CN109102103A - A kind of multi-class energy consumption prediction technique based on Recognition with Recurrent Neural Network - Google Patents
A kind of multi-class energy consumption prediction technique based on Recognition with Recurrent Neural Network Download PDFInfo
- Publication number
- CN109102103A CN109102103A CN201810672863.1A CN201810672863A CN109102103A CN 109102103 A CN109102103 A CN 109102103A CN 201810672863 A CN201810672863 A CN 201810672863A CN 109102103 A CN109102103 A CN 109102103A
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- neural network
- data
- recognition
- class
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Computational Linguistics (AREA)
- Development Economics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Game Theory and Decision Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention discloses a kind of multi-class energy consumption prediction technique based on Recognition with Recurrent Neural Network, training forms a Recognition with Recurrent Neural Network model in advance, include: step S1, load original energy consumption data on the basis of Recognition with Recurrent Neural Network model, and missing values and exceptional value are judged from original energy consumption data, and missing values are detected and handled with exceptional value；Step S2, based on original energy consumption data, temporal aspect data is extracted from original energy consumption data and establish the characteristic set of Recognition with Recurrent Neural Network model, characteristic set is normalized；Step S3, the characteristic set after normalized is trained in batches, establishes multi-class output nerve network in conjunction with non-sequential characteristic, the multi-class energy consumption data of multi-class output nerve network output is predicted.The utility model has the advantages that multi-class output nerve network is established using the temporal aspect data of original energy consumption data, and in conjunction with non-sequential characteristic, to be predicted.
Description
Technical field
The present invention relates to building energy consumption electric powder prediction more particularly to a kind of multi-class energy based on Recognition with Recurrent Neural Network
Consume prediction technique.
Background technique
With quickly propelling for urbanization process, heavy construction energy consumption as caused by hotel, market etc. in city is got over
Come bigger, such energy consumption is also continuously increased in the accounting of national total energy consumption.In response to national energy conservation and emission reduction, building section
The significant concern point of each Development of large city and planning can be had become.Building energy consumption predicts the important ring as Building Energy Analysis
Section is conducive to find energy for building existing various problems in the process in time, the environmental protection and energy saving for instructing building can also be helped to set
Meter guarantees with energy safety.
Generally, energy consumption forecasting problem can be regarded as a regression problem, i.e., is fitted point to energy consumption historical data
Analysis, to predict the power consumption values of subsequent time.Common energy consumption prediction technique includes linear regression prediction, Support vector regression
Prediction, regression forecasting based on BP neural network etc., but these methods are usually free of and utilize energy consumption historical data well
Temporal aspect data, especially by when energy consumption prediction scene under, temporal aspect data are more obvious and important.In addition, most texts
The energy consumption prediction model mentioned in offering can only all predict the energy consumption of a node or a classification, more if necessary to predict
A classification energy consumption then needs to establish multiple models and is trained, and process is cumbersome, is unfavorable for the rapid deployment of Energy Consumption Prediction System
And efficient operation.
Summary of the invention
For the above-mentioned problems in the prior art, it is pre- now to provide a kind of multi-class energy consumption based on Recognition with Recurrent Neural Network
Survey method.
Specific technical solution is as follows:
A kind of multi-class energy consumption prediction technique based on Recognition with Recurrent Neural Network, wherein training in advance forms a circulation nerve
Network model predicts multi-class energy consumption using the Recognition with Recurrent Neural Network model, described based on Recognition with Recurrent Neural Network
Multi-class energy consumption prediction technique includes:
Step S1, original energy consumption data is loaded on the basis of the Recognition with Recurrent Neural Network model, and from the original energy
Missing values and exceptional value are filtered out in consumption data, and the missing values are detected and handled with the exceptional value；
Step S2, based on the original energy consumption data, temporal aspect data are extracted from the original energy consumption data
And the characteristic set of the Recognition with Recurrent Neural Network model is established, then the characteristic set is normalized；
Step S3, the characteristic set after normalized is trained in batches, in conjunction with non-sequential characteristic
Multi-class output nerve network is established, the multi-class energy consumption data of multi-class output nerve network output is predicted.
Preferably, in the step S1, the missing values and the exceptional value are detected using box-shaped map analysis method.
Preferably, the step S1 includes:
Step S10, the original energy consumption data is loaded on the basis of the Recognition with Recurrent Neural Network model, and from the original
Judge the original energy consumption data with the presence or absence of missing values in beginning energy consumption data；
If it exists, then the missing values processing is carried out；
If it does not exist, then step S11 is carried out；
The step S11, the relative energy consumption value for calculating the original energy consumption data, and whether judge the relative energy consumption value
There are exceptional values；
If it exists, then the outlier processing is carried out；
If it does not exist, then step S12 is carried out；
The step S12, sub-category processing is carried out to the relative energy consumption value；
Step S13, calculating and store processing after the relative energy consumption value into the original energy consumption data model.
Preferably, it in the step S10, to the original energy consumption data with the constraint relationship is preset, utilizes
Valid data count and calculate the ratio for accounting for corresponding period total energy consumption at the corresponding moment, then carry out filling up the missing in proportion
Value；
To the original energy consumption data without the constraint relationship is preset, front and back moment corresponding valid data are utilized
Average value carry out filling up the missing values.
Preferably, the step S2 includes:
Step S20, based on the original energy consumption data, and the original energy consumption data is divided into test data and instruction
Practice data；
Step S21, the temporal aspect data are extracted to the test data and the training data respectively, and establishes institute
The characteristic set of Recognition with Recurrent Neural Network model is stated, place then is normalized to the training data using linear normalization method
Reason, is then handled the test data using the parameter of the training data normalized.
Preferably, the step S3 includes:
Step S30, the test data is trained in batches；
Step S31, the default first stage uses the historical energy consumption data as the first essential characteristic, according to described first
Essential characteristic is established Recognition with Recurrent Neural Network and is exported；
Step S32, default second stage uses the Recognition with Recurrent Neural Network of output as essential characteristic, and described in reference
Non-sequential characteristic establishes the multi-class output nerve network, to the multi-class energy of multi-class output nerve network output
Consumption data are predicted.
Preferably, a gating cycle unit neural network, the gating cycle unit nerve are established in the first stage
Network models the historical energy consumption data；
The gating cycle unit neural network includes updating door control unit and resetting door control unit, and the update gate is single
Member determines that the status information of the historical energy consumption data previous moment is admitted to the degree of current state, the resetting door control unit
The degree that the status information of the historical energy consumption data previous moment arrives is ignored in decision.
Preferably, a multi-class output nerve network, the multi-class output nerve network are established in the second stage
It is constructed jointly by the gating cycle unit neural network and the non-sequential characteristic.
Preferably, the non-sequential characteristic includes temporal characteristics, weather characteristics.
Preferably, the original energy consumption data include lighting energy consumption data, air conditioning energy consumption data, power energy consumption data and its
Its energy consumption data classification.
Technical solution of the present invention beneficial effect is: being carried out using Recognition with Recurrent Neural Network model to multi-class energy consumption pre-
It surveys, establishes multi-class output nerve network using the temporal aspect data of original energy consumption data, and in conjunction with non-sequential characteristic,
The multi-class energy consumption data of multi-class output nerve network output is predicted.
Detailed description of the invention
With reference to appended attached drawing, more fully to describe the embodiment of the present invention.However, appended attached drawing be merely to illustrate and
It illustrates, and is not meant to limit the scope of the invention.
Fig. 1 is the flow chart about the multi-class energy consumption prediction technique based on Recognition with Recurrent Neural Network in the present invention；
Fig. 2 is in the present invention, about the specific flow chart of step S1；
Fig. 3 is in the present invention, about the specific flow chart of step S2；
Fig. 4 is in the present invention, about the specific flow chart of step S3；
Fig. 5 is the flow chart in the present invention, about the building of multi-class neural network；
Fig. 6 is the comparative graph about lighting energy consumption predicted value and true value in the present invention；
Fig. 7 is the comparative graph about Energy consumption forecast for air conditioning value and true value in the present invention；
Fig. 8 is the comparative graph about power energy consumption predicted value and true value in the present invention；
Fig. 9 is the comparative graph about other energy consumption predicted values and true value in the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art without creative labor it is obtained it is all its
His embodiment, shall fall within the protection scope of the present invention.
It should be noted that in the absence of conflict, the feature in embodiment and embodiment in the present invention can phase
Mutually combination.
The present invention will be further explained below with reference to the attached drawings and specific examples, but not as the limitation of the invention.
The present invention includes a kind of multi-class energy consumption prediction technique based on Recognition with Recurrent Neural Network, wherein training is formed in advance
One Recognition with Recurrent Neural Network model predicts multi-class energy consumption using Recognition with Recurrent Neural Network model, is based on Recognition with Recurrent Neural Network
Multi-class energy consumption prediction technique include:
Step S1, original energy consumption data is loaded on the basis of Recognition with Recurrent Neural Network model, and from original energy consumption data
Judge missing values and exceptional value, and the missing values are detected and handled with the exceptional value；
Step S2, based on original energy consumption data, temporal aspect data are extracted from original energy consumption data and are established and are followed
The characteristic set of ring neural network model, is then normalized characteristic set；
Step S3, the characteristic set after normalized is trained in batches, is established in conjunction with non-sequential characteristic
Multi-class output nerve network predicts the multi-class energy consumption data of multi-class output nerve network output.
Using the above scheme, as shown in Figure 1, the multi-class energy consumption prediction technique based on Recognition with Recurrent Neural Network, applied to following
In ring neural network model, original energy consumption data is loaded on the basis of Recognition with Recurrent Neural Network model first, and use box-shaped figure
Analytic approach judges missing values therein and exceptional value from original energy consumption data, and to the missing values and the exceptional value into
Row detection and processing；
Further, based on original energy consumption data, temporal aspect data are extracted from original energy consumption data and are established
The characteristic set of Recognition with Recurrent Neural Network model, is then normalized characteristic set；Then to normalized after
Characteristic set trained in batches, multi-class output nerve network is established in conjunction with non-sequential characteristic, to multi-class output
The multi-class energy consumption data of neural network output predicted.
In a kind of preferably embodiment, step S1 includes:
Step S10, original energy consumption data is loaded on the basis of Recognition with Recurrent Neural Network model, and from original energy consumption data
Judge original energy consumption data with the presence or absence of missing values；
If it exists, then missing values processing is carried out；
If it does not exist, then step S11 is carried out；
Step S11, the relative energy consumption value of original energy consumption data is calculated, and judges relative energy consumption value with the presence or absence of exceptional value；
If it exists, then outlier processing is carried out；
If it does not exist, then step S12 is carried out；
Step S12, sub-category processing is carried out to relative energy consumption value；
Step S13, calculating and store processing after relative energy consumption value into original energy consumption data model.
Specifically, as shown in Fig. 2, loading original energy consumption data on the basis of Recognition with Recurrent Neural Network model, and from original
Judge original energy consumption data with the presence or absence of missing values in energy consumption data, and if it exists, then to preset with the constraint relationship
Original energy consumption data counts using valid data and calculates the ratio that the corresponding moment accounts for corresponding period total energy consumption, then by than
Example carries out filling up missing values；To presetting the original energy consumption data without the constraint relationship, corresponding have using the front and back moment
The average value of effect data carries out filling up missing values；
Further, the relative energy consumption value of original energy consumption data is calculated, i.e., the energy consumption data in each hour section makes later
With box-shaped map analysis method and cooperates the means analysis of active observation with the presence or absence of abnormal data and judges exception class value, and if it exists,
Classification processing is then carried out, classification energy consumption is then calculated according to actual needs, the classification of monitoring described in ammeter is divided into illumination, sky
Tune, power and other classification energy consumptions, finally store the relative energy consumption value after processing into original energy consumption data model.
In a kind of preferably embodiment, step S2 includes:
Step S20, using original energy consumption data, and original energy consumption data is divided into test data and training data；
Step S21, temporal aspect data are extracted to test data and training data respectively, and establishes Recognition with Recurrent Neural Network mould
Then the characteristic set of type is normalized test data using linear normalization method, training data is then utilized
The parameter of normalized handles test data.
Specifically, as shown in figure 3, using original energy consumption data first, original energy consumption data is divided into test data and instruction
Practice data, be used herein 12 months by when energy consumption data as training data, one month by when energy consumption data conduct
Then test data carries out the extraction of temporal aspect data respectively on both data sets, and place is normalized to training data
Reason, is handled test data followed by the parameter of training data normalized, then training data is introduced circulation mind
It is trained through network model；
Further, in neural network training process, the present invention uses Averaged Square Error of Multivariate (Mean Square
Error, MSE) it is used as loss function, specific in multi-class energy consumption prediction model, then use the average MSE of each classification energy consumption
As loss function, calculation is as follows:
Wherein, yijWithIt is the true value and predicted value of i-th of moment, j-th of classification energy consumption respectively, n and C are meter respectively
The energy consumption historical data entry number and energy consumption classification number used when calculation；
Further, it after Recognition with Recurrent Neural Network model training, is tested in test data, then to circulation mind
Anti-normalization processing is carried out through network model output result, can just obtain energy consumption predicted value；
In a kind of preferably embodiment, step S3 includes:
Step S30, test data is trained in batches；
Step S31, default first stage usage history energy consumption data is as the first essential characteristic, according to the first essential characteristic
It establishes Recognition with Recurrent Neural Network and exports；
Step S32, default second stage uses the Recognition with Recurrent Neural Network of output as essential characteristic, and special with reference to non-sequential
Levy Data Data, establish multi-class output nerve network, with carry out by when predict and assessment.
Specifically, as shown in figure 4, trained to test data in batches, first stage usage history energy consumption is then preset
Data are established Recognition with Recurrent Neural Network according to the first essential characteristic and are exported as the first essential characteristic, then preset second stage
Use the Recognition with Recurrent Neural Network of output as essential characteristic, and refer to non-sequential characteristic data, establishes multi-class output mind
Through network, with carry out by when predict and assessment；
Further, using square of the quadratic sum of predicted value and the difference of true value and true value and the difference of true average
Ratio R between and2As evaluation index, specific in multi-class energy consumption prediction model, then practical each classification energy consumption R2It is flat
For mean value as evaluation index, calculation formula is as follows:
Wherein,Indicate the average value of j-th of classification energy consumption true value, yijWithIt is i-th of moment, j-th of class respectively
The true value and predicted value of other energy consumption, n and C are the energy consumption historical data entry number used when calculating and energy consumption classification number respectively.
In a kind of preferably embodiment, a gating cycle unit neural network, gating cycle list are established in the first stage
First neural network models historical energy consumption data；
Gating cycle unit neural network includes updating door control unit and resetting door control unit, updates door control unit decision and goes through
The status information of history energy consumption data previous moment is admitted to the degree of current state, and history energy consumption is ignored in resetting door control unit decision
The degree that the status information of data previous moment arrives.
Specifically, different from traditional Recognition with Recurrent Neural Network, it is single that gating cycle unit neural network introduces two gates
Member, including door control unit and resetting door control unit are updated, update the state that door control unit determines historical energy consumption data previous moment
Information is admitted to the degree of current state, and resetting door control unit determines that the status information for ignoring historical energy consumption data previous moment arrives
Degree, specifically, it is assumed that the output of previous moment be ht-1, the input at current time is xt, then the output h at current timetIt is logical
Following equations are crossed to be calculated:
zt=σ (Wz·[ht-1,xt])
rt=σ (Wr·[ht-1,xt])
Wherein, [] representing matrix connects, representing matrix dot product, * representing matrix element multiplication, Wz, Wr, W~It is network
Parameter, ztTo update door, rtTo reset door, σ is usually using sigmoid function, side h
Journey are as follows:
Tanh is hyperbolic tangent function, equation are as follows:
In a kind of preferably embodiment, a multi-class output nerve network, multi-class output mind are established in second stage
It is constructed jointly by gating cycle unit neural network with non-sequential characteristic data through network.
Specifically, as shown in figure 5, the first stage establishes a gating cycle unit neural network, gating cycle unit nerve
Network integration temporal aspect data model to historical energy consumption data and then be exported, in conjunction with non-sequential characteristic data,
Multi-class output nerve network is constructed jointly, and wherein non-sequential characteristic includes temporal characteristics, weather characteristics.
In a kind of preferably embodiment, original energy consumption data includes lighting energy consumption data, air conditioning energy consumption data, kinetic force
Consume data and other energy consumption data classifications.
Specifically, as shown in Fig. 6,7,8,9, the true value and predicted value of corresponding each classification energy consumption are realized preferably
Prediction accuracy.
Technical solution of the present invention beneficial effect is: being carried out using Recognition with Recurrent Neural Network model to multi-class energy consumption pre-
It surveys, establishes multi-class output nerve network using the temporal aspect data of original energy consumption data, and in conjunction with non-sequential characteristic,
The multi-class energy consumption data of multi-class output nerve network output is predicted.
The above is only preferred embodiments of the present invention, are not intended to limit the implementation manners and the protection scope of the present invention, right
For those skilled in the art, it should can appreciate that and all replace with being equal made by description of the invention and diagramatic content
It changes and obviously changes obtained scheme, should all be included within the scope of the present invention.
Claims (10)
1. a kind of multi-class energy consumption prediction technique based on Recognition with Recurrent Neural Network, which is characterized in that training in advance forms a circulation
Neural network model predicts multi-class energy consumption using the Recognition with Recurrent Neural Network model, described based on circulation nerve net
The multi-class energy consumption prediction technique of network includes:
Step S1, original energy consumption data is loaded on the basis of the Recognition with Recurrent Neural Network model, and from the original energy consumption number
Missing values and exceptional value are judged in, and the missing values are detected and handled with the exceptional value；
Step S2, based on the original energy consumption data, temporal aspect data are extracted from the original energy consumption data and are built
The characteristic set for founding the Recognition with Recurrent Neural Network model, is then normalized the characteristic set；
Step S3, the characteristic set after normalized is trained in batches, is established in conjunction with non-sequential characteristic
Multi-class output nerve network predicts the multi-class energy consumption data of multi-class output nerve network output.
2. the multi-class energy consumption prediction technique according to claim 1 based on Recognition with Recurrent Neural Network, which is characterized in that in institute
It states in step S1, detects the missing values and the exceptional value using box-shaped map analysis method.
3. the multi-class energy consumption prediction technique according to claim 1 based on Recognition with Recurrent Neural Network, which is characterized in that described
Step S1 includes:
Step S10, the original energy consumption data is loaded on the basis of the Recognition with Recurrent Neural Network model, and from the original energy
Judge the original energy consumption data with the presence or absence of missing values in consumption data；
If it exists, then the missing values processing is carried out；
If it does not exist, then step S11 is carried out；
The step S11, the relative energy consumption value for calculating the original energy consumption data, and judge that the relative energy consumption value whether there is
Exceptional value；
If it exists, then the outlier processing is carried out；
If it does not exist, then step S12 is carried out；
The step S12, sub-category processing is carried out to the relative energy consumption value；
Step S13, calculating and store processing after the relative energy consumption value into the original energy consumption data model.
4. the multi-class energy consumption prediction technique according to claim 3 based on Recognition with Recurrent Neural Network, which is characterized in that in institute
It states in step S10, to the original energy consumption data with the constraint relationship is preset, is counted and calculated pair using valid data
Then the ratio that corresponding period total energy consumption should be accounted for constantly carries out filling up the missing values in proportion；
To the original energy consumption data without the constraint relationship is preset, the flat of front and back moment corresponding valid data is utilized
Mean value carries out filling up the missing values.
5. the multi-class energy consumption prediction technique according to claim 1 based on Recognition with Recurrent Neural Network, which is characterized in that described
Step S2 includes:
Step S20, based on the original energy consumption data, and the original energy consumption data is divided into test data and training number
According to；
Step S21, the temporal aspect data are extracted to the test data and the training data respectively, and is followed described in foundation
Then the characteristic set of ring neural network model is normalized the training data using linear normalization method,
Then the test data is handled using the parameter of the training data normalized.
6. the multi-class energy consumption prediction technique according to claim 5 based on Recognition with Recurrent Neural Network, which is characterized in that described
Step S3 includes:
Step S30, the test data is trained in batches；
Step S31, the default first stage uses the historical energy consumption data as the first essential characteristic, basic according to described first
Feature is established Recognition with Recurrent Neural Network and is exported；
Step S32, default second stage use the Recognition with Recurrent Neural Network of output as essential characteristic, and when referring to described non-
Sequence characteristics data establish the multi-class output nerve network, to the multi-class energy consumption number of multi-class output nerve network output
According to being predicted.
7. the multi-class energy consumption prediction technique according to claim 6 based on Recognition with Recurrent Neural Network, which is characterized in that in institute
It states the first stage and establishes a gating cycle unit neural network, the gating cycle unit neural network is to the history energy consumption number
According to being modeled；
The gating cycle unit neural network includes updating door control unit and resetting door control unit, and the update door control unit is determined
The status information of the fixed historical energy consumption data previous moment is admitted to the degree of current state, and the resetting door control unit determines
Ignore the degree that the status information of the historical energy consumption data previous moment arrives.
8. the multi-class energy consumption prediction technique according to claim 7 based on Recognition with Recurrent Neural Network, which is characterized in that in institute
It states second stage and establishes a multi-class output nerve network, the multi-class output nerve network passes through the gating cycle unit
Neural network constructs jointly with the non-sequential characteristic.
9. the multi-class energy consumption prediction technique according to claim 1 based on Recognition with Recurrent Neural Network, which is characterized in that described
Non-sequential characteristic includes temporal characteristics, weather characteristics.
10. the multi-class energy consumption prediction technique according to claim 1 based on Recognition with Recurrent Neural Network, which is characterized in that institute
Stating original energy consumption data includes lighting energy consumption data, air conditioning energy consumption data, power energy consumption data and other energy consumption data classifications.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810672863.1A CN109102103A (en) | 2018-06-26 | 2018-06-26 | A kind of multi-class energy consumption prediction technique based on Recognition with Recurrent Neural Network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810672863.1A CN109102103A (en) | 2018-06-26 | 2018-06-26 | A kind of multi-class energy consumption prediction technique based on Recognition with Recurrent Neural Network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109102103A true CN109102103A (en) | 2018-12-28 |
Family
ID=64845077
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810672863.1A Pending CN109102103A (en) | 2018-06-26 | 2018-06-26 | A kind of multi-class energy consumption prediction technique based on Recognition with Recurrent Neural Network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109102103A (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109961177A (en) * | 2019-03-11 | 2019-07-02 | 浙江工业大学 | A kind of general water cooled central air conditioner energy consumption prediction technique based on shot and long term memory Recognition with Recurrent Neural Network |
| CN111046027A (en) * | 2019-11-25 | 2020-04-21 | 北京百度网讯科技有限公司 | Missing value filling method and device for time series data |
| CN111766832A (en) * | 2020-06-29 | 2020-10-13 | 重庆大学 | Numerical control machine tool cutting energy consumption prediction modeling method driven by incomplete data |
| CN112070317A (en) * | 2020-09-17 | 2020-12-11 | 四川长虹电器股份有限公司 | Hotel air conditioner energy consumption prediction method |
| CN112988725A (en) * | 2021-03-15 | 2021-06-18 | 广东电网有限责任公司清远供电局 | Power transmission line data cleaning method and system, electronic equipment and storage medium |
| WO2021174881A1 (en) * | 2020-09-04 | 2021-09-10 | 平安科技（深圳）有限公司 | Multi-dimensional information combination prediction method, apparatus, computer device, and medium |
| CN116805275A (en) * | 2023-08-21 | 2023-09-26 | 中国标准化研究院 | Public institution energy consumption intelligent supervision system based on data analysis |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN104992238A (en) * | 2015-06-24 | 2015-10-21 | 国网山东省电力公司电力科学研究院 | Typical daily load characteristic-based power load forecasting method |
| CN105205558A (en) * | 2015-09-23 | 2015-12-30 | 南京磐能电力科技股份有限公司 | BP neural network model configuration method oriented to building energy consumption prediction |
| CN105205570A (en) * | 2015-10-16 | 2015-12-30 | 国网重庆铜梁区供电有限责任公司 | Power grid power sale quantity prediction method based on season time sequence analysis |
| CN106295877A (en) * | 2016-08-09 | 2017-01-04 | 四川大学 | A kind of intelligent grid electric energy usage amount Forecasting Methodology |
| CN106656637A (en) * | 2017-02-24 | 2017-05-10 | 国网河南省电力公司电力科学研究院 | Anomaly detection method and device |
-
2018
- 2018-06-26 CN CN201810672863.1A patent/CN109102103A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| 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 |
| CN104992238A (en) * | 2015-06-24 | 2015-10-21 | 国网山东省电力公司电力科学研究院 | Typical daily load characteristic-based power load forecasting method |
| CN105205558A (en) * | 2015-09-23 | 2015-12-30 | 南京磐能电力科技股份有限公司 | BP neural network model configuration method oriented to building energy consumption prediction |
| CN105205570A (en) * | 2015-10-16 | 2015-12-30 | 国网重庆铜梁区供电有限责任公司 | Power grid power sale quantity prediction method based on season time sequence analysis |
| CN106295877A (en) * | 2016-08-09 | 2017-01-04 | 四川大学 | A kind of intelligent grid electric energy usage amount Forecasting Methodology |
| CN106656637A (en) * | 2017-02-24 | 2017-05-10 | 国网河南省电力公司电力科学研究院 | Anomaly detection method and device |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109961177A (en) * | 2019-03-11 | 2019-07-02 | 浙江工业大学 | A kind of general water cooled central air conditioner energy consumption prediction technique based on shot and long term memory Recognition with Recurrent Neural Network |
| CN111046027A (en) * | 2019-11-25 | 2020-04-21 | 北京百度网讯科技有限公司 | Missing value filling method and device for time series data |
| CN111766832A (en) * | 2020-06-29 | 2020-10-13 | 重庆大学 | Numerical control machine tool cutting energy consumption prediction modeling method driven by incomplete data |
| WO2021174881A1 (en) * | 2020-09-04 | 2021-09-10 | 平安科技（深圳）有限公司 | Multi-dimensional information combination prediction method, apparatus, computer device, and medium |
| CN112070317A (en) * | 2020-09-17 | 2020-12-11 | 四川长虹电器股份有限公司 | Hotel air conditioner energy consumption prediction method |
| CN112988725A (en) * | 2021-03-15 | 2021-06-18 | 广东电网有限责任公司清远供电局 | Power transmission line data cleaning method and system, electronic equipment and storage medium |
| CN116805275A (en) * | 2023-08-21 | 2023-09-26 | 中国标准化研究院 | Public institution energy consumption intelligent supervision system based on data analysis |
| CN116805275B (en) * | 2023-08-21 | 2023-11-21 | 中国标准化研究院 | Public institution energy consumption intelligent supervision system based on data analysis |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109102103A (en) | A kind of multi-class energy consumption prediction technique based on Recognition with Recurrent Neural Network | |
| CN108510006A (en) | A kind of analysis of business electrical amount and prediction technique based on data mining | |
| Osorio et al. | A computationally efficient simulation-based optimization algorithm for large-scale urban transportation problems | |
| CN112418547A (en) | Bus stop passenger flow prediction method based on GCN-LSTM combined model | |
| CN103226741B (en) | Public supply mains tube explosion prediction method | |
| CN108280551A (en) | A kind of photovoltaic power generation power prediction method using shot and long term memory network | |
| CN113554466B (en) | Short-term electricity consumption prediction model construction method, prediction method and device | |
| CN102902882A (en) | Method for evaluating operation quality of information systems | |
| Lee et al. | Using climate classification to evaluate building energy performance | |
| Han et al. | Energy saving of buildings for reducing carbon dioxide emissions using novel dendrite net integrated adaptive mean square gradient | |
| CN102938092A (en) | Prediction method of building energy consumption in festivals and holidays based on neural network | |
| CN112434787B (en) | Terminal space energy consumption prediction method, medium and equipment based on total energy consumption of building | |
| Zou et al. | Use of skew-normal and skew-t distributions for mixture modeling of freeway speed data | |
| CN109886387A (en) | It is a kind of that the traffic time sequence forecasting method returned is promoted based on gating network and gradient | |
| CN110535149A (en) | A kind of area Gong Biantai electric load three-phase imbalance prediction technique | |
| CN113689030B (en) | Short-term wind power prediction method based on bidirectional attention and secondary optimization | |
| CN115359659B (en) | Lane opening and closing configuration method and system | |
| CN110991776A (en) | Method and system for realizing water level prediction based on GRU network | |
| Cao et al. | Short-term energy consumption prediction method for educational buildings based on model integration | |
| Vasanthkumar et al. | Improving energy consumption prediction for residential buildings using Modified Wild Horse Optimization with Deep Learning model | |
| CN110390419A (en) | Freeway toll station method for predicting based on PSO-LSSVM model | |
| CN107423854A (en) | Building population of the one kind based on GM (1,1) season trend prediction changes monitoring method | |
| CN104217296A (en) | Listed company performance comprehensive evaluation method | |
| CN109784562A (en) | A kind of wisdom grid power load forecasting method based on big data space-time cluster | |
| CN110489893B (en) | Variable weight-based bus load prediction method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20181228 |