CN110046743B - Public building energy consumption prediction method and system based on GA-ANN - Google Patents
Public building energy consumption prediction method and system based on GA-ANN Download PDFInfo
- Publication number
- CN110046743B CN110046743B CN201910168405.9A CN201910168405A CN110046743B CN 110046743 B CN110046743 B CN 110046743B CN 201910168405 A CN201910168405 A CN 201910168405A CN 110046743 B CN110046743 B CN 110046743B
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- data
- value
- model
- 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.)
- Active
Links
- 238000005265 energy consumption Methods 0.000 title claims abstract description 155
- 238000000034 method Methods 0.000 title claims abstract description 49
- 238000012549 training Methods 0.000 claims abstract description 52
- 238000012360 testing method Methods 0.000 claims abstract description 44
- 238000013528 artificial neural network Methods 0.000 claims abstract description 39
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 39
- 230000002068 genetic effect Effects 0.000 claims abstract description 28
- 238000012544 monitoring process Methods 0.000 claims abstract description 16
- 238000007781 pre-processing Methods 0.000 claims abstract description 12
- 230000002159 abnormal effect Effects 0.000 claims description 29
- 238000005457 optimization Methods 0.000 claims description 24
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 claims description 19
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 claims description 19
- 238000004364 calculation method Methods 0.000 claims description 16
- 210000000349 chromosome Anatomy 0.000 claims description 12
- 210000002569 neuron Anatomy 0.000 claims description 10
- 238000012217 deletion Methods 0.000 claims description 8
- 230000037430 deletion Effects 0.000 claims description 8
- 238000007689 inspection Methods 0.000 claims description 7
- 230000035772 mutation Effects 0.000 claims description 6
- 238000012795 verification Methods 0.000 claims description 6
- 108090000623 proteins and genes Proteins 0.000 claims description 5
- 238000011835 investigation Methods 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 230000000694 effects Effects 0.000 abstract description 2
- 238000012216 screening Methods 0.000 abstract 1
- 238000004458 analytical method Methods 0.000 description 3
- 238000005286 illumination Methods 0.000 description 3
- 238000012986 modification Methods 0.000 description 3
- 230000004048 modification Effects 0.000 description 3
- 238000004378 air conditioning Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 150000001875 compounds Chemical class 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 238000011156 evaluation Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
Images
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—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/08—Construction
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Biophysics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- General Business, Economics & Management (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Marketing (AREA)
- Evolutionary Biology (AREA)
- Artificial Intelligence (AREA)
- Tourism & Hospitality (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Primary Health Care (AREA)
- Physiology (AREA)
- Genetics & Genomics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a GA-ANN-based public building energy consumption prediction method and a system, which are used for collecting data of time-by-time energy consumption and influence factors of the time-by-time energy consumption of a public building, sorting the data and preprocessing the data; dividing a training set and a test set, and screening input variables by a correlation coefficient method; inputting test data, optimizing relevant parameters of the artificial neural network model through a genetic algorithm, and then training the model by using training set data; predicting the energy consumption of the public building by inputting the input variable of the predicted period; and finally, evaluating the prediction effect on the test set through the error index, and giving out an allowable error range. The invention provides a flow and a method for predicting energy consumption of public buildings by using a genetic algorithm and an artificial neural network. The method realizes high-precision prediction for public buildings and provides basis for monitoring, managing and diagnosing the energy consumption of the public buildings.
Description
Technical Field
The invention relates to the technical field of building energy intelligence, in particular to a public building energy consumption prediction method and system based on GA-ANN, and particularly relates to a method for predicting public building energy consumption by using a machine learning method.
Background
With the development of society, the energy consumption of buildings is continuously increased, and public buildings as energy consumers receive attention from the society. The prediction of the energy consumption of the public buildings is an important component in the public building management and building energy saving process, and has important significance for improving the utilization rate of building energy and protecting the environment. In recent years, energy consumption monitoring systems and data platforms for public buildings are developed continuously, relevant data are collected through sensors and the internet, and a predicted data base is obtained in recent years. Based on historical data, prediction and control are driven through a data method, and data support is provided for scientific management of building energy.
The traditional building energy consumption prediction analysis method is generally a physical modeling method and has the defects of long time consumption for modeling and calculation, high model complexity, complicated application and the like. The building energy consumption model based on the data driving method does not need physical parameters and thermodynamic equilibrium equations, can make more accurate prediction on the building energy consumption only by means of analysis of past historical data, and can continuously improve the performance of the model to obtain better prediction precision.
As an important prediction method, the artificial neural network has the advantages of short simulation time, suitability for nonlinear problems and the like, but the performance of the model is greatly influenced by the selection of model parameters in the modeling process. Selecting proper model parameters is a key step for predicting building energy consumption by applying an ANN model.
The prior art related to the present application is patent document CN 104134097 a, which discloses a method for designing office building shape energy saving in severe cold region based on GANN-BIM model. The invention combines a neural network model (ANN) and a Building Information Model (BIM) by applying a genetic optimization algorithm (GA), realizes the office building form generation process based on energy consumption data, achieves continuous and quantitative search of office building form solution space in severe cold regions, and introduces the neural network model (ANN) into the GANN-BIM platform constructed by the method for energy consumption prediction.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a method and a system for predicting the energy consumption of public buildings based on GA-ANN.
The invention provides a public building energy consumption prediction method based on GA-ANN, which comprises the following steps:
step S1: monitoring energy consumption to obtain energy consumption data of a public building, obtaining influence factor information, and preprocessing the energy consumption data and the influence factor information to obtain normalized data;
step S2: dividing the normalized data into a training set and a test set, wherein the training set can train a model, the test set can carry out prediction precision test on the model, calculate correlation coefficients among variables in the normalized data, and determine input variables of the model based on the magnitude of the correlation coefficients;
step S3: establishing an artificial neural network model by using a genetic optimization algorithm, and training the model by using a training set to obtain an energy consumption prediction model;
step S4: inputting the set input variable into the energy consumption prediction model to obtain a public building energy consumption prediction value corresponding to the output value;
step S5: and performing model verification on the public building energy consumption predicted value through a test set to obtain an average absolute percentage error, setting an error allowable range according to the average absolute percentage error, and jointly predicting and diagnosing the public building energy consumption based on the public building energy consumption predicted value and the error allowable range.
Preferably, the influence factor information includes outdoor weather parameters and building operation information;
the outdoor weather parameters are from weather station historical data acquired by a network, and the building operation information is acquired by means of on-site investigation or similar building operation rules.
Preferably, by the preprocessing, determination and processing including an abnormal value, a missing value are performed.
When abnormal values are judged, the abnormal values are judged through interval inspection, the interval inspection is to sort the energy consumption data in size, use a percentile positioned at alpha% as an upper limit of an interval, the lower limit of the interval is 0, expand the interval to 125% to ensure that the data is not misjudged when the building normally fluctuates, and the expression of a final judgment interval S is as follows:
S＝[0,125％×X(α％)]
wherein S is an interval for judging an abnormal value;
X(α％)the alpha percentile is the value at which the energy consumption data are arranged according to size and alpha percent is accumulated;
data not in the section is determined as abnormal data, and the abnormal data is removed.
When the missing value is judged, comparing the complete time axis with the time axis of the historical data, and recording the time point without data as missing; recording the data with data 0 at all times in a whole day as missing; and under the condition that the continuous missing points are less than 5, completing by using a linear interpolation method, wherein the calculation formula is as follows:
a (A is less than or equal to 5) is the number of deletion points, Ya(a is 1,2, …, a) is the data value of the a-th missing point, Y0And YA+1The data immediately before and after the deletion sequence are shown respectively.
Preferably, the calculation of the correlation coefficient between the variables adopts pearson correlation coefficient calculation, and is represented by the following formula:
wherein r (X, Y) is the correlation coefficient of the variables X and Y;
cov (X, Y) is the covariance of X, Y;
var (X) is the variance of variable X, and Var (Y) is the variance of variable Y;
and when the absolute value of the correlation coefficient is larger than 0.3, setting the correlation between the variable and the building energy consumption, and determining the variable as an input variable.
Preferably, the genetic optimization algorithm is to optimize the ANN by using a genetic algorithm, and the genetic algorithm optimizes the ANN to be the multi-layered perceptron MLP, and the specific optimization steps are as follows:
firstly, initializing algorithm parameters, wherein the parameters needing to be initialized comprise: the number of hidden layers, the number of neurons of each hidden layer, an L2 regularization penalty coefficient alpha and an initial learning rate eta; determining the value range and the accuracy of the optimized parameters, and simultaneously setting the number N of population individuals to be 20, the cross probability Pc to be 80%, the variation probability Pm to be 10% and the maximum iteration number to be 30;
secondly, converting the decimal system into binary codes to generate a random generation initial population, wherein the code length l meets the following formula:
2l-1<(Lupper-Llower)/δ≤2l-1
in the formula, LupperAnd LlowerRespectively representing the upper limit and the lower limit of the value range of the decision variable, and delta represents the calculation accuracy;
using randomly generated 0/1 filled sequences, an initial population of 20 individuals was generated, each sequence being referred to as the individual chromosome;
and thirdly, calculating the individual fitness, wherein the adopted fitness criterion is the reciprocal of a prediction error, and the prediction error refers to the average absolute percentage error of the prediction result of the artificial neural network trained by the same training set data.
Fi＝1/MAPEi
In the formula, FiDenotes fitness, MAPE, of the ith individualiIs the absolute percentage error of the ith individual;
MAPE refers to the mean absolute percentage error, i is the individual number of the contemporary population;
refers to the predicted value of the model to the training data, ykThe real value of the energy consumption data is k, the number n of the samples in the training set is the total number of the samples;
fourthly, judging whether the maximum fitness reaches a set value or the cycle number reaches an upper limit, and stopping the cycle when the fitness is more than 50 or the cycle number reaches 30;
fifthly, selecting good individuals in the parent generation to enable good genes of the good individuals to be inherited into the next generation, wherein the fitness is proportionally distributed to the selected probability, and the probability is calculated according to the following formula
In the formula, PiRepresenting the probability that the ith individual is selected, FiRepresenting the fitness of the ith individual, and M representing the number of individuals; and randomly selecting individuals according to the probability to form a new population.
Sixthly, crossing and mutation, wherein the crossing is the part after randomly selecting two individuals and one node and exchanging the two chromosome nodes; the variation is that a point is randomly selected, the numerical value of the point is inverted, i.e. 0 is replaced by 1, and 1 is replaced by 0.
And seventhly, returning the new population generated in the sixth step to the third step again for circulation.
And eighthly, inputting the optimal individual after the fourth step of judging and completing the circulation, decoding the optimal individual into a decimal number again, and outputting the decimal number which is the model parameter of the optimized MLP artificial neural network.
Preferably, the average absolute percentage error is calculated by:
where MAPE refers to the mean absolute percent error;
refers to the predicted value, y, of the model to the jth test data setjThe real value of the building energy consumption of the jth test data set; j represents the number of the test set sample; upper limit e of the error allowable rangemaxIs composed of
emax＝200％×MAPE
And when the building energy consumption exceeds the upper error limit of the error allowable range, an abnormal alarm is given.
The invention provides a GA-ANN-based public building energy consumption prediction system, which comprises:
module S1: monitoring energy consumption to obtain energy consumption data of a public building, obtaining influence factor information, and preprocessing the energy consumption data and the influence factor information to obtain normalized data;
module S2: dividing the normalized data into a training set and a test set, wherein the training set can train a model, the test set can carry out prediction precision test on the model, calculate correlation coefficients among variables in the normalized data, and determine input variables of the model based on the magnitude of the correlation coefficients;
module S3: establishing an artificial neural network model by using a genetic optimization algorithm, and training the model by using a training set to obtain an energy consumption prediction model;
module S4: inputting the set input variable into the energy consumption prediction model to obtain a public building energy consumption prediction value corresponding to the output value;
module S5: and performing model verification on the public building energy consumption predicted value through a test set to obtain an average absolute percentage error, setting an error allowable range according to the average absolute percentage error, and jointly predicting and diagnosing the public building energy consumption based on the public building energy consumption predicted value and the error allowable range.
Compared with the prior art, the invention has the following beneficial effects:
1. the prediction speed is high, the public building energy consumption is predicted through GA-ANN modeling, the public building energy consumption prediction can be rapidly provided after training is completed, and the problems of long physical modeling period and large engineering quantity are solved.
2. The prediction precision is high, parameters of the artificial neural network are optimized through a genetic algorithm, the model performance is improved, and the energy consumption prediction precision of the public building is obviously improved.
3. The method is high in practicability, can be widely applied to energy consumption prediction of various public buildings, the building energy management strategy is optimized according to the prediction result, and meanwhile, the provided error analysis function can provide reference for building operation diagnosis.
Drawings
Other features, objects and advantages of the invention will become more apparent upon reading of the detailed description of non-limiting embodiments with reference to the following drawings:
FIG. 1 is a schematic diagram of a public building energy consumption prediction process according to the present invention;
FIG. 2 is a flow chart of genetic algorithm optimization of an artificial neural network;
fig. 3 is a diagram illustrating the effect of the present invention on energy consumption prediction in an office building.
Detailed Description
The present invention will be described in detail with reference to specific examples. The following examples will assist those skilled in the art in further understanding the invention, but are not intended to limit the invention in any way. It should be noted that it would be obvious to those skilled in the art that various changes and modifications can be made without departing from the spirit of the invention. All falling within the scope of the present invention.
The method and the system for predicting the energy consumption of the public building based on the GA-ANN have the advantages that through the complete processes of data preprocessing, variable selection, data set division, parameter optimization, model training, energy consumption prediction and error evaluation, the algorithm can predict the real-time and future energy consumption of the public building with high precision.
As shown in fig. 1, the method for predicting the energy consumption of the public building based on the GA-ANN according to the present invention includes:
step S1: monitoring energy consumption to obtain energy consumption data of a public building, obtaining influence factor information, and preprocessing the energy consumption data and the influence factor information to obtain normalized data;
step S2: dividing the normalized data into a training set and a test set, wherein the training set can train a model, the test set can carry out prediction precision test on the model, calculate correlation coefficients among variables in the normalized data, and determine input variables of the model based on the magnitude of the correlation coefficients; preferably, calculating a correlation coefficient among the variables, comparing to obtain a parameter with high energy consumption correlation, and determining the parameter as an input variable of the model;
step S3: establishing an artificial neural network model by using a genetic optimization algorithm, and training the model by using a training set to obtain an energy consumption prediction model; preferably, the performance of an Artificial Neural Network (ANN) model is related to a hidden layer structure, and an optimal model parameter is selected according to fitness by using a Genetic Algorithm (GA) to establish the artificial neural network model; then training the model by using the training set data to obtain an energy consumption prediction model;
step S4: inputting the set input variable into the energy consumption prediction model to obtain a public building energy consumption prediction value corresponding to the output value;
step S5: and performing model verification on the public building energy consumption predicted value through a test set to obtain an average absolute percentage error, setting an error allowable range according to the average absolute percentage error, and jointly predicting and diagnosing the public building energy consumption based on the public building energy consumption predicted value and the error allowable range.
Specifically, the influence factor information includes outdoor weather parameters and building operation information;
the outdoor weather parameters are from weather station historical data acquired by a network, and the building operation information is acquired by means of on-site investigation or similar building operation rules. Preferably, the historical energy consumption data of the public buildings used in the method is derived from an energy consumption monitoring system. The energy consumption monitoring system collects the total energy consumption, the branch energy consumption data of the illumination and socket, the air conditioning system, the power system, the special energy consumption and the like of the public building.
Specifically, by the preprocessing, determination and processing including an abnormal value and a missing value are performed.
When abnormal values are judged, the abnormal values are judged through interval inspection, the interval inspection is to sort the energy consumption data in size, use a percentile positioned at alpha% as an upper limit of an interval, the lower limit of the interval is 0, expand the interval to 125% to ensure that the data is not misjudged when the building normally fluctuates, and the expression of a final judgment interval S is as follows:
S＝[0,125％×X(α％)]
wherein S is an interval for judging an abnormal value;
X(α％)the alpha percentile is the value at which the energy consumption data are arranged according to size and alpha percent is accumulated; preferably, the default α size is 95, which correspondingly improves when data reliability is high.
Data not in the section is determined as abnormal data, and the abnormal data is removed.
When the missing value is judged, comparing the complete time axis with the time axis of the historical data, and recording the time point without data as missing; recording the data with data 0 at all times in a whole day as missing; and under the condition that the continuous missing points are less than 5, completing by using a linear interpolation method, wherein the calculation formula is as follows:
a (A is less than or equal to 5) is the number of deletion points, Ya(a is 1,2, …, a) is the data value of the a-th missing point, Y0And YA+1The data immediately before and after the deletion sequence are shown respectively.
Specifically, the correlation coefficient between the variables is calculated by using a pearson correlation coefficient, and is calculated by the following formula:
wherein r (X, Y) is the correlation coefficient of the variables X and Y;
cov (X, Y) is the covariance of X, Y;
var (X) is the variance of variable X, and Var (Y) is the variance of variable Y;
and when the absolute value of the correlation coefficient is larger than 0.3, setting the correlation between the variable and the building energy consumption, and determining the variable as an input variable. Preferably, the variables include building outdoor temperature, outdoor humidity, building operating conditions, previous hour energy consumption.
Specifically, when the genetic optimization algorithm is used, the genetic algorithm is used for optimizing the ANN, the genetic algorithm optimized ANN is a multilayer perceptron (MLP), the network model of the selected and used ANN is a multilayer perceptron (MLP), and the model parameters to be optimized include: implicit layer number, neuron number of each layer, L2 regularization penalty coefficient alpha and learning rate eta.
As shown in fig. 2, the specific optimization steps are as follows:
firstly, initializing algorithm parameters, wherein the parameters needing to be initialized comprise: the number of hidden layers, the number of neurons of each hidden layer, an L2 regularization penalty coefficient alpha and an initial learning rate eta; determining the value range and the accuracy of the optimized parameters, and simultaneously setting the number N of population individuals to be 20, the cross probability Pc to be 80%, the variation probability Pm to be 10% and the maximum iteration number to be 30; preferably, the value range and the accuracy of the optimized parameters are determined, the range of the number of the hidden layers is 1-4, and the accuracy is 1; the value range of the number of neurons in each layer is 2-200, and the precision is 1; l2 regularization penaltyCoefficient alpha in the range of 10-5～10-3Accuracy of 10-5(ii) a The learning rate ranges from 0.0001 to 0.01, and the accuracy is 0.0001. Meanwhile, the number N of population individuals is set to be 20, the cross probability Pc is set to be 80%, the variation probability Pm is set to be 10%, and the maximum iteration number is set to be 30.
Secondly, converting the decimal system into binary codes to generate a random generation initial population, wherein the code length l meets the following formula:
2l-1<(Lupper-Llower)/δ≤2l-1
in the formula, LupperAnd LlowerRespectively representing the upper limit and the lower limit of the value range of the decision variable, and delta represents the calculation accuracy;
using randomly generated 0/1 filled sequences, an initial population of 20 individuals was generated, each sequence being referred to as the individual chromosome;
thirdly, calculating individual fitness which is a standard for evaluating the quality of the chromosome and a basis for optimizing an algorithm; the fitness criterion is the reciprocal of the prediction error, wherein the prediction error refers to the average absolute percentage error of the prediction result of the artificial neural network trained by the same training set data.
Fi＝1/MAPEi
In the formula, FiDenotes fitness, MAPE, of the ith individualiIs the absolute percentage error of the ith individual;
MAPE refers to the mean absolute percentage error, i is the individual number of the contemporary population;
refers to the predicted value of the model to the training data, ykThe real value of the energy consumption data is k, the number n of the samples in the training set is the total number of the samples;
fourthly, judging whether the maximum fitness reaches a set value or the cycle number reaches an upper limit, and stopping the cycle when the fitness is more than 50 or the cycle number reaches 30;
and fifthly, selecting excellent individuals in the parent to enable the excellent genes of the excellent individuals to be inherited into the next generation, wherein the selection is based on the individual fitness, and the probability of selecting the individuals with higher individual fitness is higher. The fitness is proportionally distributed to the selected probability, and the probability is calculated as follows
In the formula, PiRepresenting the probability that the ith individual is selected, FiRepresenting the fitness of the ith individual, and M representing the number of individuals; and randomly selecting individuals according to the probability to form a new population.
Sixthly, crossing and mutation, wherein the crossing is the part after randomly selecting two individuals and one node and exchanging the two chromosome nodes; the variation is that a point is randomly selected, the numerical value of the point is inverted, i.e. 0 is replaced by 1, and 1 is replaced by 0.
And seventhly, returning the new population generated in the sixth step to the third step again for circulation.
And eighthly, inputting the optimal individual after the fourth step of judging and completing the circulation, decoding the optimal individual into a decimal number again, and outputting the decimal number which is the model parameter of the optimized MLP artificial neural network.
Specifically, the average absolute percentage error is calculated by the following formula:
where MAPE refers to the mean absolute percent error;
refers to the predicted value, y, of the model to the jth test data setjThe real value of the building energy consumption of the jth test data set; j represents the test setThe number of the book;
upper limit e of the error allowable rangemaxIs composed of
emax＝200％×MAPE
And when the building energy consumption exceeds the upper error limit of the error allowable range, an abnormal alarm is given. In practical application, the error of the obtained data exceeds the upper limit, the operation of the public building can be abnormal, and an alarm can be sent to the monitoring system.
The invention provides a GA-ANN-based public building energy consumption prediction system, which comprises:
module S1: monitoring energy consumption to obtain energy consumption data of a public building, obtaining influence factor information, and preprocessing the energy consumption data and the influence factor information to obtain normalized data;
module S2: dividing the normalized data into a training set and a test set, wherein the training set can train a model, the test set can carry out prediction precision test on the model, calculate correlation coefficients among variables in the normalized data, and determine input variables of the model based on the magnitude of the correlation coefficients;
module S3: establishing an artificial neural network model by using a genetic optimization algorithm, and training the model by using a training set to obtain an energy consumption prediction model;
module S4: inputting the set input variable into the energy consumption prediction model to obtain a public building energy consumption prediction value corresponding to the output value;
module S5: and performing model verification on the public building energy consumption predicted value through a test set to obtain an average absolute percentage error, setting an error allowable range according to the average absolute percentage error, and jointly predicting and diagnosing the public building energy consumption based on the public building energy consumption predicted value and the error allowable range.
The system for predicting the energy consumption of the public building based on the GA-ANN can be realized through the step flow of the method for predicting the energy consumption of the public building based on the GA-ANN. The GA-ANN based public building energy consumption prediction method can be understood as a preferred example of the GA-ANN based public building energy consumption prediction system by the technical personnel in the field.
According to the method, the real-time and future energy consumption of the public building is predicted with high precision aiming at the demand of high-precision prediction of the energy consumption of the public building and the goal of parameter optimization of the ANN algorithm based on the thought of predicting the energy consumption of the building based on the data base, a basis is provided for scientific management of building energy, the genetic algorithm is applied to parameter optimization of the artificial neural network, and the prediction precision of the model is effectively improved. The genetic algorithm is an optimization algorithm for searching the optimal solution in the model biological evolution process, and the global optimal solution is easier to obtain because the multiple individuals search together.
In a specific embodiment, the implementation is performed on the premise of the technical scheme of the invention, and a detailed implementation mode and a specific operation process are given by depending on historical data and an open source Python code of an office building in Shanghai. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention.
The method for predicting the energy consumption of the public building based on the GA-ANN comprises the following main implementation steps:
step S1, acquiring energy consumption data of the public building by using an energy consumption monitoring system, wherein the energy consumption monitoring system acquires four major sub-item energy consumption data of total energy consumption, illumination and sockets, an air conditioning system, a power system and special energy consumption of the public building, the data length is totally two years in the example, and the data of 3 months are used; the influence factors comprise outdoor meteorological parameters and building operation information, and the influence factors comprise the outdoor meteorological parameters and are acquired by the meteorological monitoring station data to acquire meteorological data of corresponding time points. Preprocessing the acquired data, including distinguishing and processing abnormal values and missing values, and finally performing data normalization;
specifically, the used historical energy consumption data of the public building are derived from an energy consumption monitoring system, and the outdoor meteorological parameters are obtained from the historical data of a meteorological station acquired by a network and comprise outdoor temperature and humidity, daily illumination, wind speed, rainfall and the like. The building operation information is obtained by on-site investigation or similar building operation rules; when an abnormal value is judged through pretreatment, the abnormal energy consumption is judged through interval inspection, the used interval obtaining method is that according to the fact that most of energy consumption is normal data, after the data are sorted from small to large, a percentile located at alpha% is used as an interval upper limit, an interval lower limit is 0, then the interval is expanded to 125% to ensure that the data are not misjudged when the building normally fluctuates, and the expression of a final judgment interval S is as follows:
S＝[0,125％×X(α％)]
wherein S is an interval for judging an abnormal value; x(α％)The index data is arranged according to the size, the value at the alpha% position is accumulated, the default alpha size is 95, and the data reliability is correspondingly improved when being high;
and judging the data not in the interval as abnormal data, and removing the data points. The missing value judging method is to compare a complete time axis with a time axis of historical data, and the time point without data is marked as missing; data with data 0 at all times throughout the day are marked as missing. And under the condition that the continuous missing points are less than 5, completing by using a linear interpolation method, wherein the calculation formula is as follows:
a (A is less than or equal to 5) is the number of deletion points, Ya(a is 1,2, …, a) is the data value of the a-th missing point, Y0And YA+1The data immediately before and after the deletion sequence are shown respectively.
Step S2, dividing a training set and a test set, wherein the training set is used for training the model, and the test set is used for testing the prediction accuracy of the model; in this example, the training set is the historical energy consumption in two months in summer, and the test set is a one-week continuous energy consumption sequence in the same summer. Calculating correlation coefficients among the variables, comparing to obtain parameters with high energy consumption correlation, and determining the parameters meeting the correlation standard as input variables of the model;
specifically, the correlation coefficient calculation method used in step S2 is a pearson correlation coefficient, and the calculation formula is:
where r (X, Y) is the variable X, the correlation coefficient of Y, Cov (X, Y) is the covariance of X, Y, and Var (X) is the variance of the variable X.
When the absolute value of the correlation coefficient is greater than 0.3, the variable is considered to be related to the energy consumption of the building, and the input variables determined in this example include the outdoor temperature of the building, the outdoor humidity, the operating state of the building, and the energy consumption of the previous hour.
Step S3, the performance of an Artificial Neural Network (ANN) model is related to a hidden layer structure, a Genetic Algorithm (GA) is used, optimal model parameters are selected according to fitness, the artificial neural network model is established, and the specific flow is given in detail in the following description; then, training a model by using training set data to obtain an energy consumption prediction model;
the selection of parameters for the ANN is optimized using genetic algorithms. The network model of the selected ANN is a multilayer perceptron (MLP), and the model parameters to be optimized are as follows: implicit layer number, number of neurons in each layer and learning rate eta. The specific optimization steps are as follows:
firstly, initializing algorithm parameters, wherein the parameters needing to be initialized comprise: the number of hidden layers, the number of neurons of each hidden layer, an L2 regularization penalty coefficient alpha and an initial learning rate eta; determining the value range and the accuracy of the optimized parameters, wherein the range of the number of hidden layers is 1-4, and the accuracy is 1; the value range of the number of neurons in each layer is 2-200, and the precision is 1; l2 regularization penalty factor alpha range of 10-5～10-3Accuracy of 10-5(ii) a The learning rate ranges from 0.0001 to 0.01, and the accuracy is 0.0001. Meanwhile, the number N of population individuals is set to be 20, the cross probability Pc is set to be 80%, the variation probability Pm is set to be 10%, and the maximum iteration number is set to be 30.
And secondly, converting the decimal system into binary codes to generate a random generation initial population. The code length l satisfies the following formula:
2l-1<(Lupper-Llower)/δ≤2l-1
in the formula, LupperAnd LlowerRespectively represent the value of the decision variableThe upper and lower limits of the range, δ, indicate the accuracy of the calculation.
An initial population of 20 individuals was then generated using randomly generated 0/1 filled sequences, each sequence being referred to as the individual chromosome.
And thirdly, calculating individual fitness which is a standard for evaluating the quality of the chromosome and a basis for optimizing an algorithm. The fitness criterion is the reciprocal of the prediction error, wherein the prediction error refers to the average absolute percentage error of the prediction result of the artificial neural network trained by the same training set data.
Fi＝1/MAPEi
In the formula, FiDenotes fitness, MAPE, of the ith individualiIs the absolute percentage error of the ith individual; MAPE refers to the mean absolute percentage error, i is the individual number of the contemporary population;
refers to the predicted value of the model to the training data, ykThe real value of the energy consumption data is k, the number n of the samples in the training set is the total number of the samples;
and fourthly, judging whether the maximum fitness reaches a set value or the cycle number reaches an upper limit. And stopping the circulation when the fitness is more than 50 or the circulation number reaches 30.
And fifthly, selecting excellent individuals in the father generation to enable the excellent genes to be inherited into the next generation. The selection is based on the individual fitness, and the probability of selecting the individuals with higher individual fitness is higher. Here, the selected probabilities are proportionally distributed according to the fitness, and the calculation formula of the probabilities is as follows
In the formula, PiRepresenting the probability that the ith individual is selected, FiIndicating the fitness of the ith individual.
Then according to the probability PiAnd randomly selecting individuals to form a new population.
Sixthly, crossing and mutation, wherein the crossing is the part after randomly selecting two individuals and one node and exchanging the two chromosome nodes; the variation is that a point is randomly selected, the numerical value of the point is inverted, i.e. 0 is replaced by 1, and 1 is replaced by 0. The probability of crossover and mutation was 80% and 10%, respectively.
And step seven, returning the new population generated in the step six to the step three again to perform the next cycle until the cycle meets the condition and is terminated.
And eighthly, inputting the optimal individual after the fourth step of judging and completing the circulation, decoding the optimal individual into a decimal number again, and outputting the decimal number which is the model parameter of the optimized MLP artificial neural network. In this example, model parameters after algorithm optimization are 1 hidden layer, 32 hidden neurons, a L2 regularization penalty coefficient is 0.0518, and an initial learning rate is 0.0171.
Step S4, inputting corresponding input variables of the period needing to be predicted, namely the input variable sequence in the test set, as the output of the trained model, thereby obtaining a public building energy consumption predicted value corresponding to the output value;
step S5, verifying the model in the test set, passing the average percent error (MAPE), and giving out an allowable error range, if the energy consumption error is larger than the upper error limit, giving out an alarm; the predicted value and the error range jointly provide a basis for monitoring and diagnosing the energy consumption of the public building. The selected model error index is the mean absolute percentage error MAPE, and the calculation formula is as follows:
in the formula (I), the compound is shown in the specification,
refers to the predicted value of the model to the test data, yjTo test the true value of energy consumption for the data set. After the prediction error value of the model is obtained, the allowable error upper limit is
emax＝200％×MAPE
If the error of the data obtained in the actual application exceeds the upper limit, the operation of the building can be abnormal.
The method is applied to a certain office building in Shanghai, energy consumption data of two months are used for predicting energy consumption of the next five days, and the result is shown in figure 3, so that the method can realize high-precision energy consumption prediction.
Those skilled in the art will appreciate that, in addition to implementing the systems, apparatus, and various modules thereof provided by the present invention in purely computer readable program code, the same procedures can be implemented entirely by logically programming method steps such that the systems, apparatus, and various modules thereof are provided in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers and the like. Therefore, the system, the device and the modules thereof provided by the present invention can be considered as a hardware component, and the modules included in the system, the device and the modules thereof for implementing various programs can also be considered as structures in the hardware component; modules for performing various functions may also be considered to be both software programs for performing the methods and structures within hardware components.
The foregoing description of specific embodiments of the present invention has been presented. It is to be understood that the present invention is not limited to the specific embodiments described above, and that various changes or modifications may be made by one skilled in the art within the scope of the appended claims without departing from the spirit of the invention. The embodiments and features of the embodiments of the present application may be combined with each other arbitrarily without conflict.
Claims (8)
1. A prediction method for energy consumption of public buildings based on GA-ANN is characterized by comprising the following steps:
step S1: monitoring energy consumption to obtain energy consumption data of a public building, obtaining influence factor information, and preprocessing the energy consumption data and the influence factor information to obtain normalized data;
step S2: dividing the normalized data into a training set and a test set, wherein the training set can train a model, the test set can carry out prediction precision test on the model, calculate correlation coefficients among variables in the normalized data, and determine input variables of the model based on the magnitude of the correlation coefficients;
step S3: establishing an artificial neural network model by using a genetic optimization algorithm, and training the model by using a training set to obtain an energy consumption prediction model;
step S4: inputting the set input variable into the energy consumption prediction model to obtain a public building energy consumption prediction value corresponding to the output value;
step S5: performing model verification on the public building energy consumption predicted value through a test set to obtain an average absolute percentage error, setting an error allowable range according to the average absolute percentage error, and jointly predicting and diagnosing public building energy consumption based on the public building energy consumption predicted value and the error allowable range;
optimizing the ANN by using a genetic algorithm when using the genetic optimization algorithm;
the genetic algorithm optimization ANN is a multilayer perceptron MLP, and the specific optimization steps are as follows:
firstly, initializing algorithm parameters, wherein the parameters needing to be initialized comprise: the number of hidden layers, the number of neurons of each hidden layer, an L2 regularization penalty coefficient alpha and an initial learning rate eta; determining the value range and the accuracy of the optimized parameters, and simultaneously setting the number M of population individuals to be 20, the cross probability Pc to be 80%, the variation probability Pm to be 10% and the maximum iteration number to be 30;
secondly, converting the decimal system into binary codes to generate a random generation initial population, wherein the code length l meets the following formula:
2l-1<(Lupper-Llower)/δ≤2l-1
in the formula, LupperAnd LlowerRespectively representing the upper limit and the lower limit of the value range of the decision variable, and delta represents the calculation accuracy;
using randomly generated 0/1 filled sequences, an initial population of 20 individuals was generated, each sequence being referred to as the individual chromosome;
calculating individual fitness, wherein the adopted fitness criterion is the reciprocal of a prediction error, and the prediction error refers to the average absolute percentage error of the prediction result of the artificial neural network trained by the same training set data;
Fi＝1/MAPEi
in the formula, FiDenotes fitness, MAPE, of the ith individualiIs the absolute percentage error of the ith individual; MAPE refers to the mean absolute percentage error, i is the individual number of the contemporary population;
refers to the predicted value of the model to the training data, ykThe real value of the energy consumption data, k is the number of the samples in the training set, and n is the total number of the samples;
fourthly, judging whether the maximum fitness reaches a set value or the cycle number reaches an upper limit, and stopping the cycle when the fitness is more than 50 or the cycle number reaches 30;
fifthly, selecting good individuals in the parent generation to enable good genes of the good individuals to be inherited into the next generation, wherein the fitness is proportionally distributed to the selected probability, and the probability is calculated according to the following formula
In the formula, PiRepresenting the probability that the ith individual is selected, FiRepresenting the fitness of the ith individual, and M representing the number of individuals; randomly selecting individuals according to the probability to form a new population;
sixthly, crossing and mutation, wherein the crossing is the part after randomly selecting two individuals and one node and exchanging the two chromosome nodes; the variation is that a point is randomly selected, the numerical value of the point is inverted, namely 0 is replaced by 1, and 1 is replaced by 0;
step seven, returning the new population generated in the step six to the step three again for circulation;
and eighthly, inputting the optimal individual after the fourth step of judging and completing the circulation, decoding the optimal individual into a decimal number again, and outputting the decimal number which is the model parameter of the optimized MLP artificial neural network.
2. A GA-ANN based public building energy consumption prediction method according to claim 1, wherein the influence factor information comprises outdoor weather parameters and building operation information;
the outdoor weather parameters are from weather station historical data acquired by a network, and the building operation information is acquired by means of on-site investigation or similar building operation rules.
3. A GA-ANN based public building energy consumption prediction method according to claim 1, wherein the determination and processing including an abnormal value and a missing value are performed by the preprocessing.
4. A GA-ANN based public building energy consumption prediction method according to claim 3,
when the abnormal value is judged, the abnormal value is judged through interval inspection, the interval inspection is to sort the energy consumption data in size, use a percentile positioned at alpha% as an upper limit of an interval, the lower limit of the interval is 0, expand the interval to 125% to ensure that the data is not misjudged when the building normally fluctuates, and the expression of an interval S for finally judging the abnormal value is as follows:
S＝[0,125％×X(α％)]
wherein S is an interval for judging an abnormal value;
X(α％)the alpha percentile is the value at which the energy consumption data are arranged according to size and alpha percent is accumulated;
data not in the section is determined as abnormal data, and the abnormal data is removed.
5. A GA-ANN based public building energy consumption prediction method according to claim 3,
when the missing value is judged, comparing the complete time axis with the time axis of the historical data, and recording the time point without data as missing; recording the data with data 0 at all times in a whole day as missing; and under the condition that the continuous missing points are less than 5, completing by using a linear interpolation method, wherein the calculation formula is as follows:
a (A is less than or equal to 5) is the number of deletion points, Ya(a is 1,2, …, a) is the data value of the a-th missing point, Y0And YA+1The data immediately before and after the deletion sequence are shown respectively.
6. A GA-ANN based public building energy consumption prediction method according to claim 1, wherein the correlation coefficient between variables is calculated using pearson's correlation coefficient by the following formula:
wherein r (X, Y) is the correlation coefficient of the variables X and Y;
cov (X, Y) is the covariance of X, Y;
var (X) is the variance of variable X, and Var (Y) is the variance of variable Y;
and when the absolute value of the correlation coefficient is larger than 0.3, setting the correlation between the variable and the building energy consumption, and determining the variable as an input variable.
7. A GA-ANN based public building energy consumption prediction method according to claim 1, wherein the average absolute percentage error is calculated by:
where MAPE refers to the mean absolute percent error;
refers to the predicted value, y, of the model to the jth test data setjThe real value of the building energy consumption of the jth test data set; j represents the number of the test set sample;
upper limit e of the error allowable rangemaxIs composed of
emax＝200％×MAPE
And when the building energy consumption exceeds the upper error limit of the error allowable range, an abnormal alarm is given.
8. A GA-ANN-based public building energy consumption prediction system is characterized by comprising:
module S1: monitoring energy consumption to obtain energy consumption data of a public building, obtaining influence factor information, and preprocessing the energy consumption data and the influence factor information to obtain normalized data;
module S2: dividing the normalized data into a training set and a test set, wherein the training set can train a model, the test set can carry out prediction precision test on the model, calculate correlation coefficients among variables in the normalized data, and determine input variables of the model based on the magnitude of the correlation coefficients;
module S3: establishing an artificial neural network model by using a genetic optimization algorithm, and training the model by using a training set to obtain an energy consumption prediction model;
module S4: inputting the set input variable into the energy consumption prediction model to obtain a public building energy consumption prediction value corresponding to the output value;
module S5: performing model verification on the public building energy consumption predicted value through a test set to obtain an average absolute percentage error, setting an error allowable range according to the average absolute percentage error, and jointly predicting and diagnosing public building energy consumption based on the public building energy consumption predicted value and the error allowable range;
optimizing the ANN by using a genetic algorithm when using the genetic optimization algorithm;
the genetic algorithm optimization ANN is a multilayer perceptron MLP, and the specific optimization steps are as follows:
firstly, initializing algorithm parameters, wherein the parameters needing to be initialized comprise: the number of hidden layers, the number of neurons of each hidden layer, an L2 regularization penalty coefficient alpha and an initial learning rate eta; determining the value range and the accuracy of the optimized parameters, and simultaneously setting the number M of population individuals to be 20, the cross probability Pc to be 80%, the variation probability Pm to be 10% and the maximum iteration number to be 30;
secondly, converting the decimal system into binary codes to generate a random generation initial population, wherein the code length l meets the following formula:
2l-1<(Lupper-Llower)/δ≤2l-1
in the formula, LupperAnd LlowerRespectively representing the upper limit and the lower limit of the value range of the decision variable, and delta represents the calculation accuracy;
using randomly generated 0/1 filled sequences, an initial population of 20 individuals was generated, each sequence being referred to as the individual chromosome;
calculating individual fitness, wherein the adopted fitness criterion is the reciprocal of a prediction error, and the prediction error refers to the average absolute percentage error of the prediction result of the artificial neural network trained by the same training set data;
Fi＝1/MAPEi
in the formula, FiDenotes fitness, MAPE, of the ith individualiIs the absolute percentage error of the ith individual; MAPE refers to the mean absolute percentage error, i is the individual number of the contemporary population;
refers to the predicted value of the model to the training data, ykThe real value of the energy consumption data, k is the serial number of the samples in the training set, and n is the total number of the samples for calculating the error;
fourthly, judging whether the maximum fitness reaches a set value or the cycle number reaches an upper limit, and stopping the cycle when the fitness is more than 50 or the cycle number reaches 30;
fifthly, selecting good individuals in the parent generation to enable good genes of the good individuals to be inherited into the next generation, wherein the fitness is proportionally distributed to the selected probability, and the probability is calculated according to the following formula
In the formula, PiRepresenting the probability that the ith individual is selected, FiRepresenting the fitness of the ith individual, and M representing the number of individuals; randomly selecting individuals according to the probability to form a new population;
sixthly, crossing and mutation, wherein the crossing is the part after randomly selecting two individuals and one node and exchanging the two chromosome nodes; the variation is that a point is randomly selected, the numerical value of the point is inverted, namely 0 is replaced by 1, and 1 is replaced by 0;
step seven, returning the new population generated in the step six to the step three again for circulation;
and eighthly, inputting the optimal individual after the fourth step of judging and completing the circulation, decoding the optimal individual into a decimal number again, and outputting the decimal number which is the model parameter of the optimized MLP artificial neural network.
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 CN110046743A (en) | 2019-07-23 |
| CN110046743B true 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) |
Families Citing this family (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 |
| CN110988556B (en) * | 2019-12-20 | 2022-03-29 | 上海市建筑科学研究院有限公司 | Diagnosis method for architectural lighting socket support in non-operation period |
| 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 |
| CN111597609B (en) * | 2020-04-28 | 2023-03-24 | 马辰龙 | Basic operation unit containing solar radiation and building energy consumption rapid simulation method applying same |
| CN111766832B (en) * | 2020-06-29 | 2022-05-31 | 重庆大学 | Numerical control machine tool cutting energy consumption prediction modeling method driven by incomplete data |
| 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 |
| CN112712213B (en) * | 2021-01-15 | 2023-07-04 | 上海交通大学 | Method and system for predicting deep migration learning energy consumption of concentrated 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 |
| 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 |
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 |
|---|
| "A zone-level，building energy optimisation combining an artificial neural network，a genetic algorithm，and model predictive control";Jonathan Reynolds等;《Energy》;20180320;第1-11页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110046743A (en) | 2019-07-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110046743B (en) | Public building energy consumption prediction method and system based on GA-ANN | |
| CN112633604B (en) | Short-term power consumption prediction method based on I-LSTM | |
| CN109063366A (en) | A kind of building performance online data preprocess method based on time and spatial weighting | |
| CN111310968A (en) | LSTM neural network circulation hydrological forecasting method based on mutual information | |
| CN109376960A (en) | Load Forecasting based on LSTM neural network | |
| CN113205203A (en) | CNN-LSTM-based building energy consumption prediction method and system | |
| CN110084398A (en) | A kind of Industrial Cycle self-adapting detecting method based on enterprise's electric power big data | |
| CN113505923A (en) | Regional power grid short-term load prediction method and system | |
| CN114819102A (en) | GRU-based air conditioning equipment fault diagnosis method | |
| CN112733307A (en) | Urban natural gas load prediction method and device in heating season | |
| CN114862032B (en) | XGBoost-LSTM-based power grid load prediction method and device | |
| CN112990587A (en) | Method, system, equipment and medium for accurately predicting power consumption of transformer area | |
| CN113762591B (en) | Short-term electric quantity prediction method and system based on GRU and multi-core SVM countermeasure learning | |
| CN113268929B (en) | Short-term load interval prediction method and device | |
| Gao et al. | A multifactorial framework for short-term load forecasting system as well as the jinan’s case study | |
| Wang et al. | A transformer-based multi-entity load forecasting method for integrated energy systems | |
| CN116245259B (en) | Photovoltaic power generation prediction method and device based on depth feature selection and electronic equipment | |
| CN112766548A (en) | Order completion time prediction method based on GASA-BP neural network | |
| CN112767692A (en) | Short-term traffic flow prediction system based on SARIMA-GA-Elman combined model | |
| CN117350146A (en) | GA-BP neural network-based drainage pipe network health evaluation method | |
| CN111310974A (en) | Short-term water demand prediction method based on GA-ELM | |
| CN115481788B (en) | Phase change energy storage system load prediction method and system | |
| CN117033923A (en) | Method and system for predicting crime quantity based on interpretable machine learning | |
| CN114065646B (en) | Energy consumption prediction method based on hybrid optimization algorithm, cloud computing platform and system | |
| CN113191526A (en) | Short-term wind speed interval multi-objective optimization prediction method and system based on random sensitivity |
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 |