CN110046743A - Energy Consumption of Public Buildings prediction technique and system based on GA-ANN - Google Patents
Energy Consumption of Public Buildings prediction technique and system based on GA-ANN Download PDFInfo
- Publication number
- CN110046743A CN110046743A CN201910168405.9A CN201910168405A CN110046743A CN 110046743 A CN110046743 A CN 110046743A CN 201910168405 A CN201910168405 A CN 201910168405A CN 110046743 A CN110046743 A CN 110046743A
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- data
- public buildings
- prediction
- error
- 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.)
- Granted
Links
Classifications
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- 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
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/08—Construction
Abstract
The present invention provides a kind of Energy Consumption of Public Buildings prediction technique and system based on GA-ANN, acquisition public building by when energy consumption and its influence factor data, arrange data simultaneously pre-processed；Training set and test set are divided, and input variable is screened by correlation coefficient process；Input test data use training set data training pattern by the relevant parameter of genetic algorithm optimization artificial nerve network model later；By inputting the input variable in the period that is predicted, Energy Consumption of Public Buildings is predicted；Finally by error criterion evaluation to the prediction effect of test set, range of allowable error is provided.The present invention gives the processes and method predicted using genetic algorithm and artificial neural network Energy Consumption of Public Buildings.For public building, high-precision prediction is realized, provides foundation for the monitoring of Energy Consumption of Public Buildings, management and diagnosis.
Description
Technical field
The present invention relates to building energy intellectualized technology fields, and in particular, to a kind of public building based on GA-ANN
Energy consumption prediction technique and system, the method that Energy Consumption of Public Buildings is predicted especially with the method for machine learning.
Background technique
With the development of society, building energy consumption constantly increases, public building is as energy consumption rich and influential family its energy management by society
The attention of meeting.And the prediction of Energy Consumption of Public Buildings is the important component during public building management and building energy conservation, it is right
Improve building energy utilization rate, protection environment is of great significance.It is directed to the energy consumption monitoring system sum number of public building in recent years
It is continued to develop according to platform, related data is acquired by sensor and internet, has been obtained for the data basis of prediction in recent years.
Based on historical data, prediction and control are driven by data method, provide data supporting for the scientific management of building energy.
Traditional building energy consumption prediction analysis method is usually Method of Physical Modeling, it has modeling and calculating, and time-consuming,
Model complexity is high, using it is cumbersome the disadvantages of.And the building energy consumption model based on data-driven method, do not need physical parameter and
Thermodynamic balance equations, the analysis that can only rely on passing historical data make accurate prediction to building energy consumption, while can
Constantly to improve model performance, preferable precision of prediction is obtained.
Artificial neural network has that simulated time is short as a kind of important prediction technique, and be suitable for nonlinear problem etc.
Advantage, but the selection of model parameter will greatly influence model performance in modeling process.Suitable model parameter is chosen, is to answer
With the committed step of ANN model prediction building energy consumption.
The prior art relevant to the application is patent document CN 104134097A, is disclosed a kind of based on GANN-BIM
The severe cold area office building form energy-saving design method of model.The invention Genetic Optimization Algorithm (GA) is by neural network mould
Type (ANN) combines with Building Information Model (BIM), realizes the office building morphogenesis process based on energy consumption data,
Continuous, the quantization search to severe cold area office building form solution space, the GANN-BIM platform of this method institute construction are reached
It introduces neural network model (ANN) and carries out energy consumption prediction.
Summary of the invention
For the defects in the prior art, the object of the present invention is to provide a kind of Energy Consumption of Public Buildings based on GA-ANN is pre-
Survey method and system.
A kind of Energy Consumption of Public Buildings prediction technique based on GA-ANN provided according to the present invention, comprising:
Step S1: monitoring energy consumption obtains influence factor information to obtain the energy consumption data of public building, to energy consumption data and
Influence factor information is pre-processed, and normalization data is obtained；
Step S2: divide training set and test set to normalization data, the training set can training pattern, it is described
Test set can carry out precision of prediction inspection to model, the related coefficient in normalization data between each variable be calculated, based on correlation
The size of coefficient determines the input variable of model；
Step S3: using genetic Optimization Algorithm, establishes artificial nerve network model, and use the training set training mould
Type obtains energy consumption prediction model；
Step S4: the input variable of setting is inputted into energy consumption prediction model, obtains the corresponding Energy Consumption of Public Buildings of output valve
Predicted value；
Step S5: model verifying is carried out to Energy Consumption of Public Buildings predicted value by test set, obtains average absolute percentage
Error, and according to mean absolute percentage error step-up error allowed band, permitted based on Energy Consumption of Public Buildings predicted value and error
Perhaps range is predicted and diagnosis Energy Consumption of Public Buildings jointly.
Preferably, the influence factor information includes out door climatic parameter and constructing operation information；
The weather station historical data that out door climatic parameter is obtained from network, constructing operation information are obtained by investigation on the spot
It obtains or similar constructing operation rule obtains.
Preferably, by the pretreatment, carry out include exceptional value, missing values judgement and processing.
When carrying out judgement exceptional value, determined by Interval Test, the Interval Test is to carry out energy consumption data
After size sequence, use the percentile being located at α % as the section upper limit, interval limit 0, then be by the interval extension
125% to guarantee that data when building normal fluctuation are not misjudged, the expression formula of final judgement section S are as follows:
S=[0,125% × X(α %)]
In formula, S is the section for determining exceptional value；
X(α %)For α percentile, refer to that energy consumption data is sized, accumulates the value at α %；
The data being not in section are judged as abnormal data, and abnormal data is removed.
When carrying out missing values judgement, compared using full time axis and the time shaft of historical data, the time of no data
Point is denoted as missing；A whole day institute's having time data are that 0 data are denoted as missing；In the case that consecutive miss point is less than 5, use
Linear interpolation method carries out completion, calculation formula are as follows:
N (N≤5) is missing point number, Yn(n=1,2 ..., N) is the data value of n-th of missing point, Y0And YN+1Respectively
Adjacent data before and after deletion sequence.
Preferably, the calculating of the related coefficient between each variable is calculated using Pearson correlation coefficient, passes through following formula:
R (X, Y) is variable X, the related coefficient of Y in formula；
Cov (X, Y) is X, the covariance of Y；
Var (X) is the variance of variable X, and Var (Y) is the variance of variable Y；
When related coefficient absolute value is greater than 0.3, variable property relevant with building energy consumption is set, the variable is true
It is set to input variable.
Preferably, the use genetic Optimization Algorithm is using genetic algorithm optimization ANN, the genetic algorithm optimization ANN
For multi-layer perception (MLP) MLP, specific Optimization Steps are as follows:
The first step, the initialization of algorithm parameter, the parameter for needing to initialize have: the hidden layer number of plies, the mind of every layer of hidden layer
Through first number, L2 regularization penalty coefficient α, initial learning rate η；Determine the value range and accuracy of Optimal Parameters, simultaneously
Population at individual quantity N is set as 20, crossover probability Pc is that 80% and mutation probability Pm is 10%, maximum number of iterations 30；
The decimal system is converted to binary coding and generates random generation initial population by second step, and code length l meets as follows
Formula:
2l-1<(Lupper-Llower)/δ≤2l-1
In formula, LupperWith LlowerThe upper limit and lower limit of the decision variable value range are respectively indicated, δ then indicates calculating
Accuracy；
Sequence is filled up using generate at random 0/1, is generated containing 20 individual initial populations, each sequence is known as should
Individual chromosome；
Third step, calculates individual adaptation degree, and the fitness criterion used is predicts the inverse of error, prediction herein misses
Difference refers to the mean absolute percentage error of the neural network prediction result after the training of identical training set data.
Fi=1/MAPEi
In formula, FiIndicate the fitness of i-th of individual, MAPEiIt is the absolute percent error of i-th of individual；MAPE refers to flat
Equal absolute percent error,Refer to predicted value of the model to training data, ykFor the true value of energy consumption data；
4th step judges whether maximum adaptation degree has reached setting value or cycle-index reaches the upper limit, and fitness is greater than
50 or cycle-index stop circulation when reaching 30；
5th step selects the defect individual in parent, the excellent genes of defect individual is enable to be genetic in the next generation, described
The size of fitness is divided in portion the probability selected, and the calculation formula of probability is as follows
In formula, PiIndicate the probability that i-th of individual is selected, FiIndicate that the fitness of the individual, N indicate the number of individual；
Individual is randomly selected by probability, forms new population.
6th step, intersects and variation, intersects for a random selection two individuals and node, two chromosome nodes of exchange it
Part afterwards；Variation a bit, the numerical value of the position is overturn, i.e., 0, which is replaced into 1,1, is replaced into 0 for random selection.
7th step, the new population that the 6th step is generated, comes back to third step and is recycled.
8th step inputs individual optimal at this time, is decoded as decimal number again after the 4th step judges to complete circulation,
Output is the model parameter of the MLP artificial neural network after optimizing.
Preferably, the mean absolute percentage error, is calculate by the following formula:
In formula, MAPE refers to mean absolute percentage error；Refer to predicted value of the model to i-th of test data set, yiFor
The true value of the building energy consumption of i-th of test data set；The number of n expression test data set；
The error upper limit e of the allowable range of errormaxFor
emax=200% × MAPE
When building energy consumption exceeds the error upper limit of allowable range of error, abnormality alarm is issued.
A kind of Energy Consumption of Public Buildings forecasting system based on GA-ANN provided according to the present invention, comprising:
Module S1: monitoring energy consumption obtains influence factor information to obtain the energy consumption data of public building, to energy consumption data and
Influence factor information is pre-processed, and normalization data is obtained；
Module S2: divide training set and test set to normalization data, the training set can training pattern, it is described
Test set can carry out precision of prediction inspection to model, the related coefficient in normalization data between each variable be calculated, based on correlation
The size of coefficient determines the input variable of model；
Module S3: using genetic Optimization Algorithm, establishes artificial nerve network model, and use the training set training mould
Type obtains energy consumption prediction model；
Module S4: the input variable of setting is inputted into energy consumption prediction model, obtains the corresponding Energy Consumption of Public Buildings of output valve
Predicted value；
Module S5: model verifying is carried out to Energy Consumption of Public Buildings predicted value by test set, obtains average absolute percentage
Error, and according to mean absolute percentage error step-up error allowed band, permitted based on Energy Consumption of Public Buildings predicted value and error
Perhaps range is predicted and diagnosis Energy Consumption of Public Buildings jointly.
Compared with prior art, the present invention have it is following the utility model has the advantages that
1. predetermined speed is fast, by GA-ANN modeling and forecasting Energy Consumption of Public Buildings, can quickly be provided after completion training
Energy Consumption of Public Buildings prediction, solves that the physical modeling period is long, the big problem of project amount.
2. precision of prediction is high, by the parameter of genetic algorithm optimization artificial neural network, model performance is improved, to public affairs
The energy consumption precision of prediction built of building together significantly improves.
3. practical, this method can be widely applied to be predicted with the energy consumption of all kinds of public buildings, excellent according to this prediction result
Change building energy management strategy, while the error analysis function of providing can diagnose for constructing operation and provide reference.
Detailed description of the invention
Upon reading the detailed description of non-limiting embodiments with reference to the following drawings, other feature of the invention,
Objects and advantages will become more apparent upon:
Fig. 1 is that Energy Consumption of Public Buildings predicts process schematic in the present invention；
Fig. 2 is the flow chart of genetic algorithm optimization artificial neural network；
Fig. 3 is energy consumption prediction effect figure of the present invention to certain office building.
Specific embodiment
The present invention is described in detail combined with specific embodiments below.Following embodiment will be helpful to the technology of this field
Personnel further understand the present invention, but the invention is not limited in any way.It should be pointed out that the ordinary skill of this field
For personnel, without departing from the inventive concept of the premise, several changes and improvements can also be made.These belong to the present invention
Protection scope.
The present invention is based on the Energy Consumption of Public Buildings prediction technique and system of GA-ANN, by data prediction, variables choice,
Division data set, Optimal Parameters, training pattern, the entire flow of prediction of energy consumption and error assessment, the algorithm can be built to public
The real-time and following energy consumption built carries out high-precision forecast.
As shown in Figure 1, a kind of Energy Consumption of Public Buildings prediction technique based on GA-ANN provided according to the present invention, comprising:
Step S1: monitoring energy consumption obtains influence factor information to obtain the energy consumption data of public building, to energy consumption data and
Influence factor information is pre-processed, and normalization data is obtained；
Step S2: divide training set and test set to normalization data, the training set can training pattern, it is described
Test set can carry out precision of prediction inspection to model, the related coefficient in normalization data between each variable be calculated, based on correlation
The size of coefficient determines the input variable of model；Preferably, the related coefficient between each variable is calculated, comparison obtains related to energy consumption
The high parameter of property, is determined as the input variable of model；
Step S3: using genetic Optimization Algorithm, establishes artificial nerve network model, and use the training set training mould
Type obtains energy consumption prediction model；Preferably, artificial neural network (ANN) model performance is related to hidden layer structure, uses heredity
Algorithm (GA), optimal model parameter is selected according to fitness, it is established that artificial nerve network model；Training set is reused later
Data training pattern obtains energy consumption prediction model；
Step S4: the input variable of setting is inputted into energy consumption prediction model, obtains the corresponding Energy Consumption of Public Buildings of output valve
Predicted value；
Step S5: model verifying is carried out to Energy Consumption of Public Buildings predicted value by test set, obtains average absolute percentage
Error, and according to mean absolute percentage error step-up error allowed band, permitted based on Energy Consumption of Public Buildings predicted value and error
Perhaps range is predicted and diagnosis Energy Consumption of Public Buildings jointly.
Specifically, the influence factor information includes out door climatic parameter and constructing operation information；
The weather station historical data that out door climatic parameter is obtained from network, constructing operation information are obtained by investigation on the spot
It obtains or similar constructing operation rule obtains.Preferably, wherein the public building historical energy consumption data used derives from energy consumption
Monitoring system.Energy consumption monitoring system acquires the total energy consumption of public building and illuminates and socket, air-conditioning system, dynamical system and special
The subitem energy consumption data such as energy consumption.
Specifically, by the pretreatment, carry out include exceptional value, missing values judgement and processing.
When carrying out judgement exceptional value, determined by Interval Test, the Interval Test is to carry out energy consumption data
After size sequence, use the percentile being located at α % as the section upper limit, interval limit 0, then be by the interval extension
125% to guarantee that data when building normal fluctuation are not misjudged, the expression formula of final judgement section S are as follows:
S=[0,125% × X(α %)]
In formula, S is the section for determining exceptional value；
X(α %)For α percentile, refer to that energy consumption data is sized, accumulates the value at α %；Preferably, default α is big
Small is 95, corresponding when data reliability is high to improve.
The data being not in section are judged as abnormal data, and abnormal data is removed.
When carrying out missing values judgement, compared using full time axis and the time shaft of historical data, the time of no data
Point is denoted as missing；A whole day institute's having time data are that 0 data are denoted as missing；In the case that consecutive miss point is less than 5, use
Linear interpolation method carries out completion, calculation formula are as follows:
N (N≤5) is missing point number, Yn(n=1,2 ..., N) is the data value of n-th of missing point, Y0And YN+1Respectively
Adjacent data before and after deletion sequence.
Specifically, the calculating of the related coefficient between each variable is calculated using Pearson correlation coefficient, passes through following formula:
R (X, Y) is variable X, the related coefficient of Y in formula；
Cov (X, Y) is X, the covariance of Y；
Var (X) is the variance of variable X, and Var (Y) is the variance of variable Y；
When related coefficient absolute value is greater than 0.3, variable property relevant with building energy consumption is set, the variable is true
It is set to input variable.Preferably, the variable includes building outdoor temperature, outside humidity, constructing operation state, previous hour energy
Consumption.
Specifically, genetic algorithm optimization ANN, the genetic algorithm optimization ANN are used when the use genetic Optimization Algorithm
For multi-layer perception (MLP) MLP, the network model of the selected ANN used is multi-layer perception (MLP) (MLP), the model parameter for needing to optimize
Have: the hidden layer number of plies and each layer neuron number, L2 regularization penalty coefficient α, learning rate η.
As shown in Fig. 2, specific Optimization Steps are as follows:
The first step, the initialization of algorithm parameter, the parameter for needing to initialize have: the hidden layer number of plies, the mind of every layer of hidden layer
Through first number, L2 regularization penalty coefficient α, initial learning rate η；Determine the value range and accuracy of Optimal Parameters, simultaneously
Population at individual quantity N is set as 20, crossover probability Pc is that 80% and mutation probability Pm is 10%, maximum number of iterations 30；It is excellent
Selection of land determines the value range and accuracy of Optimal Parameters, 1~4 layer of hidden layer number of plies range, precision 1；Every layer of neuron
Number value range 2~200, precision 1；L2 regularization penalty coefficient α range is 10-5~10-3, precision 10-5；Learning rate
Range 0.0001 to 0.01, precision 0.0001.Concurrently setting population at individual quantity N is 20, and crossover probability Pc is 80%, and
Mutation probability Pm is 10%, maximum number of iterations 30.
The decimal system is converted to binary coding and generates random generation initial population by second step, and code length l meets as follows
Formula:
2l-1<(Lupper-Llower)/δ≤2l-1
In formula, LupperWith LlowerThe upper limit and lower limit of the decision variable value range are respectively indicated, δ then indicates calculating
Accuracy；
Sequence is filled up using generate at random 0/1, is generated containing 20 individual initial populations, each sequence is known as should
Individual chromosome；
Third step, calculate individual adaptation degree, individual adaptation degree be evaluate chromosome superiority and inferiority standard and algorithm optimizing according to
According to；The fitness criterion used for, predict the inverse of error, prediction error herein refer to by identical training set data training after
Neural network prediction result mean absolute percentage error.
Fi=1/MAPEi
In formula, FiIndicate the fitness of i-th of individual, MAPEiIt is the absolute percent error of i-th of individual；MAPE refers to flat
Equal absolute percent error,Refer to predicted value of the model to training data, ykFor the true value of energy consumption data；
4th step judges whether maximum adaptation degree has reached setting value or cycle-index reaches the upper limit, and fitness is greater than
50 or cycle-index stop circulation when reaching 30；
5th step selects the defect individual in parent, the excellent genes of defect individual is enable to be genetic in the next generation, selects
Foundation be individual adaptation degree, the higher individual of individual adaptation degree, the probability selected is also higher.The fitness it is big
Small to be divided in portion the probability selected, the calculation formula of probability is as follows
In formula, PiIndicate the probability that i-th of individual is selected, FiIndicate that the fitness of the individual, N indicate the number of individual；
Individual is randomly selected by probability, forms new population.
6th step, intersects and variation, intersects for a random selection two individuals and node, two chromosome nodes of exchange it
Part afterwards；Variation a bit, the numerical value of the position is overturn, i.e., 0, which is replaced into 1,1, is replaced into 0 for random selection.
7th step, the new population that the 6th step is generated, comes back to third step and is recycled.
8th step inputs individual optimal at this time, is decoded as decimal number again after the 4th step judges to complete circulation,
Output is the model parameter of the MLP artificial neural network after optimizing.
Specifically, the mean absolute percentage error, is calculate by the following formula:
In formula, MAPE refers to mean absolute percentage error；Refer to predicted value of the model to i-th of test data set, yiFor
The true value of the building energy consumption of i-th of test data set；The number of n expression test data set；
The error upper limit e of the allowable range of errormaxFor
emax=200% × MAPE
When building energy consumption exceeds the error upper limit of allowable range of error, abnormality alarm is issued.It is counted in practical application
According to error be more than the upper limit, the operation of the public building is likely to occur exception, can sound an alarm to monitoring system.
A kind of Energy Consumption of Public Buildings forecasting system based on GA-ANN provided according to the present invention, comprising:
Module S1: monitoring energy consumption obtains influence factor information to obtain the energy consumption data of public building, to energy consumption data and
Influence factor information is pre-processed, and normalization data is obtained；
Module S2: divide training set and test set to normalization data, the training set can training pattern, it is described
Test set can carry out precision of prediction inspection to model, the related coefficient in normalization data between each variable be calculated, based on correlation
The size of coefficient determines the input variable of model；
Module S3: using genetic Optimization Algorithm, establishes artificial nerve network model, and use the training set training mould
Type obtains energy consumption prediction model；
Module S4: the input variable of setting is inputted into energy consumption prediction model, obtains the corresponding Energy Consumption of Public Buildings of output valve
Predicted value；
Module S5: model verifying is carried out to Energy Consumption of Public Buildings predicted value by test set, obtains average absolute percentage
Error, and according to mean absolute percentage error step-up error allowed band, permitted based on Energy Consumption of Public Buildings predicted value and error
Perhaps range is predicted and diagnosis Energy Consumption of Public Buildings jointly.
Energy Consumption of Public Buildings forecasting system provided by the invention based on GA-ANN, can be by based on the public of GA-ANN
The step process of building energy consumption prediction technique is realized.Those skilled in the art can be pre- by the Energy Consumption of Public Buildings based on GA-ANN
Survey method is interpreted as the preference of the Energy Consumption of Public Buildings forecasting system based on GA-ANN.
According to the present invention with the thought of data basis prediction building energy consumption, for the need of Energy Consumption of Public Buildings high-precision forecast
The target for ANN algorithm parameter optimization of summing carries out high-precision forecast to the real-time and following energy consumption of public building, is building energy
Scientific management foundation is provided, genetic algorithm is applied to the parameter optimization of artificial neural network, it is effective to improve the pre- of model
Survey precision.Genetic algorithm is a kind of optimization algorithm of model organism evolutionary process search optimal solution, because it is that multiple individuals are total
It is same to scan for, so being easier to obtain globally optimal solution.
In a particular embodiment, implemented down based on the technical solution of the present invention, by going through for office in Shanghai
History data and open source Python code, the detailed implementation method and specific operation process are given.It should be pointed out that this
For the those of ordinary skill in field, without departing from the inventive concept of the premise, various modifications and improvements can be made, this
Belong to protection scope of the present invention.
Energy Consumption of Public Buildings prediction technique based on GA-ANN, it includes following main implementation steps:
Step S1, Energy in use monitoring system obtain the energy consumption data of public building, including energy consumption monitoring system acquisition public affairs
It builds together the total energy consumption built and illumination and socket, air-conditioning system, dynamical system and special energy consumption four subitem energy consumption data greatly, in this example
Data length 2 years in total, use wherein 3 months data；Influence factor includes out door climatic parameter and constructing operation information,
Influence factor includes out door climatic parameter, is obtained by weather monitoring station data, and the meteorological data at corresponding time point is obtained.To acquirement
Data pre-processed, differentiation and processing including exceptional value, missing values finally carry out data normalization；
Specifically, the public building historical energy consumption data used derives from energy consumption monitoring system, and out door climatic parameter comes from
In the weather station historical data that network obtains, including outdoor temperature humidity, day illumination, wind speed, rainfall etc..Constructing operation information according to
It obtains or by investigating on the spot similar to constructing operation rule；When pretreatment determines exceptional value, determine that energy consumption is abnormal by Interval Test,
It is normal data that institute, which is according to most of energy consumption using section preparation method, after the ascending sequence of data, using positioned at α %
Then the interval extension is 125% to guarantee to build positive ordinary wave as the section upper limit, interval limit 0 by the percentile at place
Data when dynamic are not misjudged, the expression formula of final judgement section S are as follows:
S=[0,125% × X(α %)]
In formula, S is the section for determining exceptional value；X(α %)For α percentile, refer to that data are sized, accumulates at α %
Value, default α size is 95, corresponding when data reliability is high to improve；
The data being not in section are judged as abnormal data, remove this partial data point.Missing values determination method is
It is compared using full time axis and the time shaft of historical data, the time point of no data is denoted as missing；A whole day institute's having time number
Missing is denoted as according to the data for being 0.In the case that consecutive miss point is less than 5, completion, calculation formula are carried out using linear interpolation method
Are as follows:
N (N≤5) is missing point number, Yn(n=1,2 ..., N) is the data value of n-th of missing point, Y0And YN+1Respectively
Adjacent data before and after deletion sequence.
Step S2 divides training set and test set, and training set is used for training pattern, and test set is for test model prediction essence
Degree；Training set is summer bimestrial history energy consumption in this example, and test set is one week continuous energy consumption sequence for being all summer.It calculates
Related coefficient between each variable, comparison obtain the parameter high with energy consumption correlation, meet the parameter of correlation criterion, be determined as mould
The input variable of type；
Specifically, related coefficient calculation method used in the step S2 is Pearson correlation coefficient, calculation formula are as follows:
R (X, Y) is variable X in formula, and the related coefficient of Y, Cov (X, Y) is X, and the covariance of Y, Var (X) is the side of variable X
Difference.
When related coefficient absolute value is greater than 0.3, it is believed that variable property relevant with building energy consumption, determination is defeated in this example
Enter variable, including building outdoor temperature, outside humidity, constructing operation state, previous hour energy consumption.
Step S3, artificial neural network (ANN) model performance is related to hidden layer structure, uses genetic algorithm (GA), root
Optimal model parameter is selected according to fitness, it is established that artificial nerve network model, detailed process are detailed in explanation later
It provides；Then training set data training pattern is reused, energy consumption prediction model is obtained；
Use the parameter selection of genetic algorithm optimization ANN.The network model of the selected ANN used is multi-layer perception (MLP)
(MLP), the model parameter for needing to optimize has: the hidden layer number of plies and each layer neuron number, learning rate η.Specific optimization step
It is rapid as follows:
The first step, the initialization of algorithm parameter, the parameter for needing to initialize have: the hidden layer number of plies, the mind of every layer of hidden layer
Through first number, L2 regularization penalty coefficient α, initial learning rate η；It determines the value range and accuracy of Optimal Parameters, hides
1~4 layer of range, precision 1 are counted layer by layer；Every layer of neuron number value range 2~200, precision 1；L2 regularization punishment system
Number α range is 10-5~10-3, precision 10-5；Learning rate range 0.0001 to 0.01, precision 0.0001.Concurrently set kind
Group individual amount N is 20, and crossover probability Pc is that 80% and mutation probability Pm is 10%, maximum number of iterations 30.
The decimal system is converted to binary coding and generates random generation initial population by second step.Code length l meets as follows
Formula:
2l-1<(Lupper-Llower)/δ≤2l-1
In formula, LupperWith LlowerThe upper limit and lower limit of the decision variable value range are respectively indicated, δ then indicates calculating
Accuracy.
Sequence is filled up using generate at random 0/1 later, is generated containing 20 individual initial populations, each sequence claims
For the individual chromosome.
Third step, calculate individual adaptation degree, individual adaptation degree be evaluate chromosome superiority and inferiority standard and algorithm optimizing according to
According to.The fitness criterion used for, predict the inverse of error, prediction error herein refer to by identical training set data training after
Neural network prediction result mean absolute percentage error.
Fi=1/MAPEi
In formula, FiIndicate that the fitness of i-th of individual, MAPE refer to mean absolute percentage error,Refer to model to training
The predicted value of data, ykFor the true value of energy consumption data.
4th step judges whether maximum adaptation degree has reached setting value or cycle-index reaches the upper limit.Fitness is greater than
50 or cycle-index stop circulation when reaching 30.
5th step selects the defect individual in parent, and being allowed to its excellent genes can be genetic in the next generation.The foundation of selection
It is individual adaptation degree, the higher individual of individual adaptation degree, the probability selected is also higher.This sentence the size of fitness by
The calculation formula of the probability that pro rate is selected, probability is as follows
In formula, PiIndicate the probability that i-th of individual is selected, FiIndicate the fitness of the individual.
Probability P is pressed lateriIndividual is randomly selected, new population is formed.
6th step, intersects and variation, intersects for a random selection two individuals and node, two chromosome nodes of exchange it
Part afterwards；Variation a bit, the numerical value of the position is overturn, i.e., 0, which is replaced into 1,1, is replaced into 0 for random selection.Intersect and becomes
Different probability is respectively 80% and 10%.
7th step, the new population that the 6th step is generated back within third step and carry out next circulation, full until recycling
Sufficient condition terminates.
8th step inputs individual optimal at this time, is decoded as decimal number again after the 4th step judges to complete circulation,
Output is the model parameter of the MLP artificial neural network after optimizing.Model parameter is 1 layer hidden after algorithm optimization in this example
Layer, 32 hidden neurons are hidden, L2 regularization penalty coefficient is 0.0518, and initial learning rate is 0.0171.
Step S4, input need the corresponding input variable in period predicted, i.e., by the input variable sequence in test set,
As the output of model after training, the corresponding Energy Consumption of Public Buildings predicted value of the output valve obtained from；
Step S5, verifies model in test set, by mean percent ratio error (MAPE), and provides the error model of permission
It encloses, if there is energy consumption error is greater than the error upper limit, then sounds an alarm；Predicted value and error range are Energy Consumption of Public Buildings jointly
Monitoring provides foundation with diagnosis.The model error index of selection is mean absolute percentage error MAPE, and calculating formula is as follows:
In formula,Refer to predicted value of the model to test data, yiFor the true value of test data set energy consumption.Obtain model certainly
After the prediction error value of body, the error upper limit of permission is
emax=200% × MAPE
The error that data are obtained in practical application is more than the upper limit, then the constructing operation is likely to occur exception.
The above method is applied to office in Shanghai, predicts following five days energy using bimestrial energy consumption data
Consumption, as a result as shown in Figure 3, it can be seen that high-precision energy consumption prediction may be implemented in this method.
One skilled in the art will appreciate that in addition to realizing system provided by the invention in a manner of pure computer readable program code
It, completely can be by the way that method and step be carried out programming in logic come so that provided by the invention other than system, device and its modules
System, device and its modules are declined with logic gate, switch, specific integrated circuit, programmable logic controller (PLC) and insertion
The form of controller etc. realizes identical program.So system provided by the invention, device and its modules may be considered that
It is a kind of hardware component, and the knot that the module for realizing various programs for including in it can also be considered as in hardware component
Structure；It can also will be considered as realizing the module of various functions either the software program of implementation method can be Hardware Subdivision again
Structure in part.
Specific embodiments of the present invention are described above.It is to be appreciated that the invention is not limited to above-mentioned
Particular implementation, those skilled in the art can make a variety of changes or modify within the scope of the claims, this not shadow
Ring substantive content of the invention.In the absence of conflict, the feature in embodiments herein and embodiment can any phase
Mutually combination.
Claims (10)
1. a kind of Energy Consumption of Public Buildings prediction technique based on GA-ANN characterized by comprising
Step S1: energy consumption is monitored to obtain the energy consumption data of public building, influence factor information is obtained, on energy consumption data and influence
Factor information is pre-processed, and normalization data is obtained；
Step S2: divide training set and test set to normalization data, the training set can training pattern, the test
Collection can carry out precision of prediction inspection to model, calculate the related coefficient in normalization data between each variable, be based on related coefficient
Size determine the input variable of model；
Step S3: using genetic Optimization Algorithm, establishes artificial nerve network model, and using the training set training model, obtains
To energy consumption prediction model；
Step S4: inputting energy consumption prediction model for the input variable of setting, obtains the corresponding Energy Consumption of Public Buildings prediction of output valve
Value；
Step S5: carrying out model verifying to Energy Consumption of Public Buildings predicted value by test set, obtain mean absolute percentage error,
And according to mean absolute percentage error step-up error allowed band, it is based on Energy Consumption of Public Buildings predicted value and allowable range of error
Common prediction and diagnosis Energy Consumption of Public Buildings.
2. the Energy Consumption of Public Buildings prediction technique according to claim 1 based on GA-ANN, which is characterized in that the influence
Factor information includes out door climatic parameter and constructing operation information；
The weather station historical data that out door climatic parameter is obtained from network, constructing operation information by the spot investigation obtain or
Person obtains similar to constructing operation rule.
3. the Energy Consumption of Public Buildings prediction technique according to claim 1 based on GA-ANN, which is characterized in that by described
Pretreatment, carry out include exceptional value, missing values judgement and processing.
4. the Energy Consumption of Public Buildings prediction technique according to claim 3 based on GA-ANN, which is characterized in that
When carrying out judgement exceptional value, determined by Interval Test, the Interval Test is that energy consumption data is carried out size
After sequence, use the percentile being located at α % as the section upper limit, interval limit 0, then be by the interval extension
125% to guarantee that data when building normal fluctuation are not misjudged, the expression formula of the section S of final judgement exceptional value are as follows:
S=[0,125% × X(α %)]
In formula, S is the section for determining exceptional value；
X(α %)For α percentile, refer to that energy consumption data is sized, accumulates the value at α %；
The data being not in section are judged as abnormal data, and abnormal data is removed.
5. the Energy Consumption of Public Buildings prediction technique according to claim 3 based on GA-ANN, which is characterized in that
It when carrying out missing values judgement, is compared using full time axis and the time shaft of historical data, the time point note of no data
For missing；A whole day institute's having time data are that 0 data are denoted as missing；In the case that consecutive miss point is less than 5, using linear
Interpolation method carries out completion, calculation formula are as follows:
N (N≤5) is missing point number, Yn(n=1,2 ..., N) is the data value of n-th of missing point, Y0And YN+1Respectively lack
Adjacent data before and after out-of-sequence column.
6. the Energy Consumption of Public Buildings prediction technique according to claim 1 based on GA-ANN, which is characterized in that each change
The calculating of related coefficient between amount is calculated using Pearson correlation coefficient, passes through following formula:
R (X, Y) is variable X, the related coefficient of Y in formula；
Cov (X, Y) is X, the covariance of Y；
Var (X) is the variance of variable X, and Var (Y) is the variance of variable Y；
When related coefficient absolute value is greater than 0.3, variable property relevant with building energy consumption is set, the variable is determined as
Input variable.
7. the Energy Consumption of Public Buildings prediction technique according to claim 1 based on GA-ANN, which is characterized in that the use
Genetic algorithm optimization ANN is used when genetic Optimization Algorithm.
8. the Energy Consumption of Public Buildings prediction technique according to claim 7 based on GA-ANN, which is characterized in that the heredity
Algorithm optimization ANN is multi-layer perception (MLP) MLP, and specific Optimization Steps are as follows:
The first step, the initialization of algorithm parameter, the parameter for needing to initialize have: the hidden layer number of plies, the neuron of every layer of hidden layer
Number, L2 regularization penalty coefficient α, initial learning rate η；The value range and accuracy for determining Optimal Parameters, concurrently set
Population at individual quantity N is 20, and crossover probability Pc is that 80% and mutation probability Pm is 10%, maximum number of iterations 30；
The decimal system is converted to binary coding and generates random generation initial population by second step, and code length l meets following public
Formula:
2l-1< (Lupper-Llower)/δ≤2l-1
In formula, LupperWith LlowerThe upper limit and lower limit of the decision variable value range are respectively indicated, δ then indicates to calculate accurate
Degree；
Sequence is filled up using generate at random 0/1, is generated containing 20 individual initial populations, each sequence is known as the individual
Chromosome；
Third step, calculates individual adaptation degree, and the fitness criterion used is predicts the inverse of error, prediction error herein refers to
The mean absolute percentage error of neural network prediction result after the training of identical training set data.
Fi=1/MAPEi
In formula, FiIndicate the fitness of i-th of individual, MAPEiIt is the absolute percent error of i-th of individual；MAPE refers to averagely absolutely
To percentage error,Refer to predicted value of the model to training data, ykFor the true value of energy consumption data, n indicates the number of individual；
4th step judges whether maximum adaptation degree has reached setting value or cycle-index reaches the upper limit, fitness be greater than 50 or
Person's cycle-index stops circulation when reaching 30；
5th step selects the defect individual in parent, so that the excellent genes of defect individual is genetic in the next generation, the adaptation
The size of degree is divided in portion the probability selected, and the calculation formula of probability is as follows
In formula, PiIndicate the probability that i-th of individual is selected, FiIndicate that the fitness of the individual, N indicate the number of individual；By general
Rate randomly selects individual, forms new population.
6th step, intersects and variation, intersects for a random selection two individuals and node, after exchanging two chromosome nodes
Part；Variation a bit, the numerical value of the position is overturn, i.e., 0, which is replaced into 1,1, is replaced into 0 for random selection.
7th step, the new population that the 6th step is generated, comes back to third step and is recycled.
8th step inputs individual optimal at this time, is decoded as decimal number again after the 4th step judges to complete circulation, exports
The model parameter of MLP artificial neural network after as optimizing.
9. the Energy Consumption of Public Buildings prediction technique according to claim 1 based on GA-ANN, which is characterized in that described average
Absolute percent error, is calculate by the following formula:
In formula, MAPE refers to mean absolute percentage error；Refer to predicted value of the model to i-th of test data set, yiIt is i-th
The true value of the building energy consumption of test data set；The number of n expression test data set；
The error upper limit e of the allowable range of errormaxFor
emax=200% × MAPE
When building energy consumption exceeds the error upper limit of allowable range of error, abnormality alarm is issued.
10. a kind of Energy Consumption of Public Buildings forecasting system based on GA-ANN characterized by comprising
Module S1: energy consumption is monitored to obtain the energy consumption data of public building, influence factor information is obtained, on energy consumption data and influence
Factor information is pre-processed, and normalization data is obtained；
Module S2: divide training set and test set to normalization data, the training set can training pattern, the test
Collection can carry out precision of prediction inspection to model, calculate the related coefficient in normalization data between each variable, be based on related coefficient
Size determine the input variable of model；
Module S3: using genetic Optimization Algorithm, establishes artificial nerve network model, and using the training set training model, obtains
To energy consumption prediction model；
Module S4: inputting energy consumption prediction model for the input variable of setting, obtains the corresponding Energy Consumption of Public Buildings prediction of output valve
Value；
Module S5: carrying out model verifying to Energy Consumption of Public Buildings predicted value by test set, obtain mean absolute percentage error,
And according to mean absolute percentage error step-up error allowed band, it is based on Energy Consumption of Public Buildings predicted value and allowable range of error
Common prediction and diagnosis Energy Consumption of Public Buildings.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910168405.9A CN110046743B (en) | 2019-03-06 | 2019-03-06 | Public building energy consumption prediction method and system based on GA-ANN |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910168405.9A CN110046743B (en) | 2019-03-06 | 2019-03-06 | Public building energy consumption prediction method and system based on GA-ANN |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110046743A true CN110046743A (en) | 2019-07-23 |
| CN110046743B CN110046743B (en) | 2021-06-04 |
Family
ID=67274563
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910168405.9A Active CN110046743B (en) | 2019-03-06 | 2019-03-06 | Public building energy consumption prediction method and system based on GA-ANN |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110046743B (en) |
Cited By (18)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110443425A (en) * | 2019-08-09 | 2019-11-12 | 长江慧控科技(武汉)有限公司 | Intelligent railway station electric energy energy consumption prediction technique based on Prophet |
| CN110471380A (en) * | 2019-08-15 | 2019-11-19 | 四川长虹电器股份有限公司 | A kind of air conditioning failure monitoring and method for early warning for smart home system |
| CN111126707A (en) * | 2019-12-26 | 2020-05-08 | 华自科技股份有限公司 | Energy consumption equation construction and energy consumption prediction method and device |
| CN111597609A (en) * | 2020-04-28 | 2020-08-28 | 马辰龙 | Basic operation unit containing solar radiation and building energy consumption rapid simulation method applying same |
| CN111766832A (en) * | 2020-06-29 | 2020-10-13 | 重庆大学 | Numerical control machine tool cutting energy consumption prediction modeling method driven by incomplete data |
| CN112182720A (en) * | 2020-10-13 | 2021-01-05 | 北京建筑技术发展有限责任公司 | Building energy consumption model evaluation method based on building energy management application scene |
| CN112365082A (en) * | 2020-11-25 | 2021-02-12 | 马鞍山学院 | Public energy consumption prediction method based on machine learning |
| CN112464554A (en) * | 2020-11-03 | 2021-03-09 | 桂林理工大学 | Operating parameter optimization method of gasoline refining equipment |
| CN112686442A (en) * | 2020-12-29 | 2021-04-20 | 博锐尚格科技股份有限公司 | Air conditioner tail end energy consumption prediction method and system based on operation diversity |
| CN112700037A (en) * | 2020-12-28 | 2021-04-23 | 成都国铁电气设备有限公司 | Method and device for predicting number of foreign body defects of high-speed railway contact net |
| CN112712213A (en) * | 2021-01-15 | 2021-04-27 | 上海交通大学 | Method and system for predicting energy consumption of deep migration learning of centralized air-conditioning house |
| CN112906956A (en) * | 2021-02-05 | 2021-06-04 | 希盟泰克(重庆)实业发展有限公司 | BIM and CIM combined urban energy consumption prediction method |
| WO2021120829A1 (en) * | 2019-12-20 | 2021-06-24 | 上海市建筑科学研究院有限公司 | Method for diagnosing energy consumption of lighting socket branches of buildings during non-operational periods |
| CN113486433A (en) * | 2020-12-31 | 2021-10-08 | 上海东方低碳科技产业股份有限公司 | Method for calculating energy consumption shortage number of net zero energy consumption building and filling system |
| CN113705867A (en) * | 2021-08-16 | 2021-11-26 | 浙江浙大中控信息技术有限公司 | Equipment abnormity diagnosis method based on energy consumption interval prediction |
| CN116128124A (en) * | 2023-01-09 | 2023-05-16 | 北京建筑大学 | Building energy consumption prediction method based on abnormal energy value processing and time sequence decomposition |
| CN116579506A (en) * | 2023-07-13 | 2023-08-11 | 陕西通信规划设计研究院有限公司 | Building energy consumption data intelligent management method and system based on big data |
| CN117688846A (en) * | 2024-02-02 | 2024-03-12 | 杭州经纬信息技术股份有限公司 | Reinforced learning prediction method and system for building energy consumption and storage medium |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101173823B1 (en) * | 2011-07-01 | 2012-08-20 | 연세대학교 산학협력단 | System and method for predicting an energy consumption of multi-family housing |
| CN104134097A (en) * | 2014-07-01 | 2014-11-05 | 哈尔滨工业大学 | Severe cold area office building form energy-saving design method based on GANN-BIM model |
| CN106920006A (en) * | 2017-02-23 | 2017-07-04 | 北京工业大学 | A kind of subway station air conditioning energy consumption Forecasting Methodology based on ISOA LSSVM |
| CN109242088A (en) * | 2018-07-23 | 2019-01-18 | 大冶特殊钢股份有限公司 | Heat treatment method and device based on GA-ANN artificial nerve network model |
-
2019
- 2019-03-06 CN CN201910168405.9A patent/CN110046743B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101173823B1 (en) * | 2011-07-01 | 2012-08-20 | 연세대학교 산학협력단 | System and method for predicting an energy consumption of multi-family housing |
| CN104134097A (en) * | 2014-07-01 | 2014-11-05 | 哈尔滨工业大学 | Severe cold area office building form energy-saving design method based on GANN-BIM model |
| CN106920006A (en) * | 2017-02-23 | 2017-07-04 | 北京工业大学 | A kind of subway station air conditioning energy consumption Forecasting Methodology based on ISOA LSSVM |
| CN109242088A (en) * | 2018-07-23 | 2019-01-18 | 大冶特殊钢股份有限公司 | Heat treatment method and device based on GA-ANN artificial nerve network model |
Non-Patent Citations (1)
| Title |
|---|
| JONATHAN REYNOLDS等: ""A zone-level，building energy optimisation combining an artificial neural network，a genetic algorithm，and model predictive control"", 《ENERGY》 * |
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110443425A (en) * | 2019-08-09 | 2019-11-12 | 长江慧控科技(武汉)有限公司 | Intelligent railway station electric energy energy consumption prediction technique based on Prophet |
| CN110471380A (en) * | 2019-08-15 | 2019-11-19 | 四川长虹电器股份有限公司 | A kind of air conditioning failure monitoring and method for early warning for smart home system |
| WO2021120829A1 (en) * | 2019-12-20 | 2021-06-24 | 上海市建筑科学研究院有限公司 | Method for diagnosing energy consumption of lighting socket branches of buildings during non-operational periods |
| CN111126707A (en) * | 2019-12-26 | 2020-05-08 | 华自科技股份有限公司 | Energy consumption equation construction and energy consumption prediction method and device |
| CN111126707B (en) * | 2019-12-26 | 2023-10-27 | 华自科技股份有限公司 | Energy consumption equation construction and energy consumption prediction method and device |
| CN111597609A (en) * | 2020-04-28 | 2020-08-28 | 马辰龙 | Basic operation unit containing solar radiation and building energy consumption rapid simulation method applying same |
| CN111766832A (en) * | 2020-06-29 | 2020-10-13 | 重庆大学 | Numerical control machine tool cutting energy consumption prediction modeling method driven by incomplete data |
| CN112182720A (en) * | 2020-10-13 | 2021-01-05 | 北京建筑技术发展有限责任公司 | Building energy consumption model evaluation method based on building energy management application scene |
| CN112182720B (en) * | 2020-10-13 | 2023-07-28 | 北京建筑技术发展有限责任公司 | Building energy consumption model evaluation method based on building energy management application scene |
| CN112464554A (en) * | 2020-11-03 | 2021-03-09 | 桂林理工大学 | Operating parameter optimization method of gasoline refining equipment |
| CN112365082A (en) * | 2020-11-25 | 2021-02-12 | 马鞍山学院 | Public energy consumption prediction method based on machine learning |
| CN112700037A (en) * | 2020-12-28 | 2021-04-23 | 成都国铁电气设备有限公司 | Method and device for predicting number of foreign body defects of high-speed railway contact net |
| CN112686442A (en) * | 2020-12-29 | 2021-04-20 | 博锐尚格科技股份有限公司 | Air conditioner tail end energy consumption prediction method and system based on operation diversity |
| CN113486433A (en) * | 2020-12-31 | 2021-10-08 | 上海东方低碳科技产业股份有限公司 | Method for calculating energy consumption shortage number of net zero energy consumption building and filling system |
| CN112712213A (en) * | 2021-01-15 | 2021-04-27 | 上海交通大学 | Method and system for predicting energy consumption of deep migration learning of centralized air-conditioning house |
| CN112906956A (en) * | 2021-02-05 | 2021-06-04 | 希盟泰克(重庆)实业发展有限公司 | BIM and CIM combined urban energy consumption prediction method |
| CN113705867A (en) * | 2021-08-16 | 2021-11-26 | 浙江浙大中控信息技术有限公司 | Equipment abnormity diagnosis method based on energy consumption interval prediction |
| CN116128124A (en) * | 2023-01-09 | 2023-05-16 | 北京建筑大学 | Building energy consumption prediction method based on abnormal energy value processing and time sequence decomposition |
| CN116579506A (en) * | 2023-07-13 | 2023-08-11 | 陕西通信规划设计研究院有限公司 | Building energy consumption data intelligent management method and system based on big data |
| CN116579506B (en) * | 2023-07-13 | 2023-09-19 | 陕西通信规划设计研究院有限公司 | Building energy consumption data intelligent management method and system based on big data |
| CN117688846A (en) * | 2024-02-02 | 2024-03-12 | 杭州经纬信息技术股份有限公司 | Reinforced learning prediction method and system for building energy consumption and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110046743B (en) | 2021-06-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110046743A (en) | Energy Consumption of Public Buildings prediction technique and system based on GA-ANN | |
| CN113962364B (en) | Multi-factor power load prediction method based on deep learning | |
| CN106951611A (en) | A kind of severe cold area energy-saving design in construction optimization method based on user's behavior | |
| CN109063366A (en) | A kind of building performance online data preprocess method based on time and spatial weighting | |
| CN104484715A (en) | Neural network and particle swarm optimization algorithm-based building energy consumption predicting method | |
| CN113205203A (en) | CNN-LSTM-based building energy consumption prediction method and system | |
| CN110866640A (en) | Power load prediction method based on deep neural network | |
| Alamin et al. | An Artificial Neural Network (ANN) model to predict the electric load profile for an HVAC system | |
| CN111915092A (en) | Ultra-short-term wind power prediction method based on long-time and short-time memory neural network | |
| CN112149890A (en) | Comprehensive energy load prediction method and system based on user energy label | |
| CN113554466A (en) | Short-term power consumption prediction model construction method, prediction method and device | |
| CN115470862A (en) | Dynamic self-adaptive load prediction model combination method | |
| CN114444561A (en) | PM2.5 prediction method based on CNNs-GRU fusion deep learning model | |
| CN110555524A (en) | training sample data acquisition method and device based on indoor environment monitoring | |
| CN112365056A (en) | Electrical load joint prediction method and device, terminal and storage medium | |
| CN114912716A (en) | Short-term power load prediction method and system based on bidirectional long-term and short-term memory network | |
| CN114648176A (en) | Wind-solar power consumption optimization method based on data driving | |
| Jayashankara et al. | A novel approach for short-term energy forecasting in smart buildings | |
| CN113868954A (en) | Air conditioner load prediction method and system | |
| CN113111419A (en) | Method and system for establishing and predicting air-conditioning load prediction model in office building | |
| CN117634678A (en) | Low-carbon park carbon emission prediction method based on actual operation scene | |
| CN113762591B (en) | Short-term electric quantity prediction method and system based on GRU and multi-core SVM countermeasure learning | |
| CN112801388B (en) | Power load prediction method and system based on nonlinear time series algorithm | |
| Yang et al. | Prediction research on the energy consumption of public building based on mlr-bp neural network | |
| Balabanova et al. | Voice control and management in smart home system by artificial intelligence |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |