CN106991504A - Building energy consumption Forecasting Methodology, system and building based on metering separate time series - Google Patents
Building energy consumption Forecasting Methodology, system and building based on metering separate time series Download PDFInfo
- Publication number
- CN106991504A CN106991504A CN201710321447.2A CN201710321447A CN106991504A CN 106991504 A CN106991504 A CN 106991504A CN 201710321447 A CN201710321447 A CN 201710321447A CN 106991504 A CN106991504 A CN 106991504A
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- time series
- building
- data
- forecasting methodology
- 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
- 238000005265 energy consumption Methods 0.000 title claims abstract description 140
- 238000000034 method Methods 0.000 title claims abstract description 65
- 238000003062 neural network model Methods 0.000 claims abstract description 13
- 238000012731 temporal analysis Methods 0.000 claims abstract description 12
- 238000000700 time series analysis Methods 0.000 claims abstract description 12
- 238000013528 artificial neural network Methods 0.000 claims abstract description 10
- 238000010219 correlation analysis Methods 0.000 claims abstract description 4
- 230000006870 function Effects 0.000 claims description 27
- 238000012549 training Methods 0.000 claims description 23
- 230000005611 electricity Effects 0.000 claims description 17
- 238000005311 autocorrelation function Methods 0.000 claims description 16
- 238000012544 monitoring process Methods 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 12
- 230000008569 process Effects 0.000 claims description 11
- 238000004891 communication Methods 0.000 claims description 8
- 238000004378 air conditioning Methods 0.000 claims description 7
- 210000002569 neuron Anatomy 0.000 claims description 4
- 238000010606 normalization Methods 0.000 claims description 4
- 210000004205 output neuron Anatomy 0.000 claims description 4
- 230000005540 biological transmission Effects 0.000 claims description 3
- 238000004364 calculation method Methods 0.000 claims description 3
- 238000006243 chemical reaction Methods 0.000 claims description 3
- 210000002364 input neuron Anatomy 0.000 claims description 3
- 230000007935 neutral effect Effects 0.000 claims description 3
- 238000005096 rolling process Methods 0.000 claims description 3
- 230000005654 stationary process Effects 0.000 claims description 3
- 238000012795 verification Methods 0.000 claims description 3
- 238000012806 monitoring device Methods 0.000 claims 1
- 238000004422 calculation algorithm Methods 0.000 abstract description 11
- 238000010586 diagram Methods 0.000 description 8
- 230000000694 effects Effects 0.000 description 6
- 230000001537 neural effect Effects 0.000 description 5
- 230000008901 benefit Effects 0.000 description 3
- 238000012886 linear function Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 210000005036 nerve Anatomy 0.000 description 2
- 238000003909 pattern recognition Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 206010019133 Hangover Diseases 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000003466 anti-cipated effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000003750 conditioning effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000012937 correction Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 239000006185 dispersion Substances 0.000 description 1
- 238000004134 energy conservation Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 239000000843 powder Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/18—Complex mathematical operations for evaluating statistical data, e.g. average values, frequency distributions, probability functions, regression analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- 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)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- General Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Tourism & Hospitality (AREA)
- Pure & Applied Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Computational Mathematics (AREA)
- Operations Research (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Evolutionary Computation (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Biophysics (AREA)
- Evolutionary Biology (AREA)
- Biomedical Technology (AREA)
- Artificial Intelligence (AREA)
- Probability & Statistics with Applications (AREA)
- Primary Health Care (AREA)
- Algebra (AREA)
Abstract
The invention discloses building energy consumption Forecasting Methodology, system and the building based on metering separate time series, wherein, Forecasting Methodology includes, and gathers data and the storage of the energy consumption and temperature of building；The energy consumption and temperature data of storage will be gathered as the input parameter of Time series analysis method；According to metering separate and correlation analysis, Time series analysis method is predicted to energy consumption and temperature trend and time factor as the major influence factors of building energy consumption；Using the main factor to affect established and the energy consumption of collection as the parameter in the BP neural network model established, to predict the energy consumption of future architecture.Because BP neural network learning efficiency is low, convergence rate is slow, it is more sensitive to parameter selection, metering separate and the building energy consumption prediction algorithm of time series are added on the basis of BP neural network, the accuracy of energy consumption prediction can be substantially increased, shorten the time of prediction so that the data predicted are more accurate.
Description
Technical field
The present invention relates to a kind of Forecasting Methodology of building energy consumption, belong to building energy consumption electric powder prediction, in particular relate to
A kind of system based on metering separate time series building energy consumption Forecasting Methodology, the such method of implementation, and be equipped with so
System building.
Background technology
With the continuous acceleration of urbanization process, energy problem becomes increasingly conspicuous.Building energy consumption is shared in social total energy consumption
Ratio rises to 28% from the 10% of eighties of last century late nineteen seventies.Country's office office building and all kinds of public buildings year energy consumption
Electricity accounts for the 22% of national cities and towns total energy consumption.Unit area power consumption is 10~20 times of ordinarily resident's house, is Europe, day
1.5~2 times of the similar building of this grade developed country.The power consumption of large public building exceedes society's building total electricity consumption
30%, it is the main object of construction energy conservation monitoring and transformation.Therefore, the prediction to the energy consumption of large public building seems particularly heavy
Will, the foundation of science can be so provided for the electricity consumption amount of large public building.
Energy for building is huge, especially in megastore, laboratory, office building etc..At present for energy supply in building
The energy consumption data subitem collection of equipment is to understand building energy consumption size, finds energy consumption and wastes point premise.And as national energy-saving subtracts
Row's policy is implemented, and each department are gathered to the energy consumption data much built, but predominantly realize the metering separate of energy consumption
With the statistics displaying of subitem energy consumption data.Meanwhile, current collecting method focuses mostly on to be adopted in itself to energy consumption data
Collection, gathers less to energy consumption factor data.In addition, these energy consumption data acquisition equipment are generally regularly to each item data of equipment
Measure, read, do not focus on time and energy consumption that some equipment are excessively consumed from a state to another state.
But and understand these information, opening time of distinct device, opening order could be planned, searching optimal scheduling side
Case.
The content of the invention
The purpose of this part is some aspects for summarizing embodiments of the invention and briefly introduces some preferably implementations
Example.It may do a little simplified or be omitted to avoid making our department in this part and the description of the present application summary and denomination of invention
Point, the purpose of specification digest and denomination of invention obscure, and this simplification or omit and cannot be used for limiting the scope of the present invention.
In view of above-mentioned and/or existing based on metering separate time series building energy consumption Forecasting Methodology, based on this method foundation
System and using the system building present in problem, it is proposed that the present invention.
Therefore, one of purpose of the invention is to provide a kind of based on metering separate time series building energy consumption prediction side
Method.
In order to solve the above technical problems, the present invention provides following technical scheme：One kind is built based on metering separate time series
Energy consumption Forecasting Methodology is built, it includes, gather data and the storage of the energy consumption and temperature of building；The energy consumption and temperature of storage will be gathered
Degrees of data as Time series analysis method input parameter；According to metering separate and correlation analysis, by time series analysis
Method predicts energy consumption and temperature trend and time factor as the major influence factors of building energy consumption；It is main by what is established
Factor to affect and the energy consumption of collection are as the parameter in the BP neural network model established, to predict the energy of future architecture
Consumption.
As a kind of preferred scheme of the present invention based on metering separate time series building energy consumption Forecasting Methodology, its
In：Data and the storage of the energy consumption and temperature of the collection building, it is carried out by energy consumption data collecting system, the energy consumption
Data collecting system includes, measuring layer, including energy consumption measure collecting device and temperature monitoring equipment, and the energy consumption measure collection is set
It is standby that the electric consumption on lighting of building, power electricity consumption, air conditioning electricity and the energy consumption of special electricity consumption are acquired, the temperature monitoring equipment
Space temperature is acquired；Communication layers, set up writing to each other between measuring layer and management level；And, management level send number
Stored according to acquisition instructions and to the corresponding energy consumption and temperature data of collection；Wherein, the management level send data acquisition and referred to
Order, after the communication Protocol Conversion of the communication layers, the energy consumption measure collecting device and temperature monitoring for reaching the measuring layer are set
Standby, the energy consumption measure collecting device and temperature monitoring equipment receive instruction and responded after verification, by corresponding energy consumption with
Temperature data feeds back to the management level, is stored by handling subitem to database.
As a kind of preferred scheme of the present invention based on metering separate time series building energy consumption Forecasting Methodology, its
In：It is described will gather storage energy consumption and temperature data as Time series analysis method input parameter, wherein, the time
Sequence is the auto-correlation function and partial autocorrelation function for calculating the energy consumption and temperature data.
As a kind of preferred scheme of the present invention based on metering separate time series building energy consumption Forecasting Methodology, its
In：
The auto-correlation coefficient, it is defined：
Because having for a stationary process：
So
As k=0, there is ρ0=1, the auto-correlation coefficient row ρ by variable of lag period kkK=0,1,2 ... it is referred to as auto-correlation
Function.
As a kind of preferred scheme of the present invention based on metering separate time series building energy consumption Forecasting Methodology, its
In：The partial autocorrelation function to describe random process architectural feature, wherein,
Use φkjJ-th of regression coefficient in k rank autoregressive process is represented, then k ranks autoregression model is expressed as：xt=φk1xt-1+φk2xt-2+...+φkkxt-k+ut, wherein φkkIt is last regression coefficient；
If φkkRegard lag period k function as, then claim φkk, k=1,2 ... it is partial autocorrelation function；
If time series { yt, the item number for taking rolling average is n, then t+1 phase predicted values calculation formula
For：
Y in above formulatRepresent t phase actual values；Represent t phase Single moving average numbers；yt+1Represent t+1 phase predicted values
(t≥n)。
As a kind of preferred scheme of the present invention based on metering separate time series building energy consumption Forecasting Methodology, its
In：The BP neural network model, wherein,
According to the data processing feature of neutral net, input data is normalized, using premnmx function handles
Training sample is normalized between [0,1], and method is as follows：
Wherein, x, x ' are the forward and backward value of normalization, xmaxIt is maximum in sample, xminIt is minimum value in sample.
As a kind of preferred scheme of the present invention based on metering separate time series building energy consumption Forecasting Methodology, its
In：The BP neural network model, it, which sets up process, includes,
Establishment, training, emulation, prediction, the renormalization processing of network；Wherein,
The establishment of the network, its type selecting " feed-forward backprop "；
The training function uses trainlm；
After network output result is handled through renormalization, you can obtain the data of prediction of energy consumption；
The energy consumption data predicted using mapminmax function pairs carries out renormalization processing.
As a kind of preferred scheme of the present invention based on metering separate time series building energy consumption Forecasting Methodology, its
In：The BP neural network model, includes the network of 3 input neurons, 9 hidden neurons and 4 output neurons；Net
The neural transferring function of network hidden layer is using being tansig functions, and it is linear function that output layer neural transferring function, which is used,
purelin。
It is a further object to provide one kind using based on metering separate time series building energy consumption Forecasting Methodology
The system of foundation.
In order to solve the above technical problems, the present invention provides following technical scheme：A kind of energy expenditure for predicting building
Forecasting system, wherein, the forecasting system includes control unit, and the control unit is implemented to build energy based on metering separate time series
Consume Forecasting Methodology.
It is also another object of the present invention to provide a kind of building, it includes implementing based on metering separate time series building
The forecasting system of the energy expenditure of the prediction building of energy consumption Forecasting Methodology.
Beneficial effects of the present invention：One kind that the present invention is provided is based on metering separate time series building energy consumption prediction side
Method, because BP neural network learning efficiency is low, convergence rate is slow, more sensitive to parameter selection, on the basis of BP neural network
It is upper to add metering separate and the building energy consumption prediction algorithm of time series, the accuracy of energy consumption prediction, contracting can be substantially increased
The time of short prediction so that the data predicted are more accurate.
Brief description of the drawings
In order to illustrate the technical solution of the embodiments of the present invention more clearly, being used required in being described below to embodiment
Accompanying drawing be briefly described, it should be apparent that, drawings in the following description are only some embodiments of the present invention, for this
For the those of ordinary skill of field, without having to pay creative labor, it can also obtain other according to these accompanying drawings
Accompanying drawing.Wherein：
Fig. 1 is energy consumption data acquisition monitoring platform system schematic of the present invention；
Fig. 2 is that energy consumption data acquisition of the present invention predicts schematic flow sheet；
Fig. 3 is stationary time series model flow schematic diagram of the present invention；
Fig. 4 is that the present invention is that the energy consumption prediction BP neural network model that Nanjing mansion is set up shows in one embodiment
It is intended to；
Fig. 5 predicts the training result schematic diagram of BP neural network model for energy consumption described in Fig. 4 of the present invention；
Fig. 6 is that mansion lighting energy consumption in Nanjing predicts the outcome schematic diagram in Fig. 4 illustrated embodiments of the present invention；
Fig. 7 is mansion Energy consumption forecast for air conditioning result schematic diagram in Nanjing in Fig. 4 illustrated embodiments of the present invention；
Fig. 8 is that mansion power consumption in Nanjing predicts the outcome schematic diagram in Fig. 4 illustrated embodiments of the present invention；
Fig. 9 is mansion special energy consumption prediction result schematic diagram in Nanjing in Fig. 4 illustrated embodiments of the present invention；
Figure 10 is the absolute error signal of the sport energy consumption prediction of Nanjing mansion four in Fig. 4 illustrated embodiments of the present invention
Figure；
Figure 11 is the relative error signal of the sport energy consumption prediction of Nanjing mansion four in Fig. 4 illustrated embodiments of the present invention
Figure；
Figure 12 is the mean absolute error schematic diagram of mansion energy consumption prediction in Nanjing in Fig. 4 illustrated embodiments of the present invention；
Figure 13 is the root-mean-square error schematic diagram of mansion energy consumption prediction in Nanjing in Fig. 4 illustrated embodiments of the present invention.
Embodiment
In order to facilitate the understanding of the purposes, features and advantages of the present invention, with reference to Figure of description pair
The embodiment of the present invention is described in detail.
Many details are elaborated in the following description to facilitate a thorough understanding of the present invention, still the present invention can be with
It is different from other manner described here using other to implement, those skilled in the art can be without prejudice to intension of the present invention
In the case of do similar popularization, therefore the present invention is not limited by following public specific embodiment.
Secondly, " one embodiment " or " embodiment " referred to herein refers to may be included at least one realization side of the invention
Special characteristic, structure or characteristic in formula." in one embodiment " that different places occur in this manual not refers both to
Same embodiment, nor the single or selective embodiment mutually exclusive with other embodiment.
Step one：As shown in figure 1, to obtain prediction data sample, energy consumption monitoring platform, energy consumption data acquisition system need to be set up
System is using the design of 3 layer architectures.
Energy consumption data collecting system is divided into situ metrology layer 100, Web communication layer 200 and management level 300.
The computer of management level 300 sends data acquisition instructions, after the protocol conversion of Web communication layer 200, reaches live meter
Measure each energy consumption measure collecting device 101 of layer 100 and temperature monitoring equipment 102.
The temperature monitoring equipment 102 of energy consumption measure collecting device 101 receives instruction and responded after verification, by corresponding energy
Consumption and temperature data feed back to the computer of management level 300, are stored by handling subitem to database.
Forecast model is by calling database to obtain training sample.
Step 2：As shown in Fig. 2 energy consumption monitoring platform gathers the electric consumption on lighting of building, power electricity consumption, idle call respectively
Electric, special electricity consumption and data and the storage of temperature.
Energy consumption monitoring platform is gathered to the electric consumption on lighting and temperature data of storage as the input of Time series analysis method
Parameter predicts following electric consumption on lighting energy consumption and future temperature trend.
According to metering separate and correlation analysis, Time series analysis method is predicted into electric consumption on lighting energy consumption and temperature becomes
Gesture and time factor as building energy consumption major influence factors.
The power electricity consumption that is gathered respectively by the main factor to affect established and by energy consumption monitoring platform, air conditioning electricity and
Special electricity consumption is as the parameter in the BP network models established, to predict electric consumption on lighting, power electricity consumption, the sky of future architecture
Call this four sports energy consumption of electric, special electricity consumption.
Step 3：The lighting energy consumption of building is predicted using Time series analysis method, in time series analysis, first
It is the stability for differentiating time series.
Time series is smoothly can just to calculate the auto-correlation function autocorr and partial autocorrelation function of these data
parcorr。
Auto-correlation coefficient is defined：
Because having for a stationary process：
So
As k=0, there is ρ0=1, the auto-correlation coefficient row ρ by variable of lag period kkK=0,1,2 ... it is referred to as auto-correlation
Function.
Partial autocorrelation function is another saying for describing random process architectural feature.
Use φkjJ-th of regression coefficient in k rank autoregressive process is represented, then k ranks autoregression model is expressed as：xt=φk1xt-1+φk2xt-2+...+φkkxt-k+ut, wherein φkkIt is last regression coefficient.
If φkkRegard lag period k function as, then claim φkk, k=1,2 ... it is partial autocorrelation function.
If time series { yt, the item number for taking rolling average is n, then t+1 phase predicted values calculation formula
For：
Y in above formulatRepresent t phase actual values；Represent t phase Single moving average numbers；yt+1Represent t+1 phase predicted values
(t≥n)。
As shown in figure 3, carrying out pattern-recognition first, judge that this model, for stationary time series model, is further joined
Number estimation, it is tail long in tow to draw auto-correlation function, and numerical value is gradually reduced, and meets hangover；And partial autocorrelation function
It is to converge to suddenly in the range of threshold levels, numerical value becomes very little suddenly, meets truncation, and n=3, further carries out mould
Type is diagnosed meets autoregression AR (3) model with test and judge this time series model, is predicted in the case that this model is desirable,
Such model is inadvisable, comes back to pattern recognition step.
Step 4：The processing of major influence factors and sample data
According to the data processing feature of neutral net, it is necessary to which input data is normalized, using premnmx letters
Several that training sample is normalized between [0,1], method is as follows：
Wherein, x, x ' are the forward and backward value of normalization, xmaxIt is maximum in sample, xminIt is minimum value in sample.
Step 5：Lighting energy consumption value will be predicted in time series models is used for the prediction of BP networks.
BP neural network model is set up, its process includes establishment, training, emulation, prediction, renormalization processing of network etc.
5 steps.
1st, the establishment of network
Implement in operation, by taking the mansion of Nanjing as an example：The establishment of network is operation circle in MATLAB artificial neural networks
On face, creating one includes the network of 3 input neurons, 9 hidden neurons and 4 output neurons.Network type is selected
Select " feed-forward backprop ", training function uses trainlm, and the learning algorithm of the function is back-propagation algorithm,
It the advantage is that convergence rate quickly.It is tansig functions, output layer nerve that the neural transferring function of network hidden layer, which is used,
First transmission function is using being linear function purelin, and the BP neural network model after establishment is as shown in Figure 4.
2nd, the training of network
To improve the accuracy of prediction, network need to can just be applied to the prediction of energy consumption after training.Present networks
It is trained using preceding 250 groups of data as sample data, training result is as shown in Figure 5.As can be seen that the training process of network
Restrain quickly, after 18 steps, network has reached requirement, available for predicting.
3rd, the emulation and prediction of network
Approximation effect of the network after training for input signal, therefore, the network trained can be detected by training
Also need to be emulated.For the mansion, its 9 groups of lighting energy consumption in July 31 day for predicting time series models, and No. 31
The temperature and time factor these three influence factors corresponding to this 9 groups of lighting energy consumption moment are used for the BP network models established
In, to predict No. 31 this 9 group of four sport energy consumptions.Obtain predicting the outcome as shown in Fig. 6~9 for four sports.Can be with by analysis
Know, Fig. 6 is due to training match value closely desired value, and match value and desired value are substantially overlapping on one wire in figure, and
The prediction effect of Fig. 7~9 can reflect greatly the change of energy consumption, simply have the deviation at indivedual moment larger, generally speaking, whole model
Prediction effect or highly desirable.
9 group of four sport energy consumption of this day is predicted using sample data, the output result of energy consumption prediction is as shown in table 1,
Table 1 is network output result.
Table 1
4th, renormalization is handled
After network output result is handled through renormalization, you can obtain the data of prediction of energy consumption；
The energy consumption data predicted using mapminmax function pairs carries out renormalization processing, draws 9 groups of energy on July 31
The result for consuming renormalization processing is as shown in table 2.First 41 groups are the predicted value and reality of the four sport energy consumptions obtained after emulation in table 2
Actual value correction data, and absolute error value, the last 9 groups energy consumption datas on July 31 for prediction.
Table 2
5th, error assessment
For the precision of prediction and stability of valuation prediction models on the whole, there is employed herein absolute error E, it is relative by mistake
Poor RE, mean absolute error MAE and tetra- performance indications of root-mean-square error RMSE are estimated to forecast model.
In formula n represent sample number, i represent forecasting sequence,Represent that the predicted value of the i-th sequence, X (i) represent the i-th sequence
The actual value of row.
Absolute error reflects the size of measured value deviation true value, relative error more can reflected measurement confidence level, it is average
Absolute error preferably reflects the actual conditions of predicted value error, and root-mean-square error reflects the dispersion degree of error distribution,
RMSE is bigger, and error sequence is more discrete, and prediction effect is poorer.Four error criterions are all smaller, and prediction effect is better.
Emulate the absolute error predicted and relative error as shown in Figure 10 and Figure 11, mean absolute error and root mean square
Error is as shown in Figure 12 and Figure 13.It can be drawn from Figure 10,11：Air conditioning energy consumption absolute error compares larger, prediction effect
Good without other three, other three energy consumption errors are fluctuated in 0 value, and Figure 12,13 are it can be seen that error has individually
Deviation is larger, but overall or more satisfactory.It is -0.0169 to emulate obtained mean absolute error maximum, and root mean square is missed
The maximum of difference is 8.3176, and error is all smaller, can carry out neural network forecast.
As can be seen from Table 2：July 31 energy consumption predict the outcome or it is more satisfactory.This is due to that time series is calculated
Method does not need complicated factor of influence, only needs the actual value of previous moment, compensate for the complicated of BP neural network algorithm needs
The shortcoming of factor of influence, has obtained following lighting energy consumption value that can not be measured so that network can Fast Training, and reach quickly
To training objective, and metering separate can preferably see the value of each single item energy consumption clearly.As can be seen here set forth herein metering separate
Time Serial Neural Network algorithm can meet requirement of the short-term forecast to real-time.Short-term forecast does not require nothing more than fast prediction,
And need to meet certain precision, reach the requirement of engineer applied.Ensuing electric energy consumption is entered using the network trained
Row prediction, is predicted ideal.All in all, the precision of prediction of metering separate Time Serial Neural Network algorithm is higher, maximum
Mean absolute error is 0.0245, minimum average B configuration absolute error 0.0001.In summary based on metering separate time series nerve
Network Prediction Model is applied to building energy consumption short-term forecast.
" feed-forward backprop ", training function uses trainlm, the study of the function for network type selection
Algorithm is back-propagation algorithm, the advantage is that convergence rate quickly.
The neural transferring function of network hidden layer is using being tansig functions, and output layer neural transferring function, which is used, is
Linear function purelin.
After network output result is handled through renormalization, you can obtain the data of prediction of energy consumption；
The energy consumption data predicted using mapminmax function pairs carries out renormalization processing.
Because time series algorithm does not need complicated factor of influence, the actual value of previous moment is only needed, BP god is compensate for
The shortcoming of the complicated factor of influence needed through network algorithm, has obtained following lighting energy consumption value that can not be measured so that net
Network can Fast Training, and quickly reach training objective, and the value of each single item energy consumption can be better anticipated out in metering separate.
The invention further relates to a kind of forecasting system for the energy expenditure for predicting building, it includes a control unit, should
Control unit implements the Forecasting Methodology of the energy expenditure of foregoing prediction building.The pre- of the energy expenditure that thing is built is built in prediction
Examining system includes illumination and/or power and/or special electricity consumption and/or heat supply and/or air-conditioning equipment and control unit, the control
Unit can adjust the internal temperature of building according to required desired temperature by being activated to relevant device.
The invention further relates to a kind of building, it includes predicting the forecasting system of the energy expenditure of building, the prediction
System implements the Forecasting Methodology of the energy expenditure of prediction building described above.
The invention further relates to the forecasting system for the energy requirement for predicting building, it includes the side for allowing to implement to be described above
The computer of method.Heat supply and air-conditioning equipment of the system advantageously with building are associated, so as to be based upon reach live it is uncommon
The comfort level of prestige and the energy expenditure that calculates implement thermal conditioning.The system includes such as control unit, the control unit bag
Include the computer for the method for implementing to be described above.This method can be implemented by the software service of storage on an information carrier.
Finally, building can be equipped with the energy expenditure forecasting system of the building for the method for implementing to be described above, and be used for
Adjust or manage its relevant device in a broad sense.
It should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention and it is unrestricted, although with reference to preferable
The present invention is described in detail embodiment, it will be understood by those within the art that, can be to technology of the invention
Scheme is modified or equivalent substitution, and without departing from the spirit and scope of technical solution of the present invention, it all should cover in this hair
Among bright right.
Claims (10)
1. a kind of building energy consumption Forecasting Methodology based on metering separate time series, it is characterised in that：Including,
Gather data and the storage of the energy consumption and temperature of building；
The energy consumption and temperature data of storage will be gathered as the input parameter of Time series analysis method；
According to metering separate and correlation analysis, by Time series analysis method predict energy consumption and temperature trend and time because
The sub major influence factors as building energy consumption；
Using the main factor to affect established and the energy consumption of collection as the parameter in the BP neural network model established, come pre-
Measure the energy consumption of future architecture.
2. metering separate time series building energy consumption Forecasting Methodology is based on as claimed in claim 1, it is characterised in that：It is described to adopt
Collect data and the storage of the energy consumption and temperature of building, it is carried out by energy consumption data collecting system,
The energy consumption data collecting system includes,
Measuring layer (100), including energy consumption measure collecting device (101) and temperature monitoring equipment (102), the energy consumption measure collection
Equipment (101) is acquired to the electric consumption on lighting of building, power electricity consumption, air conditioning electricity and the energy consumption of special electricity consumption, the temperature
Monitoring device (102) is acquired to space temperature；
Communication layers (200), set up writing to each other between measuring layer (100) and management level (300)；And,
Management level (300), send data acquisition instructions and corresponding energy consumption and temperature data to collection are stored；Wherein,
The management level (300) send data acquisition instructions, after the communication Protocol Conversion of the communication layers (200), reach institute
State the energy consumption measure collecting device (101) and temperature monitoring equipment (102) of measuring layer (100), the energy consumption measure collecting device
(101) receive instruction with temperature monitoring equipment (102) and responded after verification, corresponding energy consumption and temperature data are fed back to
The management level (300), store to database by handling subitem.
3. metering separate time series building energy consumption Forecasting Methodology is based on as claimed in claim 2, it is characterised in that：It is described to incite somebody to action
The energy consumption and temperature data of storage are gathered as the input parameter of Time series analysis method, wherein,
The time series is the auto-correlation function and partial autocorrelation function for calculating the energy consumption and temperature data.
4. metering separate time series building energy consumption Forecasting Methodology is based on as claimed in claim 3, it is characterised in that：
The auto-correlation coefficient, it is defined：
Because having for a stationary process：
So
As k=0, there is ρ0=1, the auto-correlation coefficient row ρ by variable of lag period kkK=0,1,2 ... it is referred to as auto-correlation letter
Number.
5. as described in claim 3 or 4 based on metering separate time series building energy consumption Forecasting Methodology, it is characterised in that：
The partial autocorrelation function to describe random process architectural feature, wherein,
Use φkjJ-th of regression coefficient in k rank autoregressive process is represented, then k ranks autoregression model is expressed as：xt=φk1xt-1+
φk2xt-2+...+φkkxt-k+ut, wherein φkkIt is last regression coefficient；
If φkkRegard lag period k function as, then claim φkk, k=1,2 ... it is partial autocorrelation function；
If time series { yt, the item number for taking rolling average is n, then the calculation formula of t+1 phase predicted values is：
Y in above formulatRepresent t phase actual values；Represent t phase Single moving average numbers；yt+1Expression t+1 phases predicted value (t >=
n)。
6. as described in Claims 1 to 4 is any based on metering separate time series building energy consumption Forecasting Methodology, its feature exists
In：The BP neural network model, wherein,
According to the data processing feature of neutral net, input data is normalized, using premnmx functions training
Samples normalization is between [0,1], and method is as follows：
Wherein, x, x ' are the forward and backward value of normalization, xmaxIt is maximum in sample, xminIt is minimum value in sample.
7. metering separate time series building energy consumption Forecasting Methodology is based on as claimed in claim 6, it is characterised in that：The BP
Neural network model, it, which sets up process, includes,
Establishment, training, emulation, prediction, the renormalization processing of network；Wherein,
The establishment of the network, its type selecting BP neural network；
The training, its function uses trainlm；
The renormalization processing, it is that the energy consumption data predicted using mapminmax function pairs is handled.
8. metering separate time series building energy consumption Forecasting Methodology is based on as claimed in claim 7, it is characterised in that：The BP
Neural network model,
Include the network of three input neurons, nine hidden neurons and four output neurons；
The hidden neuron transmission function is using being tansig functions, and it is linear letter that the output neuron transmission function, which is used,
Number purelin.
9. a kind of forecasting system for the energy expenditure for predicting building, it is characterised in that：The forecasting system includes control unit, should
Control unit implement as described in Claims 1 to 4,7 or 8 are any based on metering separate time series building energy consumption prediction side
Method.
10. a kind of building, it is characterised in that：Including implementing being counted based on subitem as described in Claims 1 to 4,7 or 8 are any
Measure the forecasting system of the energy expenditure of the prediction building of time series building energy consumption Forecasting Methodology.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710321447.2A CN106991504B (en) | 2017-05-09 | 2017-05-09 | Building energy consumption prediction method and system based on subentry measurement time sequence and building |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710321447.2A CN106991504B (en) | 2017-05-09 | 2017-05-09 | Building energy consumption prediction method and system based on subentry measurement time sequence and building |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106991504A true CN106991504A (en) | 2017-07-28 |
| CN106991504B CN106991504B (en) | 2021-07-16 |
Family
ID=59418606
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710321447.2A Active CN106991504B (en) | 2017-05-09 | 2017-05-09 | Building energy consumption prediction method and system based on subentry measurement time sequence and building |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106991504B (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107590565A (en) * | 2017-09-08 | 2018-01-16 | 北京首钢自动化信息技术有限公司 | A kind of method and device for building building energy consumption forecast model |
| CN107730006A (en) * | 2017-09-13 | 2018-02-23 | 重庆电子工程职业学院 | A kind of nearly zero energy consumption controller of building based on regenerative resource big data deep learning |
| CN108320016A (en) * | 2018-03-08 | 2018-07-24 | 南京工业大学 | A kind of building energy consumption short term prediction method |
| CN108764568A (en) * | 2018-05-28 | 2018-11-06 | 哈尔滨工业大学 | A kind of data prediction model tuning method and device based on LSTM networks |
| CN109063903A (en) * | 2018-07-19 | 2018-12-21 | 山东建筑大学 | A kind of building energy consumption prediction technique and system based on deeply study |
| CN109063366A (en) * | 2018-08-22 | 2018-12-21 | 深圳市建筑科学研究院股份有限公司 | A kind of building performance online data preprocess method based on time and spatial weighting |
| CN110544180A (en) * | 2019-08-22 | 2019-12-06 | 万洲电气股份有限公司 | Building energy-saving system based on energy consumption prediction and analysis diagnosis |
| CN111753470A (en) * | 2020-06-30 | 2020-10-09 | 桂林电子科技大学 | Building energy consumption prediction method based on rough set and deep belief neural network |
| CN112183826A (en) * | 2020-09-15 | 2021-01-05 | 湖北大学 | Building energy consumption prediction method based on deep cascade generation countermeasure network and related product |
| CN114593375A (en) * | 2022-03-30 | 2022-06-07 | 常州通用自来水有限公司 | Secondary water supply community pipeline leakage monitoring and positioning method based on pump house energy consumption |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202649806U (en) * | 2012-07-03 | 2013-01-02 | 上海同杭节能科技有限公司 | Building energy consumption subentry metering system |
| CN102938092A (en) * | 2012-10-08 | 2013-02-20 | 珠海派诺科技股份有限公司 | Prediction method of building energy consumption in festivals and holidays based on neural network |
| CN104239982A (en) * | 2014-10-12 | 2014-12-24 | 刘岩 | Method for predicting energy consumption of buildings during holidays and festivals on basis of time series and neural networks |
| US20150317589A1 (en) * | 2012-11-09 | 2015-11-05 | The Trustees Of Columbia University In The City Of New York | Forecasting system using machine learning and ensemble methods |
| CN105631539A (en) * | 2015-12-25 | 2016-06-01 | 上海建坤信息技术有限责任公司 | Intelligent building energy consumption prediction method based on support vector machine |
-
2017
- 2017-05-09 CN CN201710321447.2A patent/CN106991504B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN202649806U (en) * | 2012-07-03 | 2013-01-02 | 上海同杭节能科技有限公司 | Building energy consumption subentry metering system |
| CN102938092A (en) * | 2012-10-08 | 2013-02-20 | 珠海派诺科技股份有限公司 | Prediction method of building energy consumption in festivals and holidays based on neural network |
| US20150317589A1 (en) * | 2012-11-09 | 2015-11-05 | The Trustees Of Columbia University In The City Of New York | Forecasting system using machine learning and ensemble methods |
| CN104239982A (en) * | 2014-10-12 | 2014-12-24 | 刘岩 | Method for predicting energy consumption of buildings during holidays and festivals on basis of time series and neural networks |
| CN105631539A (en) * | 2015-12-25 | 2016-06-01 | 上海建坤信息技术有限责任公司 | Intelligent building energy consumption prediction method based on support vector machine |
Non-Patent Citations (3)
| Title |
|---|
| GUANG SHI；DERONG LIU；QINGLAI WEI；: "Energy consumption prediction of office buildings based on echo state networks", 《NEUROCOMPUTING》 * |
| KADIR AMASYALI；NORAEL-GOHARY；: "Building lighting energy consumption prediction for supporting energy data analytics", 《PROCEDIA ENGINEERING》 * |
| 王文举等: "《经济预测、决策与对策》", 30 September 2013, 首都经济贸易大学出版社 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107590565A (en) * | 2017-09-08 | 2018-01-16 | 北京首钢自动化信息技术有限公司 | A kind of method and device for building building energy consumption forecast model |
| CN107730006A (en) * | 2017-09-13 | 2018-02-23 | 重庆电子工程职业学院 | A kind of nearly zero energy consumption controller of building based on regenerative resource big data deep learning |
| CN107730006B (en) * | 2017-09-13 | 2021-01-05 | 重庆电子工程职业学院 | Building near-zero energy consumption control method based on renewable energy big data deep learning |
| CN108320016A (en) * | 2018-03-08 | 2018-07-24 | 南京工业大学 | A kind of building energy consumption short term prediction method |
| CN108320016B (en) * | 2018-03-08 | 2023-09-19 | 南京工业大学 | Short-term prediction method for building energy consumption |
| CN108764568B (en) * | 2018-05-28 | 2020-10-23 | 哈尔滨工业大学 | Data prediction model tuning method and device based on LSTM network |
| CN108764568A (en) * | 2018-05-28 | 2018-11-06 | 哈尔滨工业大学 | A kind of data prediction model tuning method and device based on LSTM networks |
| CN109063903A (en) * | 2018-07-19 | 2018-12-21 | 山东建筑大学 | A kind of building energy consumption prediction technique and system based on deeply study |
| CN109063903B (en) * | 2018-07-19 | 2021-04-09 | 山东建筑大学 | Building energy consumption prediction method and system based on deep reinforcement learning |
| CN109063366B (en) * | 2018-08-22 | 2022-12-23 | 深圳市建筑科学研究院股份有限公司 | Building performance data online preprocessing method based on time and space weighting |
| CN109063366A (en) * | 2018-08-22 | 2018-12-21 | 深圳市建筑科学研究院股份有限公司 | A kind of building performance online data preprocess method based on time and spatial weighting |
| CN110544180A (en) * | 2019-08-22 | 2019-12-06 | 万洲电气股份有限公司 | Building energy-saving system based on energy consumption prediction and analysis diagnosis |
| CN111753470A (en) * | 2020-06-30 | 2020-10-09 | 桂林电子科技大学 | Building energy consumption prediction method based on rough set and deep belief neural network |
| CN111753470B (en) * | 2020-06-30 | 2022-09-16 | 桂林电子科技大学 | Building energy consumption prediction method based on rough set and deep belief neural network |
| CN112183826A (en) * | 2020-09-15 | 2021-01-05 | 湖北大学 | Building energy consumption prediction method based on deep cascade generation countermeasure network and related product |
| CN112183826B (en) * | 2020-09-15 | 2023-08-01 | 湖北大学 | Building energy consumption prediction method based on deep cascade generation countermeasure network and related products |
| CN114593375A (en) * | 2022-03-30 | 2022-06-07 | 常州通用自来水有限公司 | Secondary water supply community pipeline leakage monitoring and positioning method based on pump house energy consumption |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106991504B (en) | 2021-07-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106991504A (en) | Building energy consumption Forecasting Methodology, system and building based on metering separate time series | |
| CN113746144B (en) | Source-grid-load real-time interaction electric carbon control method and intelligent management system thereof | |
| Tian et al. | Predictive model of energy consumption for office building by using improved GWO-BP | |
| CN106779129A (en) | A kind of Short-Term Load Forecasting Method for considering meteorologic factor | |
| CN107909220A (en) | Electric heating load prediction method | |
| CN106315319A (en) | Intelligent pre-dispatching method and system for elevator | |
| CN112633571A (en) | LSTM-based ultrashort-term load prediction method under source network load interaction environment | |
| CN104376371B (en) | A kind of distribution based on topology is layered load forecasting method | |
| CN106410781A (en) | Power consumer demand response potential determination method | |
| CN110489783B (en) | QNNN-based low-voltage transformer area line loss rate reasonable range estimation method | |
| CN104463356A (en) | Photovoltaic power generation power prediction method based on multi-dimension information artificial neural network algorithm | |
| Song et al. | An indoor temperature prediction framework based on hierarchical attention gated recurrent unit model for energy efficient buildings | |
| CN103839108A (en) | Energy saving evaluation system and method for power supply and distribution networks of industrial enterprises | |
| WO2023083235A1 (en) | Optimization method and system for performing prediction on power distribution network in distributed manner | |
| CN114219212A (en) | Flexible scheduling method for demand side resources considering ubiquitous power internet of things and edge calculation | |
| CN109403425B (en) | Optimized dispatching system of water supply pipe network | |
| CN113872228A (en) | Electric vehicle scheduling method and device applied to power grid peak shaving frequency modulation | |
| Wang et al. | Aggregated large-scale air-conditioning load: Modeling and response capability evaluation of virtual generator units | |
| CN113237135A (en) | SD-LSTMs-based apartment independent heating system heat demand prediction and energy-saving control method | |
| Shilin et al. | BP neural network for the prediction of urban building energy consumption based on Matlab and its application | |
| CN112307675A (en) | Neural network-based temperature-sensitive load separation identification method and system | |
| Wu et al. | A deep learning-based strategy to the energy management-advice for time-of-use rate of household electricity consumption | |
| CN116307437A (en) | Virtual power plant optimal scheduling method and system based on carbon emission right exchange mechanism | |
| CN104463683B (en) | A kind of Mid-long term load forecasting method of power grid containing multi-source | |
| Guo et al. | Research on short-term traffic demand of taxi in large cities based on BP neural network algorithm |
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 |