CN114065335A - Building energy consumption prediction method based on multi-scale convolution cyclic neural network - Google Patents

Building energy consumption prediction method based on multi-scale convolution cyclic neural network Download PDF

Info

Publication number
CN114065335A
CN114065335A CN202111020470.0A CN202111020470A CN114065335A CN 114065335 A CN114065335 A CN 114065335A CN 202111020470 A CN202111020470 A CN 202111020470A CN 114065335 A CN114065335 A CN 114065335A
Authority
CN
China
Prior art keywords
layer
output
gru
energy consumption
building energy
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
Application number
CN202111020470.0A
Other languages
Chinese (zh)
Inventor
马武彬
顾桐菲
吴亚辉
邓苏
周浩浩
皇甫先鹏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
National University of Defense Technology
Original Assignee
National University of Defense Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by National University of Defense Technology filed Critical National University of Defense Technology
Priority to CN202111020470.0A priority Critical patent/CN114065335A/en
Publication of CN114065335A publication Critical patent/CN114065335A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • 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
    • G06N3/08—Learning methods

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Evolutionary Computation (AREA)
  • General Physics & Mathematics (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Hardware Design (AREA)
  • Software Systems (AREA)
  • Artificial Intelligence (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Medical Informatics (AREA)
  • Architecture (AREA)
  • Civil Engineering (AREA)
  • Structural Engineering (AREA)
  • Computational Mathematics (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a building energy consumption prediction method based on a multi-scale convolution cyclic neural network, which comprises the following steps of: building energy consumption prediction models based on the multi-scale convolution cyclic neural network are built; training the building energy consumption prediction model by using training set data; and inputting the test set data into the trained building energy consumption prediction model, and calculating to obtain the predicted value of the building energy consumption. The method introduces the multi-scale convolutional layer into the recurrent neural network, and attention mechanisms are distributed from different scales, so that the model can acquire historical information from different scales; the bidirectional GRU layer can more fully acquire context information of sequence data, the whole model adopts a convolution structure to fuse recognition outputs of attention mechanisms of different scales, and output is screened and recognized by different scales through convolution connection, so that better accuracy is acquired for prediction of building energy consumption values.

Description

Building energy consumption prediction method based on multi-scale convolution cyclic neural network
Technical Field
The invention belongs to the technical field of building energy consumption prediction, and particularly relates to a building energy consumption prediction method based on a multi-scale convolution cyclic neural network.
Background
The problem of energy consumption is one of the important issues of social widespread concern. The proportion of the building power consumption to the total social power consumption exceeds 50%, and the problem of power consumption prediction of a certain building or a family is one of the key problems, so that attention of vast personnel is attracted. The prediction of the future power consumption can provide early warning for the abnormal use of the power supply, and meanwhile, the power supply system can also provide decision support for power supply strategies and scheduling of power supply departments, and has great significance.
The prediction accuracy for energy consumption is still insufficient at present. The traditional machine learning methods such as linear regression, Support Vector Regression (SVR), random forest, XBBboost, ensemble learning and the like can predict the energy consumption, but because the factors influencing the energy consumption are more and the relationship is more complex, the traditional machine learning methods are difficult to capture the long-term dependence relationship, and the time sequence importance among the factors is not well acquired. Recently, researchers have adopted deep learning methods (RNN, LSTM, GRU, Bi-LSTM, etc.) to predict energy consumption, and the method has a good effect. However, both the conventional machine learning method and the deep learning method which is popular in recent years do not capture the correlation characteristics between the elements from the time sequence, and the prediction accuracy is not ideal.
Disclosure of Invention
In view of the above, in order to solve the problem of accurate prediction of building energy consumption, the present invention aims to provide a method for predicting building energy consumption based on a multi-scale convolution cyclic neural network, which predicts the power consumption of a certain building by combining outdoor air pressure, temperature, humidity, wind power and visible light sensor data with indoor temperature and humidity sensors of the building.
Based on the purpose, the building energy consumption prediction method based on the multi-scale convolution cyclic neural network is provided, and comprises the following steps:
step 1, constructing a building energy consumption prediction model based on a multi-scale convolution cyclic neural network;
step 2, training the building energy consumption prediction model by using training set data, wherein the training set data comprises influence factor data and known building energy consumption data;
and 3, inputting the test set data into the trained building energy consumption prediction model, and calculating to obtain a predicted value of the building energy consumption.
Specifically, the building energy consumption prediction model comprises a first convolution layer, a first bidirectional GRU layer, a first multi-scale convolution layer, a second bidirectional GRU layer, a second multi-scale convolution layer, a first full-connection layer and a second full-connection layer, wherein the layers are sequentially connected, the output of the first convolution layer and the output of the first bidirectional GRU layer are connected and then simultaneously used as the input of the first multi-scale convolution layer and the second multi-scale convolution layer, the bidirectional GRU layer is formed by connecting a forward GRU model and a backward GRU model in parallel to form a bidirectional structure, the bidirectional GRU layer outputs two combined GRU signals, the output layer of the first full connection layer is 100, the output layer of the second full connection layer is 1, the input of the first convolution layer in the building energy consumption prediction model is an influence factor data sequence, and the output of the second full-connection layer is a building energy consumption value.
Specifically, the energy consumption prediction model for the building is
Figure RE-GDA0003299092970000021
x0,...,xTTo influence the factor sequence data, (y)0,...,yK),K<T is a known building energy consumption value, (y)K+1,...,yT) To the extent that a predicted building energy consumption value is required,
Figure RE-GDA0003299092970000022
for the corresponding estimated value, the input is x0,...,xT,y0,...,yKAnd (3) sequentially inputting variables into the building energy consumption prediction model to start training, wherein the loss function adopts standard normalized MSE, and the activation function adopts Relu function.
Specifically, the analytical expression of the building energy consumption prediction model is as follows:
Figure RE-GDA0003299092970000023
Figure RE-GDA0003299092970000024
Figure RE-GDA0003299092970000031
Figure RE-GDA0003299092970000032
Figure RE-GDA0003299092970000033
Figure RE-GDA0003299092970000034
C4 t＝η2([C2 t,C3 t])
Figure RE-GDA0003299092970000035
wherein x istIs an input to the model at time t, η1(. eta.) and eta2(. is) two convolution operations, [, ]]For merge join operations, MutiScalConv (. circle., Scale1) and MutiScalConv (. circle., Scale2) are two multi-ruler scales Scale1 and Scale2, respectivelyThe degree convolution operation, the specific process of fusion convolution, is as follows:
first convolution layer η1(xt) Accepting sequence data xtInput and output of
Figure RE-GDA0003299092970000036
Simultaneously as inputs to the first multi-scale convolutional layer and the second multi-scale convolutional layer;
Figure RE-GDA0003299092970000037
is the output of the first bi-directional GRU layer,
Figure RE-GDA0003299092970000038
indicating the output to be GRU in forward direction
Figure RE-GDA0003299092970000039
And backward GRU output
Figure RE-GDA00032990929700000310
Carrying out merging connection;
Figure RE-GDA00032990929700000311
is to multiply the first bidirectional GRU layer by a weight vector
Figure RE-GDA00032990929700000312
And adding the offset vector
Figure RE-GDA00032990929700000313
The result of (1);
will be provided with
Figure RE-GDA00032990929700000314
And η1(xt) Output of (2)
Figure RE-GDA00032990929700000315
Are combined into
Figure RE-GDA00032990929700000316
As an input to the first multi-scale convolutional layer;
Figure RE-GDA00032990929700000317
is the output of the first multi-Scale convolutional layer with Scale1, connected to the second bidirectional GRU layer;
Figure RE-GDA00032990929700000318
is the output of the second bidirectional GRU layer,
Figure RE-GDA00032990929700000319
indicating the output of the GRU in forward direction therein
Figure RE-GDA00032990929700000320
And backward GRU output
Figure RE-GDA00032990929700000321
Carrying out merging connection;
Figure RE-GDA00032990929700000322
is to multiply the second bidirectional GRU layer by a weight vector
Figure RE-GDA00032990929700000323
And adding the offset vector
Figure RE-GDA00032990929700000324
The result of (1);
by analogy, the expression is obtained
Figure RE-GDA00032990929700000325
By a convolution operation with Scale2, for [ C2 t,C3 t]Extracting to reserve the more important scale information of the target and obtain output
Figure RE-GDA00032990929700000326
Then is fully connectedOperation to obtain output Ot；
Wherein the content of the first and second substances,
Figure RE-GDA00032990929700000327
and
Figure RE-GDA00032990929700000328
all are obtained by learning and training.
Specifically, the bidirectional GRU layer is a bidirectional structure formed by connecting two GRU models of a forward GRU and a backward GRU, and the first layer of the forward GRU is forgotten to be output by a gate: f. of1 t＝σ(W1 f[H1 t-1,x”t]+B1 f)，σ(x)＝1/(1+e-x) In the forward GRU update gate, the first output is: z is a radical of1 t＝σ(W1 z[H1 t-1,x”t]+B1 z) And the second output is:
Figure RE-GDA0003299092970000041
similarly, the corresponding first output to the GRU is: z is a radical of2 t＝σ(W2 z[H2 t-1,x”t]+B2 z) And a second output:
Figure RE-GDA0003299092970000042
intermediate output of forward GRU
Figure RE-GDA0003299092970000043
And backward GRU intermediate output
Figure RE-GDA0003299092970000044
Obtaining an output by an aggregation operation on the intermediate output
Figure RE-GDA0003299092970000045
Figure RE-GDA0003299092970000046
Show thatForward GRU output
Figure RE-GDA0003299092970000047
And backward GRU output
Figure RE-GDA0003299092970000048
Performing merged connection as output of bidirectional GRU layer
Figure RE-GDA0003299092970000049
x”tFor the input of the bidirectional GRU layer, [ W ]1 f,B1 f]， [W1 Z,B1 Z]，[W1 h,B1 h]For forward GRU model parameters, [ W ]2 f,B2 f],[W2 Z,B2 Z]，[W2 h,B2 h]For the inverse GRU model parameters, [ W ]12 o,B12 o]Are output layer parameters.
Figure RE-GDA00032990929700000410
The output of the first bidirectional GRU layer, correspondingly,
Figure RE-GDA00032990929700000411
is the output of the second bidirectional GRU layer.
Preferably, the convolutional layer is a 1-dimensional convolutional network.
Specifically, the influencing factor data includes: the temperature and humidity of each room in the building, as well as the outside air pressure, outside humidity and outside wind speed.
The building energy consumption prediction model in the method mainly comprises a multi-scale convolution layer, a bidirectional GRU layer and a full-connection layer, the multi-scale convolution layer is introduced into a recurrent neural network, a plurality of improved attention units are connected in series, attention mechanisms are distributed on different scales, therefore, the model can collect historical information from different scales, distinguish the influence of different input elements on a prediction result, the bidirectional GRU layer can more fully acquire the context information of sequence data on the basis of GRUs, the whole model adopts a convolution structure to fuse the recognition output of attention mechanisms of different scales, and the output is screened and identified by different scales through convolution connection, so that the scale information which is more important to the target can be reserved, and the better precision is obtained for predicting the energy consumption value of the building.
Drawings
FIG. 1 is a schematic flow diagram of the process of the present invention;
FIG. 2 is a data processing flow diagram of the method of the present invention;
FIG. 3 is a schematic diagram of a 1-dimensional convolutional network in the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
As shown in fig. 1, the method for predicting the energy consumption of a building based on the multi-scale convolution cyclic neural network is provided, and includes the following steps:
step 1, constructing a building energy consumption prediction model based on a multi-scale convolution cyclic neural network;
step 2, training the building energy consumption prediction model by using training set data, wherein the training set data comprises influence factor data and known building energy consumption data;
and 3, inputting the test set data into the trained building energy consumption prediction model, and calculating to obtain a predicted value of the building energy consumption.
Specifically, the building energy consumption prediction model comprises a first convolution layer, a first bidirectional GRU layer, a first multi-scale convolution layer, a second bidirectional GRU layer, a second multi-scale convolution layer, a first full-connection layer and a second full-connection layer, wherein the layers are sequentially connected in sequence, the output of the first convolution layer and the output of the first bidirectional GRU layer are connected and then simultaneously used as the input of the first multi-scale convolution layer and the input of the second multi-scale convolution layer, the bidirectional GRU layer is formed by connecting a forward GRU model and a backward GRU model in parallel to form a bidirectional structure, the bidirectional GRU layer outputs two combined GRU signals, the output layer of the first full-connection layer is 100, the output layer of the second full-connection layer is 1, the input of the first convolution layer in the building energy consumption prediction model is an influence factor data sequence, and the output of the full-connection layer is a building energy consumption value.
Specifically, the energy consumption prediction model for the building is
Figure RE-GDA0003299092970000061
x0,...,xTTo influence the factor sequence data, (y)0,...,yK),K<T is a known building energy consumption value, (y)K+1,...,yT) To the extent that a predicted building energy consumption value is required,
Figure RE-GDA0003299092970000062
for the corresponding estimated value, the input is x0,...,xT,y0,...,yKAnd (3) sequentially inputting variables into the building energy consumption prediction model to start training, wherein the loss function adopts standard normalized MSE, and the activation function adopts Relu function.
Specifically, the analytical expression of the building energy consumption prediction model is as follows:
Figure RE-GDA0003299092970000063
Figure RE-GDA0003299092970000064
Figure RE-GDA0003299092970000065
Figure RE-GDA0003299092970000066
Figure RE-GDA0003299092970000067
Figure RE-GDA0003299092970000068
C4 t＝η2([C2 t,C3 t])
Figure RE-GDA0003299092970000069
wherein x istIs an input to the model at time t, η1(. eta.) and eta2(. is) two convolution operations, [, ]]For merge join operations, mutiscalaconv (·, Scale1) and mutiscalaconv (·, Scale2) are two multi-Scale convolution operations with scales Scale1 and Scale2, respectively, and the specific process of fusion convolution is as follows:
first convolution layer η1(xt) Accepting sequence data xtInput and output of
Figure RE-GDA00032990929700000610
Simultaneously as inputs to the first multi-scale convolutional layer and the second multi-scale convolutional layer;
Figure RE-GDA00032990929700000611
is the output of the first bi-directional GRU layer,
Figure RE-GDA00032990929700000612
indicating the output to be GRU in forward direction
Figure RE-GDA00032990929700000613
And backward GRU output
Figure RE-GDA00032990929700000614
Carrying out merging connection;
Figure RE-GDA00032990929700000615
is to multiply the first bidirectional GRU layer by a weight vector
Figure RE-GDA00032990929700000616
And adding the offset vector
Figure RE-GDA00032990929700000617
The result of (1);
will be provided with
Figure RE-GDA0003299092970000071
And η1(xt) Output of (2)
Figure RE-GDA0003299092970000072
Are combined into
Figure RE-GDA0003299092970000073
As an input to the first multi-scale convolutional layer;
Figure RE-GDA0003299092970000074
is the output of the first multi-Scale convolutional layer with Scale1, connected to the second bidirectional GRU layer;
Figure RE-GDA0003299092970000075
is the output of the second bidirectional GRU layer,
Figure RE-GDA0003299092970000076
indicating the output of the GRU in forward direction therein
Figure RE-GDA0003299092970000077
And backward GRU output
Figure RE-GDA0003299092970000078
Carrying out merging connection;
Figure RE-GDA0003299092970000079
is to multiply the second bidirectional GRU layer by a weight vector
Figure RE-GDA00032990929700000710
And adding the offset vector
Figure RE-GDA00032990929700000711
The result of (1);
by analogy, the expression is obtained
Figure RE-GDA00032990929700000712
By a convolution operation with Scale2, for [ C2 t,C3 t]Extracting to reserve the more important scale information of the target and obtain output
Figure RE-GDA00032990929700000713
Then obtaining output O through full connection operationt(ii) a The data processing flow of the method of the invention is shown in FIG. 2;
wherein the content of the first and second substances,
Figure RE-GDA00032990929700000714
and
Figure RE-GDA00032990929700000715
all are obtained by learning and training.
Specifically, the bidirectional GRU layer is a bidirectional structure formed by connecting two GRU models of a forward GRU and a backward GRU, and the first layer of the forward GRU is forgotten to be output by a gate: f. of1 t＝σ(W1 f[H1 t-1,x”t]+B1 f)，σ(x)＝1/(1+e-x) In the forward GRU update gate, the first output is: z is a radical of1 t＝σ(W1 z[H1 t-1,x”t]+B1 z) And the second output is:
Figure RE-GDA00032990929700000716
similarly, the corresponding first output to the GRU is: z is a radical of2 t＝σ(W2 z[H2 t-1,x”t]+B2 z) And a second output:
Figure RE-GDA00032990929700000717
intermediate output of forward GRU
Figure RE-GDA00032990929700000718
And backward GRU intermediate output
Figure RE-GDA00032990929700000719
Obtaining an output by an aggregation operation on the intermediate output
Figure RE-GDA00032990929700000720
Figure RE-GDA00032990929700000721
Indicating the output to be GRU in forward direction
Figure RE-GDA00032990929700000722
And backward GRU output
Figure RE-GDA00032990929700000723
Performing merged connection as output of bidirectional GRU layer
Figure RE-GDA00032990929700000724
x”tFor the input of the bidirectional GRU layer, [ W ]1 f,B1 f]， [W1 Z,B1 Z]，[W1 h,B1 h]For forward GRU model parameters, [ W ]2 f,B2 f],[W2 Z,B2 Z]，[W2 h,B2 h]For the inverse GRU model parameters, [ W ]12 o,B12 o]Are output layer parameters.
Figure RE-GDA00032990929700000725
The output of the first bidirectional GRU layer, correspondingly,
Figure RE-GDA00032990929700000726
is the output of the second bidirectional GRU layer.
Preferably, the convolutional layer is a 1-dimensional convolutional network.
Specifically, the influencing factor data includes: the temperature and humidity of each room in the building, as well as the outside air pressure, outside humidity and outside wind speed.
Preferably, the convolutional layer is a 1-dimensional convolutional network. Convolutional neural networks generally include 1-dimensional convolution, 2-dimensional convolution, and 3-dimensional convolution networks. The one-dimensional convolution network is mainly used for sequence data such as audio data, equipment maintenance sampling data and the like, the two-dimensional convolution is mainly used for image processing such as image classification, target recognition, image segmentation and the like, and the three-dimensional convolution network is mainly used for video processing such as medical image video, motion detection and the like. In this embodiment, the time series data is mainly analyzed, and a 1-dimensional convolution network result is adopted. A typical 1-dimensional convolutional network results are shown in fig. 3. This includes a one-dimensional convolution kernel vector with a filter size k of 4. Convolution factors (convolution factors) d is 1.
For the element s that currently needs to be convolved, the mathematical expression of the one-dimensional convolution operation is:
Figure RE-GDA0003299092970000081
wherein f (i) represents a convolution kernel, Xs-d·iIndicating that sample values at interval d are taken forward.
To better show the details of this example, the experiment used an energy consumption data set for a building house in belgium. The data description is shown in table 1.
TABLE 1 data set data item meanings
Figure RE-GDA0003299092970000082
Figure RE-GDA0003299092970000091
The experimental parameters of the model in this example are shown in table 2:
table 2: algorithm variable parameter valuing
Figure RE-GDA0003299092970000092
The experimental environment is as follows: the experimental background used in this example is: computer experiment environment: the experimental background used in this example is: the computer is mainly configured as follows: pentium (R) Dual-core 3.06CPU, 8G RAM memory.
And (3) effect evaluation: parameters employed herein for performance evaluation of algorithms include RMSE, MAE, MAPE, and CC:
RMSE (Root Mean Square Error) is calculated as:
Figure RE-GDA0003299092970000093
MAE (Mean absolute Error) is calculated as:
Figure RE-GDA0003299092970000094
MAPE (Mean absolute percent Error) was calculated as:
Figure RE-GDA0003299092970000095
r2 (coeffient of Determination), determining the coefficient by the following calculation method:
Figure RE-GDA0003299092970000101
it should be noted that RMSE, MAE and MAPE are measures of prediction errors, and the smaller the value is, the more accurate the value is, and the R2 parameter represents the coefficient for determining the number of two sequences, and the larger the value is, the more relevant the two sequence data is, the better the prediction effect is.
For the building energy consumption data set, the periodicity of each sequence data change is not strong, which indicates that the energy consumption problem does not show periodic changes along with days. Seasonally, during the data acquisition period of 4.5 months, as the weather becomes gradually hot (the data of T1-T9 are in an ascending trend), the air humidity gradually decreases, and the overall amplitude of energy consumption is reduced.
Experiments are performed on the data to predict the future energy consumption of the building, and the prediction accuracy is shown in table 3.
Table 3: predicted result values of different algorithms
Figure RE-GDA0003299092970000102
As can be seen from table 3, the prediction accuracy of the method of the present invention is generally higher than that of the conventional machine learning model in the prediction calculation for building energy consumption. While the MCRNN model reduces the RMSE by 47.83%, 38.72%, 16.62%, 15.67%, 13.29%, 13.58%, 7.55%, 3.09% relative to the SVM, Random Forest, LSTM, GRU, Bi-LSTM, Bi-GRU, Bi-Conv-LSTM, Bi-Conv-GRU network models, from the RMSE index. The average accuracy is respectively improved by 37.81%, 70.38%, 32.50%, 16.09%, 25.59%, 27.43%, 4.93% and 4.39%, the prediction correlation is respectively improved by 83.09%, 69.38%, 8.12%, 11.47%, 8.80%, 8.73%, 5.22% and 2.08%, and the method of the invention has better performance than other network models in terms of average percentage error.
According to the invention content and the embodiment, the building energy consumption prediction model in the method mainly comprises a multi-scale convolutional layer, a bidirectional GRU layer and a full-connection layer, the multi-scale convolutional layer is introduced into a recurrent neural network, a plurality of improved attention units are connected in a series connection mode and a jump connection mode, attention mechanisms are distributed on different scales, therefore, the model can collect historical information from different scales, distinguish the influence of different input elements on a prediction result, the bidirectional GRU layer can more fully acquire the context information of sequence data on the basis of GRUs, the whole model adopts a convolution structure to fuse the recognition output of attention mechanisms of different scales, and the output is screened and identified by different scales through convolution connection, so that the scale information which is more important to the target can be reserved, and the better precision is obtained for predicting the energy consumption value of the building.

Claims (6)

1. The building energy consumption prediction method based on the multi-scale convolution cyclic neural network is characterized by comprising the following steps of:
step 1, constructing a building energy consumption prediction model based on a multi-scale convolution cyclic neural network;
step 2, training the building energy consumption prediction model by using training set data, wherein the training set data comprises influence factor data and known building energy consumption data;
step 3, inputting the test set data into the trained building energy consumption prediction model, and calculating to obtain a prediction value of the building energy consumption;
the building energy consumption prediction model comprises a first convolution layer, a first bidirectional GRU layer, a first multi-scale convolution layer, a second bidirectional GRU layer, a second multi-scale convolution layer, a first full-connection layer and a second full-connection layer, wherein the layers are sequentially connected, the output of the first convolution layer and the output of the first bidirectional GRU layer are connected and then simultaneously used as the input of the first multi-scale convolution layer and the second multi-scale convolution layer, the bidirectional GRU layer is formed by connecting a forward GRU model and a backward GRU model in parallel to form a bidirectional structure, the bidirectional GRU layer outputs two combined GRU signals, the output layer of the first full-connection layer is 100, the output layer of the second full-connection layer is 1, the input of the first convolution in the building energy consumption prediction model is an influence factor data sequence, and the output of the second full-connection layer is a building energy consumption value.
2. The building energy consumption prediction method based on the multi-scale convolution cyclic neural network as claimed in claim 1, wherein the building energy consumption prediction model is
Figure RE-FDA0003299092960000011
x0,...,xTTo influence the factor sequence data, (y)0,...,yK),K<T is a known building energy consumption value, (y)K+1,...,yT) To the extent that a predicted building energy consumption value is required,
Figure RE-FDA0003299092960000012
for the corresponding estimated value, the input is x0,...,xT,y0,...,yKAnd (3) sequentially inputting variables into the building energy consumption prediction model to start training, wherein the loss function adopts standard normalized MSE, and the activation function adopts Relu function.
3. The building energy consumption prediction method based on the multi-scale convolution cyclic neural network as claimed in claim 1, wherein the analytical expression of the building energy consumption prediction model is as follows:
Figure RE-FDA0003299092960000021
Figure RE-FDA0003299092960000022
Figure RE-FDA0003299092960000023
Figure RE-FDA0003299092960000024
Figure RE-FDA0003299092960000025
Figure RE-FDA0003299092960000026
Figure RE-FDA0003299092960000027
Figure RE-FDA0003299092960000028
wherein x istIs an input to the model at time t, η1(. eta.) and eta2(. is) two convolution operations, [, ]]For merge join operations, mutiscalaconv (·, Scale1) and mutiscalaconv (·, Scale2) are two multi-Scale convolution operations with scales Scale1 and Scale2, respectively, and the specific process of fusion convolution is as follows:
first convolution layer η1(xt) Accepting sequence data xtInput and output of
Figure RE-FDA0003299092960000029
Simultaneously as inputs to the first multi-scale convolutional layer and the second multi-scale convolutional layer;
Figure RE-FDA00032990929600000210
is the output of the first bi-directional GRU layer,
Figure RE-FDA00032990929600000211
indicating the output to be GRU in forward direction
Figure RE-FDA00032990929600000212
And backward GRU output
Figure RE-FDA00032990929600000213
Carrying out merging connection;
Figure RE-FDA00032990929600000214
is to multiply the first bidirectional GRU layer by a weight vector
Figure RE-FDA00032990929600000215
And adding the offset vector
Figure RE-FDA00032990929600000216
The result of (1);
will be provided with
Figure RE-FDA00032990929600000217
And η1(xt) Output of (2)
Figure RE-FDA00032990929600000218
Are combined into Pt 1As input to the first multi-scale convolutional layer;
Figure RE-FDA00032990929600000219
is the output of the first multi-Scale convolutional layer with Scale1, connected to the second bidirectional GRU layer;
Figure RE-FDA00032990929600000220
is the output of the second bidirectional GRU layer,
Figure RE-FDA00032990929600000221
indicating the output of the GRU in forward direction therein
Figure RE-FDA00032990929600000222
And backward GRU output
Figure RE-FDA00032990929600000223
Carrying out merging connection;
Figure RE-FDA00032990929600000224
is to multiply the second bidirectional GRU layer by a weight vector
Figure RE-FDA00032990929600000225
And adding the offset vector
Figure RE-FDA00032990929600000226
The result of (1);
by analogy, the expression is obtained
Figure RE-FDA00032990929600000227
By a convolution operation with Scale2, for [ C2 t,C3 t]Extracting to reserve the more important scale information of the target and obtain output
Figure RE-FDA0003299092960000031
Then obtaining output O through full connection operationt；
Wherein the content of the first and second substances,
Figure RE-FDA0003299092960000032
and
Figure RE-FDA0003299092960000033
all are obtained by learning and training.
4. The building energy consumption prediction method based on the multi-scale convolution cyclic neural network as claimed in claim 1 or 3, characterized in that the bidirectional GRU layer is a bidirectional structure formed by connecting two GRU models, namely a forward GRU and a backward GRU, and the first layer of the forward GRU is left to be output through a gate: f. of1 t＝σ(W1 f[H1 t-1,x”t]+B1 f)，σ(x)＝1/(1+e-x) In the forward GRU update gate, the first output is: z is a radical of1 t＝σ(W1 z[H1 t-1,x”t]+B1 z) And the second output is:
Figure RE-FDA0003299092960000034
similarly, the corresponding first output to the GRU is: z is a radical of2 t＝σ(W2 z[H2 t-1,x”t]+B2 z) And a second output:
Figure RE-FDA0003299092960000035
intermediate output of forward GRU
Figure RE-FDA0003299092960000036
And backward GRU intermediate output
Figure RE-FDA0003299092960000037
Obtaining an output by an aggregation operation on the intermediate output
Figure RE-FDA0003299092960000038
Figure RE-FDA0003299092960000039
Indicating the output to be GRU in forward direction
Figure RE-FDA00032990929600000310
And backward GRU output
Figure RE-FDA00032990929600000311
Performing merged connection as output of bidirectional GRU layer
Figure RE-FDA00032990929600000312
x”tFor the input of the bidirectional GRU layer, [ W ]1 f,B1 f]，[W1 Z,B1 Z]，[W1 h,B1 h]For forward GRU model parameters, [ W ]2 f,B2 f],[W2 Z,B2 Z]，[W2 h,B2 h]For the inverse GRU model parameters, [ W ]12 o,B12 o]Are output layer parameters.
5. The method of claim 4, wherein the convolutional layer is a 1-dimensional convolutional network.
6. The method of claim 1, wherein the influence factor data comprises: the temperature and humidity of each room in the building, as well as the outside air pressure, outside humidity and outside wind speed.
CN202111020470.0A 2021-09-01 2021-09-01 Building energy consumption prediction method based on multi-scale convolution cyclic neural network Pending CN114065335A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111020470.0A CN114065335A (en) 2021-09-01 2021-09-01 Building energy consumption prediction method based on multi-scale convolution cyclic neural network

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111020470.0A CN114065335A (en) 2021-09-01 2021-09-01 Building energy consumption prediction method based on multi-scale convolution cyclic neural network

Publications (1)

Publication Number Publication Date
CN114065335A true CN114065335A (en) 2022-02-18

Family

ID=80233595

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111020470.0A Pending CN114065335A (en) 2021-09-01 2021-09-01 Building energy consumption prediction method based on multi-scale convolution cyclic neural network

Country Status (1)

Country Link
CN (1) CN114065335A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117011092A (en) * 2023-09-28 2023-11-07 武昌理工学院 Intelligent building equipment management monitoring system and method

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117011092A (en) * 2023-09-28 2023-11-07 武昌理工学院 Intelligent building equipment management monitoring system and method
CN117011092B (en) * 2023-09-28 2023-12-19 武昌理工学院 Intelligent building equipment management monitoring system and method

Similar Documents

Publication Publication Date Title
Xu et al. Predicting pipeline leakage in petrochemical system through GAN and LSTM
CN111178611B (en) Method for predicting daily electric quantity
Hu et al. Research and application of a hybrid model based on Meta learning strategy for wind power deterministic and probabilistic forecasting
CN110020712B (en) Optimized particle swarm BP network prediction method and system based on clustering
CN114282443B (en) Residual service life prediction method based on MLP-LSTM supervised joint model
CN114358192B (en) Multi-source heterogeneous landslide data monitoring and fusing method
Shang et al. A hybrid method for traffic incident detection using random forest-recursive feature elimination and long short-term memory network with Bayesian optimization algorithm
CN112396234A (en) User side load probability prediction method based on time domain convolutional neural network
CN114015825B (en) Method for monitoring abnormal state of blast furnace heat load based on attention mechanism
CN116316599A (en) Intelligent electricity load prediction method
CN111415752A (en) Hand-foot-and-mouth disease prediction method integrating meteorological factors and search indexes
CN113642254A (en) Wind power generation power prediction method based on multi-scale attention mechanism
CN114548592A (en) Non-stationary time series data prediction method based on CEMD and LSTM
CN115271186B (en) Reservoir water level prediction and early warning method based on delay factor and PSO RNN Attention model
CN113642255A (en) Photovoltaic power generation power prediction method based on multi-scale convolution cyclic neural network
CN112613542A (en) Bidirectional LSTM-based enterprise decontamination equipment load identification method
CN113858566B (en) Injection molding machine energy consumption prediction method and system based on machine learning
CN115759461A (en) Internet of things-oriented multivariate time sequence prediction method and system
CN111582588B (en) Building energy consumption prediction method based on triple convolution fusion GRU
CN114065335A (en) Building energy consumption prediction method based on multi-scale convolution cyclic neural network
CN114357670A (en) Power distribution network power consumption data abnormity early warning method based on BLS and self-encoder
CN114240687A (en) Energy hosting efficiency analysis method suitable for comprehensive energy system
CN111242369A (en) PM2.5 data prediction method based on multiple fusion convolution GRU
CN115062764B (en) Intelligent illuminance adjustment and environmental parameter Internet of things big data system
CN114970745B (en) Intelligent security and environment big data system of Internet of things

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