training data and an original random noise limiting condition into the generator and outputting generated data by the generator;
a determination unit: the device is used for inputting the original random noise point limiting condition, the discrimination data and the generated data into a discriminator and outputting a discrimination result by the discriminator;
a counter propagation unit: the device is used for updating the parameters of the discriminator according to the discrimination result, and the discriminator reversely transmits errors to the generator and updates the parameters of the generator;
a termination unit: the generator training device is used for terminating the algorithm and finishing generator training when the display error of the judgment result reaches the preset precision or the learning frequency is more than the preset training iteration frequency;
a prediction unit: inputting the current indoor temperature, the indoor relative humidity and the indoor black ball temperature into the generator after training is finished, and outputting the predicted indoor temperature, the indoor relative humidity and the indoor black ball temperature by the generator;
thermal comfort unit: the method is used for obtaining clothing thermal resistance, human body metabolism rate and indoor wind speed, and indoor temperature, indoor relative humidity and indoor black ball temperature predicted in S7, and matching corresponding indoor thermal comfort data.
Example 4
A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, the processor implementing the indoor thermal comfort data generation method of embodiments 1 and 2 when executing the computer program.
Example 5
A computer-readable storage medium storing a computer program which, when executed by a processor, implements the indoor thermal comfort data generation method of embodiments 1 and 2.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
It will be appreciated by those skilled in the art that the invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The embodiments disclosed above are therefore to be considered in all respects as illustrative and not restrictive. All changes which come within the scope of or equivalence to the invention are intended to be embraced therein.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solutions of the present invention and not for limiting the same, and although the present invention is described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.
Claims (10)
1. An indoor thermal comfort data generation method, characterized by comprising the steps of:
s1: collecting a plurality of groups of real data in a period of time, wherein each group of real data comprises indoor temperature, indoor relative humidity and indoor black ball temperature;
s2: dividing the plurality of groups of real data into training data and discriminating data according to time, and increasing redundancy to obtain an original random noise point limiting condition;
s3: inputting training data and original random noise point limiting conditions into a generator, and outputting generated data by the generator;
s4: inputting the original random noise point limiting condition, the discrimination data and the generated data into a discriminator, and outputting a discrimination result by the discriminator;
s5: updating parameters of the discriminator according to the discrimination result, reversely propagating errors to the generator by the discriminator, and updating parameters of the generator;
s6: when the judgment result shows that the error reaches the preset precision or the learning frequency is more than the preset training iteration frequency, terminating the algorithm and finishing the generator training;
s7: inputting the current indoor temperature, the indoor relative humidity and the indoor black ball temperature into the generator after training is finished, and outputting the predicted indoor temperature, the indoor relative humidity and the indoor black ball temperature by the generator;
s8: and acquiring the thermal resistance of the clothes, the human body metabolism rate and the indoor wind speed, and the indoor temperature, the indoor relative humidity and the indoor black ball temperature predicted in the S7, and matching corresponding indoor thermal comfort data.
2. The method of claim 1, wherein the original random noise limit condition is a data interval obtained by adding 20% redundancy to the real data.
3. The method of generating indoor thermal comfort data according to claim 2, wherein the number of training iterations is 10000.
4. The method of claim 2, wherein the predetermined accuracy is close-to-error 10-3。
5. The method of claim 1, wherein the hidden layer and output layer activation functions of the generator and arbiter are f (x) max (0, x), and when x is 1, the result is 1; when x is 0, the result is 0.
6. The method for generating indoor thermal comfort data according to claim 1, wherein in step S5, the corresponding parameters are updated according to the loss functions of the discriminator D and the generator G.
7. The method of generating indoor thermal comfort data according to claim 6, wherein the loss function is:
wherein z is random noise following a gaussian distribution; x is real data; g represents a generator; d represents a discriminator; pdata(x) A probability distribution representing real data; pz(x) A probability distribution representing random noise; x to PdataMeans to randomly draw x from the distribution of the real data; z to PzRepresenting the degree of separation from gaussExtracting noise z from the distributed random noise; x | y represents x obtained under the constraint y, and z | y represents z obtained under the constraint y; d (x | y) represents the vector output by the discriminator D after receiving the input in parentheses; g (z | y) represents the vector that the generator G outputs after receiving the input in brackets.
8. An indoor thermal comfort data generation system, comprising:
a data acquisition unit: the system comprises a data acquisition module, a data processing module and a data processing module, wherein the data acquisition module is used for acquiring a plurality of groups of real data in a period of time, and each group of real data comprises indoor temperature, indoor relative humidity and indoor black ball temperature;
a data preprocessing unit: the system is used for dividing the plurality of groups of real data into training data and judging data according to time, and increasing redundancy to obtain an original random noise point limiting condition;
a generation unit: the random noise limiting condition generator is used for inputting training data and an original random noise limiting condition into the generator and outputting generated data by the generator;
a determination unit: the device is used for inputting the original random noise point limiting condition, the discrimination data and the generated data into a discriminator and outputting a discrimination result by the discriminator;
a counter propagation unit: the device is used for updating the parameters of the discriminator according to the discrimination result, and the discriminator reversely transmits errors to the generator and updates the parameters of the generator;
a termination unit: the generator training device is used for terminating the algorithm and finishing generator training when the display error of the judgment result reaches the preset precision or the learning frequency is more than the preset training iteration frequency;
a prediction unit: inputting the current indoor temperature, the indoor relative humidity and the indoor black ball temperature into the generator after training is finished, and outputting the predicted indoor temperature, the indoor relative humidity and the indoor black ball temperature by the generator;
thermal comfort unit: the method is used for obtaining clothing thermal resistance, human body metabolism rate and indoor wind speed, and indoor temperature, indoor relative humidity and indoor black ball temperature predicted in S7, and matching corresponding indoor thermal comfort data.
9. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that the processor implements the indoor thermal comfort data generation method of any one of claims 1 to 7 when executing the computer program.
10. A computer-readable storage medium, in which a computer program is stored, which, when being executed by a processor, implements the indoor thermal comfort data generation method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111436736.XA CN114139937A (en) | 2021-11-29 | 2021-11-29 | Indoor thermal comfort data generation method, system, equipment and medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111436736.XA CN114139937A (en) | 2021-11-29 | 2021-11-29 | Indoor thermal comfort data generation method, system, equipment and medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114139937A true CN114139937A (en) | 2022-03-04 |
Family
ID=80389230
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111436736.XA Pending CN114139937A (en) | 2021-11-29 | 2021-11-29 | Indoor thermal comfort data generation method, system, equipment and medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114139937A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114841464A (en) * | 2022-05-25 | 2022-08-02 | 山东大卫国际建筑设计有限公司 | Building energy-saving management method, equipment and medium based on chimpanzee algorithm |
-
2021
- 2021-11-29 CN CN202111436736.XA patent/CN114139937A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114841464A (en) * | 2022-05-25 | 2022-08-02 | 山东大卫国际建筑设计有限公司 | Building energy-saving management method, equipment and medium based on chimpanzee algorithm |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105488528B (en) | Neural network image classification method based on improving expert inquiry method | |
| CN110751318B (en) | Ultra-short-term power load prediction method based on IPSO-LSTM | |
| CN112633604B (en) | Short-term power consumption prediction method based on I-LSTM | |
| CN110059852A (en) | A kind of stock yield prediction technique based on improvement random forests algorithm | |
| CN114841257B (en) | Small sample target detection method based on self-supervision comparison constraint | |
| CN111428849A (en) | Improved particle swarm algorithm-based transfer function model parameter identification method and device | |
| CN114925938B (en) | Electric energy meter running state prediction method and device based on self-adaptive SVM model | |
| CN113903395A (en) | BP neural network copy number variation detection method and system for improving particle swarm optimization | |
| CN113722980A (en) | Ocean wave height prediction method, system, computer equipment, storage medium and terminal | |
| CN112163671A (en) | New energy scene generation method and system | |
| Tembusai et al. | K-nearest neighbor with K-fold cross validation and analytic hierarchy process on data classification | |
| CN111144462B (en) | Unknown individual identification method and device for radar signals | |
| CN114139937A (en) | Indoor thermal comfort data generation method, system, equipment and medium | |
| CN109492816A (en) | A kind of coal and gas prominent dynamic prediction method based on hybrid intelligent | |
| CN117313795A (en) | Intelligent building energy consumption prediction method based on improved DBO-LSTM | |
| CN116993548A (en) | Incremental learning-based education training institution credit assessment method and system for LightGBM-SVM | |
| CN109187898B (en) | Soft measurement method and device for ammonia nitrogen content of water in aquaculture environment | |
| CN112016774A (en) | Distribution network running state identification method and system based on data enhancement technology | |
| CN115617882A (en) | Time sequence diagram data generation method and system with structural constraint based on GAN | |
| CN112749211B (en) | Novel tea yield prediction method based on electric power big data | |
| CN114912653A (en) | Short-term load combined prediction method based on self-adaptive chirp modal decomposition and SSA-BilSTM | |
| CN103675012A (en) | Industrial melt index soft measurement instrument and method based on BP particle swarm optimization | |
| CN111369074A (en) | Corn yield prediction method based on artificial bee colony optimized BP neural network | |
| CN113408622A (en) | Non-invasive load identification method and system considering characteristic quantity information expression difference | |
| CN113177675A (en) | Air conditioner cold load prediction method based on optimization neural network of longicorn group algorithm |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |