CN116415713A - Building energy consumption prediction method based on E+ and artificial intelligence - Google Patents
Building energy consumption prediction method based on E+ and artificial intelligence Download PDFInfo
- Publication number
- CN116415713A CN116415713A CN202310173796.XA CN202310173796A CN116415713A CN 116415713 A CN116415713 A CN 116415713A CN 202310173796 A CN202310173796 A CN 202310173796A CN 116415713 A CN116415713 A CN 116415713A
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- data
- building
- building energy
- lstm
- 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
- 238000005265 energy consumption Methods 0.000 title claims abstract description 120
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 18
- 238000004088 simulation Methods 0.000 claims abstract description 36
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 22
- 238000013461 design Methods 0.000 claims abstract description 6
- 238000004134 energy conservation Methods 0.000 claims abstract description 3
- 238000012549 training Methods 0.000 claims description 47
- 230000006870 function Effects 0.000 claims description 39
- 238000012360 testing method Methods 0.000 claims description 14
- 238000010606 normalization Methods 0.000 claims description 11
- 238000012545 processing Methods 0.000 claims description 9
- 238000004378 air conditioning Methods 0.000 claims description 8
- 210000002569 neuron Anatomy 0.000 claims description 7
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 claims description 6
- 230000004913 activation Effects 0.000 claims description 6
- 230000000694 effects Effects 0.000 claims description 6
- 239000011159 matrix material Substances 0.000 claims description 6
- 230000001131 transforming effect Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000013332 literature search Methods 0.000 claims description 3
- 238000012821 model calculation Methods 0.000 claims description 3
- 238000005457 optimization Methods 0.000 claims description 3
- 238000007781 pre-processing Methods 0.000 claims description 3
- 238000011160 research Methods 0.000 claims description 3
- 238000010438 heat treatment Methods 0.000 description 9
- 210000004027 cell Anatomy 0.000 description 3
- 238000012407 engineering method Methods 0.000 description 3
- 238000005057 refrigeration Methods 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 230000002779 inactivation Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000007787 long-term memory Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000001105 regulatory effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/29—Geographical information databases
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/906—Clustering; Classification
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/067—Enterprise or organisation modelling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Databases & Information Systems (AREA)
- Tourism & Hospitality (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Marketing (AREA)
- General Business, Economics & Management (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Educational Administration (AREA)
- Operations Research (AREA)
- General Health & Medical Sciences (AREA)
- Game Theory and Decision Science (AREA)
- Health & Medical Sciences (AREA)
- Quality & Reliability (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Primary Health Care (AREA)
- Remote Sensing (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a building energy consumption prediction method based on E+ and artificial intelligence, which comprises the steps of firstly comprehensively arranging influence factors of national building energy conservation design standard, building energy consumption field documents and related enterprises and public institutions of building industry on building energy consumption, adopting energy plus building energy consumption simulation software to establish a typical building energy consumption model, and establishing an artificial intelligent building energy consumption database based on energy plus for different influence factors in a cross simulation mode; and then, an LSTM prediction algorithm is adopted, and an artificial intelligent building energy consumption database is utilized to establish an energy consumption prediction model based on the LSTM algorithm. When the subsequent energy consumption is predicted, the corresponding energy consumption prediction data of the building can be obtained only by inputting the building energy consumption simulation parameters into an energy consumption prediction model of the LSTM algorithm, so that complicated operation during each simulation of other methods is avoided, and the building energy consumption simulation is simpler and more accurate.
Description
Technical Field
The invention relates to the technical field of building energy consumption prediction, in particular to a building energy consumption prediction method based on E+ and artificial intelligence.
Background
The energy consumption prediction of the building mainly comprises three steps, namely an engineering method, statistical regression and artificial intelligence, wherein the engineering method adopts a thermophysical equation, relevant information such as the environment, operation and air conditioning equipment of the building is input, and an accurate energy consumption simulation result is obtained through a series of complex simulation calculations. However, the energy consumption simulation software such as energy plus has the characteristic of complexity and complexity in use, which is difficult for some engineering personnel. The statistical regression method has the advantages of simple structure, relatively easy establishment of a model, complex interactivity of input elements, low calculation efficiency and insufficient prediction precision.
Disclosure of Invention
In order to solve the defects of the prior art, the invention provides the building energy consumption prediction method based on E+ and artificial intelligence, which takes the problems of complicated construction of engineering method models and complicated artificial intelligence method into consideration, trains an LSTM prediction algorithm by utilizing an artificial intelligence database, and once the LSTM network model is trained, the subsequent building energy consumption prediction becomes very simple, corresponding building energy consumption prediction data can be obtained only by inputting building energy consumption simulation parameters into the trained LSTM network model, thereby avoiding complicated operation of other methods during each simulation, and enabling the building energy consumption simulation to be simpler and more accurate.
The technical scheme for solving the technical problems is as follows: the building energy consumption prediction method based on E+ and artificial intelligence is characterized by comprising the following steps of:
step 1): determining building energy consumption model parameters
According to national building energy-saving design standard, building energy consumption field literature and research results of related enterprises and government units in building industry, the determined building energy consumption simulation parameters include weather files of buildings, building window wall ratio, building enclosure structure parameters, indoor heat interference and an air conditioning system. Weather files are divided into severe cold areas, summer hot and winter warm areas and mild areas according to climate areas of China, indoor heat interference mainly comprises power density of lighting equipment, and an air conditioning system comprises setting cold and hot temperatures.
Step 2): acquiring artificial intelligence simulation databases
Building energy consumption simulation parameters determined in the step 1) are used for building a model by using Sketchup software and Openstudio software, and an IDF file which can be calculated by energy plus software is derived. And then the IDF file is imported into energy plus software for calculation, so that a plurality of groups of building time-to-time energy consumption data sets are obtained.
Step 3): arrangement of databases
The time-by-time energy consumption data sets of the plurality of groups of buildings obtained in the step 2) are arranged, and because the time-by-time data sets are numerous, in order to enable model calculation to be more efficient, each climate zone is simulated by taking one model, and five time-by-time databases which are respectively composed of a plurality of pieces of data of severe cold regions, summer hot winter warm regions and mild regions are obtained. Each piece of data comprises building energy consumption simulation parameters and time-by-time data of energy consumption.
Step 4): data preprocessing
The five climate zone hour-by-hour databases obtained in the step 3) are respectively preprocessed: firstly, a MinMaxScale scaler is used for carrying out normalization processing on a database, all data are scaled to be between [0,1], a series-to-super-visual () function is used for converting the data in a time sequence form into a form of a supervision learning set, and the following steps are carried out according to 7:3, dividing a training set and a testing set of the database, and transforming the data matrixes of the training set and the testing set into a three-dimensional matrix required by an LSTM algorithm by using a reshape () function.
Step 5): establishing LSTM energy consumption prediction network model
Adopting a tf.keras.model.sequential () function component LSTM network model in Tensorflow, setting a units value as 25, and taking input_shape as the first dimension and the second dimension of a training set train_X.shape; regularized debugging is carried out by adopting Dropout; adding a full-connection layer by Dense, setting the number of neurons as 1 and the activation function as tanh; calling a common method to configure a learning flow of an LSTM algorithm, selecting adam as an optimizer, and taking a loss function as an MAE; the batch size and the iteration number of the LSTM are set. The data call function is a fit function.
Step 6): training LSTM energy consumption prediction network model
Respectively taking the data of corresponding parts of building energy consumption simulation parameter data in the training sets of the five climate areas as the input of the LSTM network in the step 5), taking the data corresponding to the energy consumption data parts in the corresponding training sets as the reference value of the output of the LSTM network, calculating a loss function by utilizing the data corresponding to the energy consumption data parts in the training sets and the output of the LSTM network, and optimizing and updating the initial parameters of the LSTM network by utilizing an adam optimizer according to the loss function value to finish one iteration training; and taking the parameter value after the last optimization update as an initial parameter of the next iteration training of the LSTM network, inputting the training set into the LSTM network again for iteration training, and repeating the steps until the iteration training times reach a preset value, and storing the parameter value of the LSTM network when the last iteration training is finished, so as to respectively obtain a trained LSTM network model of the five climate areas. And (3) adjusting the loss function, the optimizer, the number of neurons, the batch size and the iteration number of the LSTM network in the step (5), and respectively training by using a training set by adopting the same method to obtain a plurality of trained LSTM network models of the five climate areas.
And taking the data of the corresponding parts of the building energy consumption simulation parameter data in the test set of the five climate areas as the input of a plurality of trained LSTM network models, taking the data corresponding to the energy consumption data in the corresponding test set as the reference value output by the trained LSTM network models, and respectively evaluating the prediction performance of the models of the five climate areas by taking the decision coefficients as evaluation indexes. The larger the value of the decision coefficient, the better the prediction effect of the trained LSTM network model.
Step 8): building energy consumption prediction
Acquiring time-by-time data of building energy consumption simulation parameters of a building in a period to be predicted of a climate zone to be predicted, firstly, carrying out normalization processing on a database by using a MinMaxScale scaler, scaling all data to be between [0,1], and transforming the data into a three-dimensional matrix required by an LSTM algorithm by using a reshape () function to obtain input data of an LSTM network model; inputting the input data into the LSTM network model of the climate zone with the maximum decision coefficient value in the step 7), and obtaining the output of the LSTM network model; and then carrying out inverse normalization processing on the output of the LSTM network model to obtain building energy consumption data of the to-be-predicted period of the building in the to-be-predicted climate zone.
Compared with the prior art, the invention has the beneficial effects that: the invention relates to a building energy consumption prediction method based on E+ and artificial intelligence, which comprises the steps of firstly comprehensively arranging influence factors of national building energy conservation design standard, building energy consumption field documents and related enterprises and public institutions in building industry on building energy consumption, adopting energy plus building energy consumption simulation software to establish a typical building energy consumption model, carrying out cross simulation on different influence factors, and establishing an artificial intelligent building energy consumption database based on energy plus; and then, an LSTM prediction algorithm is adopted, and an artificial intelligent building energy consumption database is utilized to establish an energy consumption prediction model based on the LSTM algorithm. When the subsequent energy consumption is predicted, the corresponding energy consumption prediction data of the building can be obtained only by inputting the building energy consumption simulation parameters into an energy consumption prediction model of the LSTM algorithm, so that complicated operation during each simulation of other methods is avoided, and the building energy consumption simulation is simpler and more accurate. The building energy consumption prediction model provided by the method can avoid the complexity of an energy plus modeling process during each prediction, the LSTM algorithm can be used for predicting the building energy consumption, the prediction accuracy can be improved, and a space is reserved for the intervention of future building detection data.
Drawings
FIG. 1 is a schematic flow chart of one embodiment of a building energy consumption prediction method based on E+ and artificial intelligence.
Fig. 2 is a schematic diagram of the LSTM algorithm.
Fig. 3 is a schematic diagram of the principle of obtaining building energy consumption data by using building energy consumption simulation parameters.
Fig. 4 is a graph showing the comparison of predicted energy consumption data and actual energy consumption data of a building in a severe cold region and a mild region in a certain time period by using the method for predicting energy consumption of a building based on e+ and artificial intelligence of the present invention, wherein (a) in fig. 4 is a graph showing the comparison of predicted energy consumption data and actual energy consumption data of a building in a severe cold region in a certain time period, and (b) in fig. 4 is a graph showing the comparison of predicted energy consumption data and actual energy consumption data of heating of the building in a severe cold region in a certain time period; fig. 4 (c) is a graph of predicted refrigeration energy consumption versus actual refrigeration energy consumption data of a certain building in a mild region over a certain period, and fig. 4 (d) is a graph of predicted heating energy consumption versus actual heating energy consumption data of the certain building in the mild region over a certain period.
Detailed Description
The technical scheme of the invention is explained in detail below with reference to the attached drawings.
The invention provides a building energy consumption prediction method based on E+ and artificial intelligence, which comprises the following steps:
step 1): determining building energy consumption model parameters
According to national building energy-saving design standard, building energy consumption field literature and research results of related enterprises and government units in building industry, the determined building energy consumption simulation parameters include weather files of buildings, building window wall ratio, building enclosure structure parameters, indoor heat interference and an air conditioning system. Weather files are divided into severe cold areas, summer hot and winter warm areas and mild areas according to climate areas of China, indoor heat interference mainly comprises power density of lighting equipment, and an air conditioning system comprises setting cold and hot temperatures.
Step 2): establishing an artificial intelligence simulation database
Building energy consumption simulation parameters determined in the step 1) are used, sketchup software and Openstudio software are used for building a model, and an IDF file which can be calculated by energy plus software (energy consumption simulation software) is derived. And then the IDF file is imported into energy plus software for calculation, so that a plurality of groups (1200 groups) of building energy consumption (refrigeration and heating) data sets are obtained all the time year by year.
Step 3): arrangement of databases
And (3) arranging a plurality of groups (1200 groups) of building time-by-time energy consumption data sets obtained in the step (2) all year round, wherein because the time-by-time data sets are numerous, in order to make model calculation more efficient, each climate zone takes one model for simulation, and five time-by-time databases consisting of a plurality of pieces (8760×60 pieces) of data of severe cold regions, summer hot winter warm regions and mild regions are obtained. Each piece of data includes building energy consumption simulation parameters (when a certain parameter does not change with time, the data of each moment is unchanged so that the dimension of each piece of data is consistent) and time-by-time data of energy consumption.
Step 4): data preprocessing
The five climate zone hour-by-hour databases obtained in the step 3) are respectively preprocessed: the LSTM algorithm adopts Python language, firstly, a MinMaxScale scaler is used for carrying out normalization processing on a database, all data are scaled to be between 0 and 1, a series_to_supervisual () function is used for converting data in a time sequence form into a form of a supervision learning set, and training set and testing set data are converted into data types which are available for LSTM; according to 7:3, dividing a training set and a testing set of the database, and transforming the data matrixes of the training set and the testing set into a three-dimensional matrix required by an LSTM algorithm by using a reshape () function.
The normalization formula is:
in the formula, X is the original data of a certain parameter; x' is the data normalized value of the parameter; x is X max Is the maximum value in the original data sequence of the parameter; x is X min Is the minimum value in the original data sequence of the parameter.
Step 5): establishing LSTM energy consumption prediction network model
The LSTM unit comprises five interaction layers, three sigmoid layers and two tanh layers, as shown in fig. 2, in the figure, A represents an LSTM unit, sigma represents a sigmoid activation function, the sigma represents a sigmoid activation function and the tanh activation function act as values in a regulating network, the sigmoid function enables the data value range in the network to be 0-1, the tanh function enables the data value in the network to be-1, the arrangement of 0 and 1 of the sigmoid function is beneficial to LSTM algorithm screening information, important information is reserved by 1, and non-important information is removed by 0.
The LSTM algorithm designs control "gates" to direct the information flow, including forget gates, input gates, and output gates. Forgetting door f t Information to be deleted is determined and input to gate i t Determining cell output in a front layer
Which parts should be updated to long-term memory C t Output gate o t Deciding which information in the cell state should be read, then generating the current output h of the LSTM cell t ，h t Is the output of the current neuron. Structure of each "door":
1) Forgetting the door:
f t ＝sigmoid(W f,x ·x t +W f,h ·h t-1 +b f )
2) An input door:
i t ＝sigmoid(W i,x ·x t +W i,h ·h t-1 +b i )
3) Output door:
o t ＝sigmoid(W o,x ·x t +W o,h ·h t-1 +b o )
h t ＝o t ·tanh(C t )
wherein f t 、i t 、o t Respectively representing an input gate, a forget gate and an output gate, W represents a weightMatrix, b represents bias vector, x t The input of the LSTM unit corresponding to the t-th moment is obtained.
Setting units (hidden layer size in each unit) to be 25 by adopting a tf.keras.model.sequential () function component LSTM network model in Tensorflow (a symbolic mathematical system programmed based on data flow), wherein input_shape is the first dimension and the second dimension of a training set train_X.shape; regularized debugging by adopting Dropout (random inactivation); adding a full connection layer by Dense (Dense neural network), setting the number of neurons as 1 and the activation function as tanh; calling a common method to configure a learning flow of an LSTM algorithm, selecting adam as an optimizer, and taking a loss function as MAE (mean absolute error, mean Absolute Error, MAE); the batch size (batch_size) and the number of iterations (epochs) of the LSTM are set. The data call function is a fit function. I.e. directly using the network layer encapsulated by the pytorch, no initialization of the model parameters is required.
Step 6): training LSTM energy consumption prediction network model
Respectively taking the data of corresponding parts of building energy consumption simulation parameter data in the training sets of the five climate areas as the input of the LSTM network in the step 5), taking the data corresponding to the energy consumption data parts in the corresponding training sets as the reference value of the output of the LSTM network, calculating a loss function by utilizing the data corresponding to the energy consumption data parts in the training sets and the output of the LSTM network, and optimizing and updating the initial parameters of the LSTM network by utilizing an adam optimizer according to the loss function value to finish one iteration training; and taking the parameter value after the last optimization update as an initial parameter of the next iteration training of the LSTM network, inputting the training set into the LSTM network again for iteration training, and repeating the steps until the iteration training times reach a preset value, and storing the parameter value of the LSTM network when the last iteration training is finished, so as to respectively obtain a trained LSTM network model of the five climate areas. And (3) adjusting the loss function, the optimizer, the number of neuron cores, the batch size and the iteration number of the LSTM network in the step (5), and respectively training by using a training set by adopting the same method to obtain a plurality of trained LSTM network models of the five climate areas.
Five climate zones are to be treatedData of corresponding parts of building energy consumption simulation parameter data in the test set are used as input of a plurality of trained LSTM network models, data of corresponding parts of energy consumption data in the test set are used as reference values output by the trained LSTM network models, and a decision coefficient (R-Squared, R) 2 ) As evaluation indexes, the predictive performance of the model of the five climate zones was evaluated, respectively. Determining the coefficient R 2 The fitting degree of the model is represented, the value is 0-1, and the closer to 1, the better the prediction effect is. When R is 2 When the index reaches more than 0.85, the LSTM energy consumption prediction network model (namely the LSTM network model) is judged to have a good prediction effect.
Through the test, the heating and refrigerating energy consumption R in the severe cold region is obtained 2 Values 0.938 and 0.902, respectively; heating and refrigerating energy consumption R in cold region 2 Values of 0.922 and 0.881, respectively; heating and refrigerating energy consumption R of summer heat and winter cold region 2 Values are 0.921 and 0.885, respectively; heating and refrigerating energy consumption R of summer heat and winter warm area 2 Values of 0.865 and 0.898, respectively; heating and refrigerating energy consumption R in mild region 2 The values are respectively 0.923 and 0.901, and the prediction effect is good.
Step 8): building energy consumption prediction
Acquiring time-by-time data of building energy consumption simulation parameters (weather files, building window wall ratios, building enclosure parameters, indoor heat disturbances and air conditioning systems) of a certain building of a climate zone to be predicted in a period to be predicted, firstly, carrying out normalization processing on a database by using a MinMaxScaler scaler, scaling all data to be between [0,1], and transforming the data into a three-dimensional matrix required by an LSTM algorithm by using a reshape () function to obtain input data of an LSTM network model; inputting the input data into the LSTM network model of the climate zone with the maximum decision coefficient value in the step 7), and obtaining the output of the LSTM network model; and then carrying out inverse normalization processing on the output of the LSTM network model to obtain building energy consumption data of the to-be-predicted period of the building in the to-be-predicted climate zone.
The building energy consumption prediction based on the E+ and LSTM algorithm is completed, and the prediction effect is good.
While the foregoing description of the embodiments of the present invention has been presented in conjunction with the drawings, it should be understood that it is not intended to limit the scope of the invention, but rather, it is intended to cover all modifications or variations within the scope of the invention as defined by the claims of the present invention.
The invention is applicable to the prior art where it is not described.
Claims (2)
1. A building energy consumption prediction method based on E+ and artificial intelligence is characterized by comprising the following steps:
step 1): determining building energy consumption model parameters
According to national building energy conservation design standard, building energy consumption field literature and research results of related enterprises and government units in building industry, the determined building energy consumption simulation parameters include weather files of buildings, building window wall ratio, building enclosure structure parameters, indoor heat interference and an air conditioning system; weather files are divided into severe cold areas, summer hot and winter warm areas and mild areas according to climate areas of China, indoor heat interference mainly comprises power density of lighting equipment, and an air conditioning system comprises setting cold and hot temperatures;
step 2): establishing an artificial intelligence simulation database
Building a model by using the building energy consumption simulation parameters determined in the step 1) and using Sketchup software and Openstudio software to derive an IDF file which can be calculated by energy plus software; then, the IDF file is imported into energy plus software for calculation, and a plurality of groups of building time-by-time energy consumption data sets are obtained;
step 3): arrangement of databases
The energy consumption data sets of the multiple groups of buildings obtained in the step 2) are arranged year by year, and because the number of the time-by-year data is large, in order to make the model calculation more efficient, each climate zone is simulated by taking one model, so that five time-by-year databases which are respectively composed of multiple pieces of data of severe cold regions, summer hot winter warm regions and mild regions are obtained; each piece of data comprises building energy consumption simulation parameters and time-by-time data of energy consumption;
step 4): data preprocessing
The five climate zone hour-by-hour databases obtained in the step 3) are respectively preprocessed: firstly, a MinMaxScale scaler is used for carrying out normalization processing on a database, all data are scaled to be between [0,1], a series-to-super-visual () function is used for converting the data in a time sequence form into a form of a supervision learning set, and the following steps are carried out according to 7:3, dividing a training set and a testing set of the database, and transforming the data matrixes of the training set and the testing set into a three-dimensional matrix required by an LSTM algorithm by using a reshape () function;
step 5): establishing LSTM energy consumption prediction network model
Adopting a tf.keras.model.sequential () function component LSTM network model in Tensorflow, setting a units value as 25, and taking input_shape as the first dimension and the second dimension of a training set train_X.shape; regularized debugging is carried out by adopting Dropout; adding a full-connection layer by Dense, setting the number of neurons as 1 and the activation function as tanh; calling a common method to configure a learning flow of an LSTM algorithm, selecting adam as an optimizer, and taking a loss function as an MAE; setting the batch size and the iteration times of the LSTM; the data calling function is a fit function;
step 6): training LSTM energy consumption prediction network model
Respectively taking the data of corresponding parts of building energy consumption simulation parameter data in the training sets of the five climate areas as the input of the LSTM network in the step 5), taking the data corresponding to the energy consumption data parts in the corresponding training sets as the reference value of the output of the LSTM network, calculating a loss function by utilizing the data corresponding to the energy consumption data parts in the training sets and the output of the LSTM network, and optimizing and updating the initial parameters of the LSTM network by utilizing an adam optimizer according to the loss function value to finish one iteration training; the parameter value after the last optimization and updating is used as the initial parameter of the next iteration training of the LSTM network, the training set is input into the LSTM network again for iteration training, the iteration training is repeated continuously until the iteration training times reach a preset value, the parameter value of the LSTM network when the last iteration training is completed is saved, and a trained LSTM network model of five climatic regions is obtained respectively; adjusting the loss function, the optimizer, the number of neurons, the batch size and the iteration number of the LSTM network in the step 5), and respectively training by using a training set by adopting the same method to obtain a plurality of trained LSTM network models of five climate areas;
taking the data of the corresponding parts of the building energy consumption simulation parameter data in the test set of the five climate areas as the input of a plurality of trained LSTM network models, taking the data corresponding to the energy consumption data in the corresponding test set as the reference value output by the trained LSTM network models, and respectively evaluating the prediction performance of the models of the five climate areas by taking the decision coefficients as evaluation indexes; the larger the value of the decision coefficient is, the better the prediction effect of the trained LSTM network model is;
step 8): building energy consumption prediction
Acquiring time-by-time data of building energy consumption simulation parameters of a building in a period to be predicted of a climate zone to be predicted, firstly, carrying out normalization processing on a database by using a MinMaxScale scaler, scaling all data to be between [0,1], and transforming the data into a three-dimensional matrix required by an LSTM algorithm by using a reshape () function to obtain input data of an LSTM network model; inputting the input data into the LSTM network model of the climate zone with the maximum decision coefficient value in the step 7), and obtaining the output of the LSTM network model; and then carrying out inverse normalization processing on the output of the LSTM network model to obtain building energy consumption data of the to-be-predicted period of the building in the to-be-predicted climate zone.
2. The building energy consumption prediction method based on E+ and artificial intelligence according to claim 1, wherein in the step 4), the normalization formula is:
in the formula, X is the original data of a certain parameter; x' is the data normalized value of the parameter; x is X max Is the maximum value in the original data sequence of the parameter; x is X min Is the minimum value in the original data sequence of the parameter.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310173796.XA CN116415713A (en) | 2023-02-28 | 2023-02-28 | Building energy consumption prediction method based on E+ and artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310173796.XA CN116415713A (en) | 2023-02-28 | 2023-02-28 | Building energy consumption prediction method based on E+ and artificial intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116415713A true CN116415713A (en) | 2023-07-11 |
Family
ID=87048825
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310173796.XA Pending CN116415713A (en) | 2023-02-28 | 2023-02-28 | Building energy consumption prediction method based on E+ and artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116415713A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117146382A (en) * | 2023-10-31 | 2023-12-01 | 西华大学 | Intelligent adaptive system optimization method |
-
2023
- 2023-02-28 CN CN202310173796.XA patent/CN116415713A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117146382A (en) * | 2023-10-31 | 2023-12-01 | 西华大学 | Intelligent adaptive system optimization method |
| CN117146382B (en) * | 2023-10-31 | 2024-01-19 | 西华大学 | Intelligent adaptive system optimization method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhou et al. | Using long short-term memory networks to predict energy consumption of air-conditioning systems | |
| CN111795484B (en) | Intelligent air conditioner control method and system | |
| May-Ostendorp et al. | Model-predictive control of mixed-mode buildings with rule extraction | |
| Yan et al. | A simplified prediction model for energy use of air conditioner in residential buildings based on monitoring data from the cloud platform | |
| Chen et al. | Modeling and optimization of complex building energy systems with deep neural networks | |
| Liu et al. | Identifying the most significant input parameters for predicting district heating load using an association rule algorithm | |
| CN114119273B (en) | Non-invasive load decomposition method and system for park comprehensive energy system | |
| Ngarambe et al. | A review on the current usage of machine learning tools for daylighting design and control | |
| CN111027772A (en) | Multi-factor short-term load prediction method based on PCA-DBILSTM | |
| CN104239982A (en) | Method for predicting energy consumption of buildings during holidays and festivals on basis of time series and neural networks | |
| CN111461466B (en) | Heating valve adjusting method, system and equipment based on LSTM time sequence | |
| CN114811713B (en) | Two-level network inter-user balanced heat supply regulation and control method based on mixed deep learning | |
| CN111561734A (en) | Heating household valve adjusting method, system and equipment based on artificial intelligence | |
| CN114370698A (en) | Indoor thermal environment learning efficiency improvement optimization control method based on reinforcement learning | |
| CN111754037A (en) | Long-term load hybrid prediction method for regional terminal integrated energy supply system | |
| CN112651543A (en) | Daily electric quantity prediction method based on VMD decomposition and LSTM network | |
| CN116415713A (en) | Building energy consumption prediction method based on E+ and artificial intelligence | |
| CN116989442A (en) | Central air conditioner load prediction method and system | |
| CN113962454A (en) | LSTM energy consumption prediction method based on dual feature selection and particle swarm optimization | |
| CN114169639A (en) | Enterprise power consumption prediction method | |
| CN114662201A (en) | Optimizing method for intelligent regulation and control of natural ventilation | |
| CN115640901A (en) | Small sample load prediction method based on hybrid neural network and generation countermeasure | |
| CN111709585A (en) | Air conditioner load prediction method and device and storage medium | |
| CN115879190B (en) | Model construction method and device and building load prediction method and device | |
| Mobarakeh et al. | A novel methodology based on artificial intelligence to achieve the formost Buildings’ heating system |
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 |