CN111594996A - Method for predicting air supply quantity of variable air volume air conditioner based on deep belief neural network - Google Patents
Method for predicting air supply quantity of variable air volume air conditioner based on deep belief neural network Download PDFInfo
- Publication number
- CN111594996A CN111594996A CN202010477527.9A CN202010477527A CN111594996A CN 111594996 A CN111594996 A CN 111594996A CN 202010477527 A CN202010477527 A CN 202010477527A CN 111594996 A CN111594996 A CN 111594996A
- Authority
- CN
- China
- Prior art keywords
- training
- air supply
- neural network
- supply quantity
- air conditioner
- 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
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 72
- 238000000034 method Methods 0.000 title claims abstract description 41
- 238000012549 training Methods 0.000 claims abstract description 112
- 238000012360 testing method Methods 0.000 claims abstract description 26
- 210000002569 neuron Anatomy 0.000 claims description 11
- 230000008569 process Effects 0.000 claims description 11
- 230000005855 radiation Effects 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 4
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000012545 processing Methods 0.000 claims description 3
- 238000013480 data collection Methods 0.000 claims description 2
- 238000005265 energy consumption Methods 0.000 abstract description 3
- 230000008878 coupling Effects 0.000 description 3
- 238000010168 coupling process Methods 0.000 description 3
- 238000005859 coupling reaction Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 description 2
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 241000282414 Homo sapiens Species 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000004378 air conditioning Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000010438 heat treatment Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000009423 ventilation Methods 0.000 description 1
Images
Classifications
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F11/00—Control or safety arrangements
- F24F11/62—Control or safety arrangements characterised by the type of control or by internal processing, e.g. using fuzzy logic, adaptive control or estimation of values
- F24F11/63—Electronic processing
- F24F11/64—Electronic processing using pre-stored data
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F11/00—Control or safety arrangements
- F24F11/50—Control or safety arrangements characterised by user interfaces or communication
- F24F11/56—Remote control
- F24F11/58—Remote control using Internet communication
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F11/00—Control or safety arrangements
- F24F11/50—Control or safety arrangements characterised by user interfaces or communication
- F24F11/61—Control or safety arrangements characterised by user interfaces or communication using timers
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F11/00—Control or safety arrangements
- F24F11/70—Control systems characterised by their outputs; Constructional details thereof
- F24F11/72—Control systems characterised by their outputs; Constructional details thereof for controlling the supply of treated air, e.g. its pressure
- F24F11/74—Control systems characterised by their outputs; Constructional details thereof for controlling the supply of treated air, e.g. its pressure for controlling air flow rate or air velocity
- F24F11/77—Control systems characterised by their outputs; Constructional details thereof for controlling the supply of treated air, e.g. its pressure for controlling air flow rate or air velocity by controlling the speed of ventilators
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F2110/00—Control inputs relating to air properties
- F24F2110/10—Temperature
- F24F2110/12—Temperature of the outside air
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F2110/00—Control inputs relating to air properties
- F24F2110/20—Humidity
- F24F2110/22—Humidity of the outside air
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F2110/00—Control inputs relating to air properties
- F24F2110/30—Velocity
- F24F2110/32—Velocity of the outside air
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F2130/00—Control inputs relating to environmental factors not covered by group F24F2110/00
- F24F2130/20—Sunlight
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F24—HEATING; RANGES; VENTILATING
- F24F—AIR-CONDITIONING; AIR-HUMIDIFICATION; VENTILATION; USE OF AIR CURRENTS FOR SCREENING
- F24F2140/00—Control inputs relating to system states
- F24F2140/10—Pressure
- F24F2140/12—Heat-exchange fluid pressure
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02B—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO BUILDINGS, e.g. HOUSING, HOUSE APPLIANCES OR RELATED END-USER APPLICATIONS
- Y02B30/00—Energy efficient heating, ventilation or air conditioning [HVAC]
- Y02B30/70—Efficient control or regulation technologies, e.g. for control of refrigerant flow, motor or heating
Abstract
The invention discloses a method for predicting the air supply quantity of a variable air volume air conditioner based on a deep belief neural network, which comprises the following steps of: (1) determining input parameters for training air supply quantity prediction of the variable air volume air conditioner based on the deep belief neural network; (2) importing the input parameters and the corresponding air supply quantity collected in the step (1) into a deep belief neural network as training samples by using MATLAB software, establishing a training model of an air supply quantity prediction model of the variable air conditioner based on the deep belief neural network, performing learning training on the training samples, and continuously adjusting model parameters to obtain an optimal training model; (3) determining a variable air volume air conditioner air supply quantity prediction model based on a deep belief neural network, introducing a test sample into the prediction model, and predicting the corresponding air supply quantity of the test sample. The design of the invention solves the problem of large building energy consumption caused by insufficient accuracy and stability in the traditional variable air volume air conditioner air supply volume prediction model.
Description
Technical Field
The invention relates to the technical field of air supply quantity prediction of variable air volume air conditioners, in particular to a method for predicting air supply quantity of a variable air volume air conditioner based on a deep belief neural network.
Background
With the increasing of the energy consumption ratio of air conditioners in buildings, variable air volume air conditioners (VAVs) have become the research object in the field of air conditioners at home and abroad with significant energy saving advantages. The variable air volume air conditioner relates to a plurality of parameters such as temperature, humidity, pressure and flow, and due to the nonlinearity and the mutual coupling characteristic among the parameters, the prediction of the air volume of the variable air volume air conditioner is inaccurate and unstable, and the energy conservation and the indoor comfort of the variable air volume air conditioner are influenced. Therefore, it is an urgent problem to accurately predict the air output of the variable air volume air conditioner.
The above background disclosure is only for the purpose of assisting understanding of the concept and technical solution of the present invention and does not necessarily belong to the prior art of the present patent application, and should not be used for evaluating the novelty and inventive step of the present application in the case that there is no clear evidence that the above content is disclosed at the filing date of the present patent application.
Disclosure of Invention
The invention provides a method for predicting the air supply quantity of the variable air volume air conditioner based on the deep belief neural network, aiming at the technical problems, so as to realize the accurate prediction of the VAV air supply quantity and improve the prediction accuracy.
In order to achieve the purpose, the invention adopts the following technical scheme:
a method for predicting the air supply quantity of a variable air volume air conditioner based on a deep belief neural network is characterized by comprising the following steps of:
the method comprises the following steps: determining input parameters for training air supply quantity prediction of the variable air volume air conditioner based on the deep belief neural network, and collecting data to obtain corresponding sample data, wherein the sample data comprises a test sample and training sample data;
step two: using MATLAB software to introduce the input parameters of the variable air volume air conditioner air supply quantity prediction based on the deep belief neural network collected in the step one and the corresponding air supply quantity into the deep belief neural network as training samples, establishing a training model of the variable air volume air conditioner air supply quantity prediction model based on the deep belief neural network, performing learning training on the training samples, and continuously adjusting the parameters of the model to obtain an optimal training model;
step three: and determining a variable air volume air conditioner air supply quantity prediction model based on the deep confidence neural network, introducing the test sample into the trained prediction model, and predicting the corresponding air supply quantity of the test sample.
Further, the input parameters for predicting the air supply quantity of the variable air volume air conditioner in the first step comprise: the air supply quantity at the time t of the previous day, the air supply quantities at the time t of the previous two days, the outdoor temperature at the time t of the previous day, the outdoor temperatures at the time t of the previous two days, the outdoor temperature at the time t-1 of the current day, the outdoor temperature at the time t-2 of the current day, the solar radiation temperature of the current day and the atmospheric humidity of the current day.
Further, in the second step, a variable air volume air conditioner air supply amount prediction model based on a deep belief neural network is established to perform learning training on the training samples, and parameters of the model are continuously adjusted to obtain an optimal training model, wherein the establishment of the training model comprises the following steps:
s1: carrying out standardization processing on model input parameters;
s2: carrying out sample classification;
s3: initializing a weight value and a threshold value of the deep belief neural network;
s4: calculating the updating rule of the weight and the threshold of the deep belief neural network;
s5: setting the number of neurons of an input layer, each hidden layer and an output layer;
s6: training a DBN model;
s7: and predicting the air supply amount corresponding to the test sample by using the trained deep confidence neural network.
Further, the sample classification in step S2 is implemented as follows: dividing the sample data into a training sample and a test sample, wherein the training sample is used for learning and training, and the test sample is used for testing; and importing the training samples into MATLAB software in a matrix form, taking 9 parameters of the air supply quantity at the t moment of the previous day, the air supply quantities at the t moments of the previous two days, the outdoor temperature at the t moment of the previous day, the outdoor temperature at the t moment of the previous two days, the outdoor temperature at the t moment of the current day, the outdoor temperature at the t-1 moment of the current day, the outdoor temperature at the t-2 moment of the current day, the solar radiation temperature of the current day and the atmospheric humidity of the current day as input parameters, taking the air supply quantities as output results, and establishing a deep belief neural network training model for deep belief neural network learning training.
Furthermore, the variable air volume air conditioner air supply quantity prediction model based on the deep belief neural network has a four-layer structure; the first layer is a display layer, namely an input layer, and consists of outdoor meteorological parameters and historical air supply quantity of a variable air conditioner; the second layer and the third layer are hidden layers, and the last layer is an output layer, namely the air supply quantity of the variable air conditioner at the predicted day t in the model.
Further, the outdoor meteorological parameters during the data collection period comprise outdoor average atmospheric pressure, outdoor average wind speed, outdoor calculated dry bulb average temperature, outdoor calculated wet bulb average humidity and outdoor calculated daily average temperature.
Further, the DBN is stacked by a number of Restricted Boltzmann Machines (RBMs); and in the RBM training process, the DBN solves the activation probability of the display layer and the hidden layer according to the energy function of the RBM, and then obtains the update rule of the RBM training parameters.
Further, the training process of the DBN model is mainly divided into two steps:
firstly, obtaining parameters through layer-by-layer training in RBM; the output end is connected with a BP neural network to obtain output data, and the forward stage of the whole DBN training is completed;
and secondly, reversely transmitting the error to the RBMs of each layer from top to bottom according to the error of the actual data and the output data, and finely adjusting the RBMs of each layer by layer, so that the parameters reach the global optimum, and the training of the whole DBN is completed.
Furthermore, in the second step of the DBN model training, the t-time actual air supply volume of the variable air volume air conditioner is taken as an output variable of the model, the t-time air supply volume of the previous day, the t-time air supply volumes of the previous two days, the t-time outdoor temperature of the previous day, the t-time outdoor temperatures of the previous two days, the t-time outdoor temperature of the current day, the t-1-time outdoor temperature of the current day, the t-2-time outdoor temperature of the current day, the solar radiation temperature of the current day, and the atmospheric humidity of the current day are taken as input variables of the model, and the output variable and the input variables form a corresponding training data set S; and then, importing the training data set S into the DBN model, performing layer-by-layer training by using the RBM training parameter updating rule, performing unsupervised pre-training, and performing supervised reverse tuning training until the training error meets the set requirement, and stopping updating to obtain the trained optimal DBN model.
Compared with the prior art, the invention has the beneficial effects that:
(1) the method is based on deep neural network modeling, classifies and inputs factors influencing the air output of the variable air volume air conditioner into the algorithm, contributes to more comprehensively reflecting the prediction of the air output, is more advanced theoretically, and can overcome the mutual coupling interference among different parameters. Secondly, the deep confidence neural network is utilized to establish the prediction of the air supply quantity of the variable air volume air conditioner, so that the problems of local convergence, low convergence speed, long training time when a large amount of data is generated, overfitting and the like of BP, Elman and fuzzy shallow neural networks can be avoided, and the multiple limitations in practical application can be overcome. The accuracy of air supply quantity prediction of the variable air volume air conditioner in actual conditions can be better ensured, and the requirements in practical application are met.
(2) The invention introduces the deep learning theory into the prediction field of the air output of the variable air volume air conditioner for the first time, and accurately predicts the air output by utilizing the DBN model, thereby reducing the energy consumption of the building. The air supply quantity of the variable air volume air conditioner can be accurately predicted, new ideas and inspirations can be provided for researching a novel energy-saving variable air volume air conditioner, the assistance is indirectly provided for human beings to provide comfortable living environment, and the method has great significance for the research of the future variable air volume air conditioner.
Drawings
FIG. 1 is a flow chart of the DBN-based variable air volume air conditioner air supply volume prediction;
FIG. 2 is a deep belief neural network structure;
FIG. 3 is a DBN training process;
FIG. 4 is a schematic diagram of a comparison between a predicted value and an actual value of a deep belief neural network;
fig. 5 is a comparison graph of DBN model prediction error.
Detailed Description
The following detailed description of the present invention is provided in conjunction with the accompanying drawings, but it should be understood that the scope of the present invention is not limited to the specific embodiments. It should be understood that the directions "up", "down", "left" and "right" mentioned in the following embodiments of the present invention are based on the positions of the corresponding drawings. These directional terms are used for convenience of description only and do not represent limitations on the particular embodiments of the present invention. Unless otherwise specified, like reference numerals in the reference numerals refer to like structures. Other embodiments, which are within the scope of the invention, are contemplated by those skilled in the art, based on the teachings herein, and are obtained without the exercise of inventive faculty.
The invention relates to a method for predicting the air supply quantity of a variable air volume air conditioner based on a deep belief neural network, which applies the deep belief neural network to the prediction of the air supply quantity of a VAV variable air conditioner for the first time, and specifically comprises the following steps:
the method comprises the following steps: determining input parameters for training air supply quantity prediction of the variable air volume air conditioner based on the deep belief neural network, and collecting data to obtain corresponding sample data, wherein the sample data comprises a test sample and training sample data;
step two: using MATLAB software to introduce the input parameters of the variable air volume air conditioner air supply quantity prediction based on the deep belief neural network collected in the step one and the corresponding air supply quantity into the deep belief neural network as training samples, establishing a training model of the variable air volume air conditioner air supply quantity prediction model based on the deep belief neural network, performing learning training on the training samples, and continuously adjusting the parameters of the model to obtain an optimal training model;
step three: and determining a variable air volume air conditioner air supply quantity prediction model based on the deep confidence neural network, introducing the test sample into the trained prediction model, and predicting the corresponding air supply quantity of the test sample.
In the scheme of the invention, the input parameters for predicting the air supply quantity of the variable air volume air conditioner in the first step comprise: the air supply quantity at the time t of the previous day, the air supply quantities at the time t of the previous two days, the outdoor temperature at the time t of the previous day, the outdoor temperatures at the time t of the previous two days, the outdoor temperature at the time t-1 of the current day, the outdoor temperature at the time t-2 of the current day, the solar radiation temperature of the current day and the atmospheric humidity of the current day. And in the first step: the mathematical relationship between each influence factor and the air supply in the VAV air supply prediction model is as follows:
y=f(x1,x2,···,xn)i=1,2,···,n (1.1)
wherein y is the air volume at time t of VAV, xiThe parameter of the ith air volume influence, namely the input parameter of the air volume prediction of the variable air volume air conditioner in the step one, n is the number of factors influencing the air volume, and f is the uncertain functional relation of DBN. See fig. 1 is based onThe flow chart for predicting the air supply quantity of the variable air volume air conditioner of the DBN comprises the following calculation steps:
(1) forming a training data set S:
and taking the actual air supply quantity at the time t of the VAV as an output variable of the model, taking the air supply quantity at the time t of the previous day, the air supply quantities at the time t of the previous two days, the outdoor temperature at the time t of the previous day, the outdoor temperature at the time t of the previous two days, the outdoor temperature at the time t of the current day, the outdoor temperature at the time t-1 of the current day, the outdoor temperature at the time t-2 of the current day, the solar radiation temperature of the current day and the atmospheric humidity of the current day as input variables of the model, and forming a corresponding training data set S by the output.
(2) Training the DBN model:
and (3) importing the training data set S into the DBN model, performing layer-by-layer training by using the RBM training parameter updating rule, performing unsupervised pre-training, and performing supervised reverse tuning training until the training error meets the set requirement, and stopping updating to obtain the trained optimal DBN model, namely the deep belief neural network model.
Test data set D corresponding VAV air supply: and (4) introducing the test sample into the trained DBN model, and predicting the VAV air supply amount y at the time t.
Furthermore, the outdoor meteorological parameters and the air supply quantity of the large-scale variable air volume air conditioner of a certain cultural sports center in Guangxi are subjected to field data acquisition, and the outdoor meteorological parameters and the building parameters are shown in the following table 1:
TABLE 1 outdoor weather and building parameters
In the data acquisition process, a temperature and humidity self-recorder, a black ball thermometer and a heating ventilation air conditioning electronic air volume measuring instrument FLY-1 are used for measuring outdoor meteorological parameters and VAV air volume of a cultural sports center hall, and the parameters of the measuring instrument are shown in a table 2:
TABLE 2. measuring instrument models and parameters
Taking the period from 6 middle of the month in 2019 to 9 middle of the month in 2019 as an example, the input parameters for predicting the air supply quantity of the variable air volume air conditioner are collected: the 3 measuring instruments are set to automatically record once every 10 minutes, recorded data are sorted, abnormal data are eliminated, and 1000 groups of data are collected to serve as training samples. In three days from 16 days to 18 days in 9 months in 2019, setting 3 measuring instruments to automatically record once every 1 hour, collecting 13 groups of data as a prediction sample, and according to the measured data of the VAV hourly air supply volume and the outdoor hourly temperature in two days of 16 days and 17 days in 9 months and the outdoor meteorological parameters in 18 days in 9 months, utilizing the optimal model which completes training to carry out the comparison on 8 days in 18 days in 9 months: and predicting the hourly air supply amount of 00-22: 00.
In the scheme of the invention, in the second step, a variable air volume air conditioner air supply quantity prediction model based on a deep belief neural network is established to carry out learning training on a training sample, and parameters of the model are continuously adjusted to obtain an optimal training model, wherein the establishment of the training model comprises the following steps:
s1: carrying out standardization processing on model input parameters;
s2: carrying out sample classification;
s3: initializing a weight value and a threshold value of the deep belief neural network;
s4: calculating the updating rule of the weight and the threshold of the deep belief neural network;
s5: setting the number of neurons of an input layer, each hidden layer and an output layer;
s6: training a DBN model;
s7: and predicting the air supply amount corresponding to the test sample by using the trained deep confidence neural network.
Wherein the sample classification in step S2 is implemented as follows: dividing the sample data into a training sample and a test sample, wherein the training sample is used for learning and training, and the test sample is used for testing; and importing the training samples into MATLAB software in a matrix form, taking 9 parameters of the air supply quantity at the t moment of the previous day, the air supply quantities at the t moments of the previous two days, the outdoor temperature at the t moment of the previous day, the outdoor temperature at the t moment of the previous two days, the outdoor temperature at the t moment of the current day, the outdoor temperature at the t-1 moment of the current day, the outdoor temperature at the t-2 moment of the current day, the solar radiation temperature of the current day and the atmospheric humidity of the current day as input parameters, taking the air supply quantities as output results, and establishing a deep belief neural network training model for deep belief neural network learning training.
Referring to fig. 2, the air volume prediction method of the variable air volume air conditioner based on the deep belief neural network of the present invention comprises the following steps: air volume of the variable air conditioner DBN model: deep belief neural network (DBN) is an important algorithm for Deep learning, and is composed of an input layer, an intermediate layer (hidden layer) and an output layer. The method is characterized in that the whole neural network is quickly trained layer by layer from bottom to top by using an unsupervised layer by layer training method, so that the network has higher training speed.
The structure of the DBN algorithm-based VAV air volume prediction model is shown in fig. 2, and has a four-layer structure. The first layer is a display layer, namely an input layer, and consists of outdoor meteorological parameters and VAV historical air supply volume; the second layer and the third layer are hidden layers, and the double-hidden-layer structure can avoid local convergence on the premise of ensuring the prediction precision. The last layer is the output layer, i.e. the predicted VAV air supply at time t of day in the model. The DBN is stacked from a number of Restricted Boltzmann Machines (RBMs). In the process of training the RBM, the DBN usually solves the activation probability of the explicit layer and the hidden layer according to the energy function of the RBM, and then obtains the update rule of the RBM training parameters. An RBM is a special neural network structure based on energy functions, and for any RBM structure, the energy function can be defined as:
wherein I and J are respectively the number of display layer and hidden layer units, viAnd hjRespectively, the ith apparent layer neuron and the jth hidden layer neuron, aiAnd bjRepresents the bias of the ith explicit layer neuron and the jth hidden layer neuron, respectively, wijIs connecting viAnd hjWeight in between.
Recalculating the connection weight and bias between the apparent layer neuron and the hidden layer neuron according to the activation probability of the hidden layer neuron and the apparent layer neuron, and obtaining RBM parameter theta ═ { w ═ij,ai,bjUpdate rule of }:
where η is the learning rate, and in a four-layer DBN structure, η generally takes the value of 0.01.
In the whole DBN training process, a vector is generated by the display layer of the first RBM, then the vector is transmitted to the hidden layer through the RBM network, the display layer can be reconstructed in turn, the weight and the offset between the hidden layer and the display layer are updated according to the difference between the reconstructed layer and the display layer, and the updating is stopped until the training error reaches the set requirement. The DBN training process is shown in FIG. 3, and the main steps are divided into two steps:
in the first step, in RBM, a parameter theta is obtained through layer-by-layer training. The output end is connected with the BP neural network to obtain output data, and the forward stage of the whole DBN training is completed.
And secondly, reversely transmitting the error to the RBM of each layer from top to bottom according to the error of the actual data and the output data, and finely adjusting the RBM of each layer by layer to ensure that the parameter theta is globally optimal, thereby finishing the training of the whole DBN.
Referring to fig. 4 and 5, the method for predicting the air supply quantity of the variable air conditioner based on the deep belief neural network of the present invention comprises the following three steps: and (4) finishing the whole DBN training by using the training sample data collected in the first step to obtain an optimal prediction model. Then, the input parameters of the test sample are led into an optimal prediction model, the air supply quantity of the corresponding variable air conditioner is predicted, the hourly air supply quantity and the actual air supply quantity of the variable air conditioner predicted by using the deep belief neural network are shown in fig. 4, and the prediction error is shown in fig. 5.
Further, the performance of the deep belief neural network model for predicting VAV air supply is evaluated, and the average absolute relative error (MAPE), the root mean square relative error (RMSPE) and a coefficient of determination (R) are adopted2) To evaluate the predictive performance of the model.
Wherein, yk、ypRepresenting the actual value and the predicted value,represents the average of the actual values and the average of the predicted values.
Generally, smaller values of MAPE indicate better prediction performance. The RMSPE is a better index in the prediction process, and the smaller the value of the RMSPE is, the smaller the deviation between the predicted value and the actual value is, and the better the prediction performance is. R2Is an effective index of prediction accuracy, R2The larger the value, the higher the linear relationship between the predicted value and the actual value. The average absolute relative error, the root mean square relative error and the decision coefficient of the depth confidence neural network prediction model for predicting the hourly air supply volume of the variable air volume air conditioner are obtained through calculation, and are respectively 1.555%, 0.789% and 0.9975. Therefore, the deep confidence neural network prediction model has global convergence and better prediction performance, and can accurately and stably predict the hourly air supply quantity of the variable air volume air conditioner.
Through the analysis of the prediction result of the variable air volume air conditioner air supply amount prediction model based on the DBN, the theory of the variable air volume air conditioner air supply amount prediction model based on the DBN is more advanced, and the prediction error is smaller after a large amount of data are trained. The method can process mass data and has the advantages of high convergence speed, global convergence and stable prediction. However, in the field of the current variable air volume air conditioner, the deep learning is less in application. The DBN model is applied to a complex model influencing the air output of the variable air volume air conditioner, so that a good effect is achieved, and the interference of mutual coupling between different influencing factors can be overcome. The DBN theory is applied to the field of variable air volume air conditioners, a more advanced intelligent algorithm can be provided for researching the air volume under the complex environment condition, and greater help is provided for the research of the novel energy-saving variable air volume air conditioner.
While there has been described and illustrated what are considered to be example embodiments of the present invention, it will be understood by those skilled in the art that various changes and substitutions may be made therein without departing from the spirit of the invention. In addition, many modifications may be made to adapt a particular situation to the teachings of the present invention without departing from the central concept described herein. Therefore, it is intended that the invention not be limited to the particular embodiments disclosed, but that the invention will include all embodiments and equivalents falling within the scope of the invention.
Claims (9)
1. A method for predicting the air supply quantity of a variable air volume air conditioner based on a deep belief neural network is characterized by comprising the following steps of:
the method comprises the following steps: determining input parameters for training air supply quantity prediction of the variable air volume air conditioner based on the deep belief neural network, and collecting data to obtain corresponding sample data, wherein the sample data comprises a test sample and training sample data;
step two: using MATLAB software to introduce the input parameters of the variable air volume air conditioner air supply quantity prediction based on the deep belief neural network collected in the step one and the corresponding air supply quantity into the deep belief neural network as training samples, establishing a training model of the variable air volume air conditioner air supply quantity prediction model based on the deep belief neural network, performing learning training on the training samples, and continuously adjusting the parameters of the model to obtain an optimal training model;
step three: and determining a variable air volume air conditioner air supply quantity prediction model based on the deep confidence neural network, introducing the test sample into the trained prediction model, and predicting the corresponding air supply quantity of the test sample.
2. The method for predicting the air supply quantity of the variable air volume air conditioner based on the deep belief neural network is characterized in that: the input parameters for predicting the air supply quantity of the variable air volume air conditioner in the first step comprise: the air supply quantity at the time t of the previous day, the air supply quantities at the time t of the previous two days, the outdoor temperature at the time t of the previous day, the outdoor temperatures at the time t of the previous two days, the outdoor temperature at the time t-1 of the current day, the outdoor temperature at the time t-2 of the current day, the solar radiation temperature of the current day and the atmospheric humidity of the current day.
3. The method for predicting the air supply quantity of the variable air volume air conditioner based on the deep belief neural network is characterized in that: in the second step, a variable air volume air conditioner air supply quantity prediction model based on a deep belief neural network is established to carry out learning training on a training sample, and parameters of the model are continuously adjusted to obtain an optimal training model, wherein the establishment of the training model comprises the following steps:
s1: carrying out standardization processing on model input parameters;
s2: carrying out sample classification;
s3: initializing a weight value and a threshold value of the deep belief neural network;
s4: calculating the updating rule of the weight and the threshold of the deep belief neural network;
s5: setting the number of neurons of an input layer, each hidden layer and an output layer;
s6: training a DBN model;
s7: and predicting the air supply amount corresponding to the test sample by using the trained deep confidence neural network.
4. The method for predicting the air supply quantity of the variable air volume air conditioner based on the deep belief neural network is characterized in that: the sample classification in step S2 is implemented as follows: dividing the sample data into a training sample and a test sample, wherein the training sample is used for learning and training, and the test sample is used for testing; and importing the training samples into MATLAB software in a matrix form, taking 9 parameters of the air supply quantity at the t moment of the previous day, the air supply quantities at the t moments of the previous two days, the outdoor temperature at the t moment of the previous day, the outdoor temperature at the t moment of the previous two days, the outdoor temperature at the t moment of the current day, the outdoor temperature at the t-1 moment of the current day, the outdoor temperature at the t-2 moment of the current day, the solar radiation temperature of the current day and the atmospheric humidity of the current day as input parameters, taking the air supply quantities as output results, and establishing a deep belief neural network training model for deep belief neural network learning training.
5. The method for predicting the air supply quantity of the variable air volume air conditioner based on the deep belief neural network is characterized in that: the variable air volume air conditioner air supply quantity prediction model based on the deep belief neural network has a four-layer structure; the first layer is a display layer, namely an input layer, and consists of outdoor meteorological parameters and historical air supply quantity of a variable air conditioner; the second layer and the third layer are hidden layers, and the last layer is an output layer, namely the air supply quantity of the variable air conditioner at the predicted day t in the model.
6. The method for predicting the air supply quantity of the variable air volume air conditioner based on the deep belief neural network as claimed in claim 5, wherein the method comprises the following steps: the outdoor meteorological parameters during the data collection period comprise outdoor average atmospheric pressure, outdoor average wind speed, outdoor calculated dry bulb average temperature, outdoor calculated wet bulb average humidity and outdoor calculated daily average temperature.
7. The method for predicting the air supply quantity of the variable air volume air conditioner based on the deep belief neural network as claimed in claim 5, wherein the method comprises the following steps: the DBN is stacked by a number of Restricted Boltzmann Machines (RBMs); and in the RBM training process, the DBN solves the activation probability of the display layer and the hidden layer according to the energy function of the RBM, and then obtains the update rule of the RBM training parameters.
8. The method for predicting the air supply quantity of the variable air volume air conditioner based on the deep belief neural network as claimed in claim 7, wherein the method comprises the following steps: the training process of the DBN model is mainly divided into two steps:
firstly, in RBM, obtaining a parameter theta through layer-by-layer training; the output end is connected with a BP neural network to obtain output data, and the forward stage of the whole DBN training is completed;
and secondly, reversely transmitting the error to the RBM of each layer from top to bottom according to the error of the actual data and the output data, and finely adjusting the RBM of each layer by layer to ensure that the parameter theta is globally optimal, thereby finishing the training of the whole DBN.
9. The method for predicting the air supply quantity of the variable air volume air conditioner based on the deep belief neural network as claimed in claim 8, wherein the method comprises the following steps: in the second step of the DBN model training, the actual air supply quantity at the t moment of the variable air volume air conditioner is taken as an output variable of the model, the air supply quantity at the t moment of the previous day, the air supply quantities at the t moments of the previous two days, the outdoor temperature at the t moment of the previous day, the outdoor temperatures at the t moments of the previous two days, the outdoor temperature at the t moment of the current day, the outdoor temperature at the t-1 moment of the current day, the outdoor temperature at the t-2 moment of the current day, the solar radiation temperature of the current day and the atmospheric humidity of the current day are taken as input variables of the model, and the output variable and the; and then, importing the training data set S into the DBN model, performing layer-by-layer training by using the RBM training parameter updating rule, performing unsupervised pre-training, and performing supervised reverse tuning training until the training error meets the set requirement, and stopping updating to obtain the trained optimal DBN model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010477527.9A CN111594996A (en) | 2020-05-29 | 2020-05-29 | Method for predicting air supply quantity of variable air volume air conditioner based on deep belief neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010477527.9A CN111594996A (en) | 2020-05-29 | 2020-05-29 | Method for predicting air supply quantity of variable air volume air conditioner based on deep belief neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN111594996A true CN111594996A (en) | 2020-08-28 |
Family
ID=72184282
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010477527.9A Pending CN111594996A (en) | 2020-05-29 | 2020-05-29 | Method for predicting air supply quantity of variable air volume air conditioner based on deep belief neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111594996A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112093025A (en) * | 2020-09-23 | 2020-12-18 | 江苏科技大学 | Control system and control method for variable air volume air distributor air valve |
CN113446683A (en) * | 2021-06-30 | 2021-09-28 | 广州地铁设计研究院股份有限公司 | Intelligent room air conditioner for subway station equipment and management |
CN113446684A (en) * | 2021-06-30 | 2021-09-28 | 广州地铁设计研究院股份有限公司 | Intelligent air conditioner for public area of subway station |
CN113671361A (en) * | 2021-06-25 | 2021-11-19 | 海南电网有限责任公司海南输变电检修分公司 | High-voltage circuit breaker characteristic parameter prediction method and system based on multi-source signal fusion |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH06147598A (en) * | 1992-11-09 | 1994-05-27 | Takasago Thermal Eng Co Ltd | Air conditioning load prediction method |
CN107223195A (en) * | 2015-02-24 | 2017-09-29 | 西门子工业公司 | Variable air quantity for HVAC system is modeled |
CN107725283A (en) * | 2017-09-19 | 2018-02-23 | 江苏方天电力技术有限公司 | A kind of fan trouble detection method based on depth belief network model |
-
2020
- 2020-05-29 CN CN202010477527.9A patent/CN111594996A/en active Pending
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JPH06147598A (en) * | 1992-11-09 | 1994-05-27 | Takasago Thermal Eng Co Ltd | Air conditioning load prediction method |
CN107223195A (en) * | 2015-02-24 | 2017-09-29 | 西门子工业公司 | Variable air quantity for HVAC system is modeled |
CN107725283A (en) * | 2017-09-19 | 2018-02-23 | 江苏方天电力技术有限公司 | A kind of fan trouble detection method based on depth belief network model |
Non-Patent Citations (3)
Title |
---|
单文丽等: "深度置信网络拟合异步电动机逆变器控制开关表", 微特电机, vol. 47, no. 2, pages 76 - 82 * |
朱栋华等: "基于模糊神经网络变风量空调系统风量预测研究", 建筑节能, vol. 38, no. 227, pages 22 - 24 * |
杨国田等: "基于变量选择的深度置信神经网络锅炉烟气NOx排放预测", 热力发电, vol. 49, no. 4, pages 34 - 40 * |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112093025A (en) * | 2020-09-23 | 2020-12-18 | 江苏科技大学 | Control system and control method for variable air volume air distributor air valve |
CN113671361A (en) * | 2021-06-25 | 2021-11-19 | 海南电网有限责任公司海南输变电检修分公司 | High-voltage circuit breaker characteristic parameter prediction method and system based on multi-source signal fusion |
CN113446683A (en) * | 2021-06-30 | 2021-09-28 | 广州地铁设计研究院股份有限公司 | Intelligent room air conditioner for subway station equipment and management |
CN113446684A (en) * | 2021-06-30 | 2021-09-28 | 广州地铁设计研究院股份有限公司 | Intelligent air conditioner for public area of subway station |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111594996A (en) | Method for predicting air supply quantity of variable air volume air conditioner based on deep belief neural network | |
CN110084367B (en) | Soil moisture content prediction method based on LSTM deep learning model | |
CN112116147A (en) | River water temperature prediction method based on LSTM deep learning | |
CN110647979B (en) | Greenhouse environment multi-parameter intelligent monitoring system based on Internet of things | |
CN112288164B (en) | Wind power combined prediction method considering spatial correlation and correcting numerical weather forecast | |
CN111639462B (en) | Building indoor thermal comfort prediction method in natural ventilation environment based on deep belief neural network | |
CN111626518A (en) | Urban daily water demand online prediction method based on deep learning neural network | |
CN105354620A (en) | Method for predicting fan generation power | |
Lee et al. | Using climate classification to evaluate building energy performance | |
CN108399470B (en) | Indoor PM2.5 prediction method based on multi-example genetic neural network | |
Han et al. | Energy saving of buildings for reducing carbon dioxide emissions using novel dendrite net integrated adaptive mean square gradient | |
CN112036658B (en) | Wind power real-time prediction method based on forgetting long-term and short-term memory with difference | |
CN112182709B (en) | Method for rapidly predicting water drainage temperature of large reservoir stoplog gate layered water taking facility | |
CN113705877A (en) | Real-time monthly runoff forecasting method based on deep learning model | |
CN113112077A (en) | HVAC control system based on multi-step prediction deep reinforcement learning algorithm | |
CN111461413A (en) | Highway road surface performance detecting system | |
CN115186803A (en) | Data center computing power load demand combination prediction method and system considering PUE | |
CN111160659A (en) | Power load prediction method considering temperature fuzzification | |
CN116167531A (en) | Photovoltaic power generation prediction method based on digital twin | |
CN116757057A (en) | Air quality prediction method based on PSO-GA-LSTM model | |
CN113960925A (en) | Building energy consumption control method and device based on artificial intelligence | |
CN113111419A (en) | Method and system for establishing and predicting air-conditioning load prediction model in office building | |
CN114386672B (en) | Environment big data Internet of things intelligent detection system | |
CN115907204A (en) | Forest transpiration water consumption prediction method for optimizing BP neural network by sparrow search algorithm | |
CN111461462B (en) | Daily load prediction method based on trellis Net-LSTM |
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 |