CN110264001A - Electro-load forecast method based on multiple timings - Google Patents
Electro-load forecast method based on multiple timings Download PDFInfo
- Publication number
- CN110264001A CN110264001A CN201910534470.9A CN201910534470A CN110264001A CN 110264001 A CN110264001 A CN 110264001A CN 201910534470 A CN201910534470 A CN 201910534470A CN 110264001 A CN110264001 A CN 110264001A
- Authority
- CN
- China
- Prior art keywords
- load
- value
- electro
- formula
- data
- 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
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Abstract
The electro-load forecast method based on multiple timings that the present invention relates to a kind of, comprising the following steps: collect and predict regional power load, the production output value, economic conditions, the historical data of weather condition, and carry out the pretreatment of data.Wavelet decomposition is carried out to the historical data of power load, trend component, periodic component and random component are obtained after decomposition.In conjunction with the characteristics of traditional load forecasting method, choose different load forecasting methods to it is different when order components carry out load prediction, finally by superposition obtain the result of electro-load forecast.The present invention is to predict regional power load as research object, it is decomposed by timing, make full use of demand history data, the self-law of power load is sufficiently excavated, load prediction can be preferably carried out, compared with traditional single load forecasting method, model prediction accuracy proposed by the present invention is higher, and the changing rule of power load can be held from multiple angles, the flexibility of Electric Power Network Planning can be enhanced.
Description
Technical field
The electro-load forecast method based on multiple timings that the present invention relates to a kind of.
Background technique
Currently, as China's economic structure Continuous optimization adjusts, the continuous promotion of re-invent industry ability, electric load demand
And new variation also has occurred in Characteristics of Electric Load.On the one hand, the continuous propulsion and high pollution, highly energy-consuming type of industrial transformation upgrading
Enterprise gradually eliminates, and causes while electricity needs increases, biggish change will occur for network load demand and part throttle characteristics
Change.On the other hand, the appearance of some Novel electric loads is but also following can occur profound change, example with electrical characteristics as compared with the past
Such as, in recent years, wideling popularize, help due to country, electric car is in daily life by more and more extensive use, directly
Result in the charge requirement of the electric car quicklyd increase；Meanwhile promoting the popularization etc. of the energy storage technology of new energy consumption application,
These all introduce the novel load with two-way adjustability in power grid, bring new choose to the stable operation of power grid
War.
Currently, load prediction research still remains, analysis level is not careful enough, model application is not accurate enough, correlative factor is examined
Consider the problems such as not comprehensive enough, therefore, inside and outside never same space-time, power grid, the multi-angles such as social economy and policy set out, establish big number
Start with the electricity needs that this carries out fining from bottom to top according to braced frame, and based on this from user side and industry load prediction
Prediction, it is imperative further investigate to the development trend of electricity needs.
Summary of the invention
For the studies above status, the present invention proposes a kind of electro-load forecast method based on multiple timings, by pre-
Geodetic area power load, the comprehensive analysis for producing the output value, economic data and environmental data, will the prediction regional power load time
Sequence carries out timing decomposition；It is negative that linear regression, time series analysis, random forest etc. is respectively adopted for different when order components
Lotus prediction technique ,-Random Forest model can be added by establishing broad sense, final to realize multiple timings electro-load forecast.
The electro-load forecast method based on multiple timings that the technical solution of the present invention is to provide a kind of, it includes following steps
It is rapid:
S1, it collects and predicts regional power load, the production output value, economic conditions, the historical data of weather condition, line number of going forward side by side
According to pretreatment；
S2, wavelet decomposition is carried out to the historical data of power load, trend component, periodic component and random is obtained after decomposition
Component；
It is pre- that the different load forecasting method of S3, selection carries out load to trend component, periodic component and random component respectively
It surveys, obtains the result of electro-load forecast finally by superposition.
The present invention proposes a kind of electro-load forecast method based on multiple timings, has the following beneficial effects:
1. method of the invention has merged a variety of existing Methods of electric load forecasting, is decomposed, given full play to by timing
The advantages of various prediction techniques.
2. the present invention elaborates the data preprocessing method of electro-load forecast, history can be preferably handled
Data, improve data using degree, there is stronger practical value.
3. mentality of designing of the present invention is clear, usage mode is relatively simple, engineering in practice, have wide applicability.
Detailed description of the invention
Fig. 1 is the flow chart of the electro-load forecast method based on multiple timings；
Fig. 2 is X-12-ARIMA Program Elementary Stream journey figure；
Fig. 3 is random forest prediction model explanatory diagram；
Fig. 4 a~Fig. 4 d is the prediction result using the embodiment of electro-load forecast method of the present invention；
Fig. 5 is the user year load data wavelet decomposition result of above-described embodiment；
Fig. 6 is user's week load data wavelet decomposition result of above-described embodiment.
Specific embodiment
As shown in Figure 1, a kind of electro-load forecast method based on multiple timings, comprising the following steps:
S1, it collects and predicts regional power load, the production output value, economic conditions, the historical data of weather condition, line number of going forward side by side
According to pretreatment.
S2, wavelet decomposition is carried out to the historical data of power load, trend component, periodic component and random is obtained after decomposition
Component.
S3, in conjunction with the characteristics of traditional load forecasting method, choose different load forecasting methods to it is different when order components into
Row load prediction obtains the result of electro-load forecast finally by superposition.
In the step S1, the history number for predicting regional power load, the production output value, economic conditions, weather condition is collected
According to, and the pretreatment of data is carried out, specific steps dismantling is as follows:
Step S1-1: handling missing data, finds reasonable numerical value and removes replacement missing data, further increases number
According to quality.Since load data has stronger periodicity, for having lacked very multidata record some day, benefit
It is weighted processing with adjacent several days normal datas, is then filled to reduce error, specific formula is as follows:
X (d, t)=ω1x(d1,t)+ω2x(d2,t)
In formula, x (d, t) indicates t hours the d days corresponding load values；ω1x(d1, t) indicate t hours the d-1 days it is right
The load value answered；ω2x(d2, t) and indicate t hours the d+1 days corresponding load values.Wherein ω1=ω2=0.5 respectively indicates
The weight of the previous day and one day after load is (due to thinking the previous day and being one day after indifference to the embodiment on missing values date
, therefore take herein 0.5).
Step S1-2: handling data outliers (such as 0 value, mutation value), directly deletes the note comprising exceptional value
Record is smoothed exceptional value.The specific formula (reference can be made to formula of step S1-1) of smoothing processing is as follows:
X (d, t)=ω1x(d1,t)+ω2x(d2,t)
In formula, x (d, t) indicates t hours the d days corresponding load values；ω1x(d1, t) indicate t hours the d-1 days it is right
The load value answered；ω2x(d2, t) and indicate t hours the d+1 days corresponding load values.Wherein ω1=ω2=0.5 respectively indicates
The weight of the previous day and one day after load.
Step S1-3: the data of separate sources carry out characteristic criterion, the method standardized using section, by characteristic value
Numerical value is normalized to [0,1] section by the boundary value in section, and specific formula for calculation is as follows:
In formula, Xmin、XmaxIt is characterized the minimum value and maximum value of X, x respectivelyiIt is characterized some value of X, x'iFor standardization
Value afterwards.
In load prediction, the data of the separate sources, include: power-related data is provided by Utilities Electric Co.；Weather
Data are provided by weather site；Economic data is by statistics bureau's offer etc..
In the step S2, wavelet decomposition is carried out to the historical data of power load, trend component, period are obtained after decomposition
Component and random component.It makes a concrete analysis of as follows:
The continuous wavelet transform of signal x (t) is defined as:
In formula, a is scale factor, and small echo Φ is required to meet admissibility condition:
In formula,For the wavelet of selection, CΦFor the obtained result of the calculation formula.
Original signal can be replied out by the wavelet transformation of signal at this time, restores publicity are as follows:
In formula,For complex conjugate, CΨSubstitute into the calculated result of a formula.
The family of functions in T/F space that wavelet function Ψ (t) is generated through integer scaling and the translation of whole node, is constituted
Discrete wavelet.If f (t) ∈ L2It (R) is signal to be decomposed, L2(R) finite energy space is indicated,With
Ψ0,n={ Ψ (t-n) }n∈ZIt is V respectively0And W0Orthonormal basis；V0、W0It is the name of two groups of orthonormal basis；
Then had according to wavelet series expansion:
fNFor all scale frequency ingredients of signal to be decomposed；VjIt is VNIn a closed subspace；VNIt has been filled with L2(R)
The closed subspace sequence in entire space；Relative to fNFor, fJIn containing the frequency content lower than scale J, but without containing between ruler
The frequency content between J and N is spent, this is the ecotopia of low-pass filtering.Equally, WjBe contained only in f (t) frequency of scale j at
Point.Linear change component and high frequency random component in this way in information on load, after Wavelet transformation, frequency spectrum will present apparent
Separation characteristic.Multi-resolution decomposition is carried out to the regional historical load sequence wavelet transformation of prediction and Mallat algorithm, using approximation
Symmetrically, smooth compact schemes biorthogonal wavelet Daubechies function is as morther wavelet.The available trend point after decomposing
Amount, periodic component and random component.
In the step S3, in conjunction with the characteristics of traditional load forecasting method, different load forecasting methods is chosen to difference
When order components carry out load prediction, finally by superposition obtain the result of electro-load forecast.
Specific steps dismantling is as follows:
Step S3-1: the method that linear regression is used to trend component, using the Generalized Multivariate linear regression containing error term
Model are as follows:
yi=β0+β1xi1+β2xi2+…+βp-1xi,p-1+ei
In formula, i indicates the sequential value of load；β0…βp-1For regression coefficient；xi1…xi,p-1For influence factor constant, wide
In adopted multiple linear regression model, influence factor variable can be continuous quantity, discrete magnitude or indicator variable；eiFor obey N (0,
σ2) distribution stochastic variable.Both sides are same to take expectation that can obtain:
E (y)=β0+β1x1+β2x2+…+βp-1xp-1
In formula, x1…xp-1For p-1 influence factor variable, observation yiObedience desired value is E (y), variance σ2Just
State distribution.Load prediction is carried out using the formula, wherein y is prediction load value, x1…xp-1For influence load correlative factor,
Regression coefficient β1…βp-1It can be found out by the method for linear regression.
Step S3-2: periodic component is predicted using Time Series Analysis Model, the specific steps are as follows:
(1) stationarity of checking sequence carries out one or many difference to it, is translated into for non-stationary series
Stationary sequence, specific formula for calculation are as follows:
▽2yt=▽ (yt-yt-1)=yt-2yt-1+yt-2
In formula, ▽ is difference operator；ytIndicate the load value of t moment；yt-1Indicate the load value y at t-1 momentt-2Indicate t-
The load value at 2 moment.
(2) d rank homogeneous nonstationary time series yt, then there is ▽dytStationary time series, then can use from ARMA (p,
Q) model, i.e.,
λ(B)(▽dyt)=θ (B) εt
In formula, λ (B)=1- λ1B-λ2B2-…-λpBp, θ (B)=1- θ1B-θ2B2-…-θqBqRespectively autoregressive coefficient is more
Item formula and sliding average coefficient polynomial.εtFor zero-mean white noise sequence.
Step S3-3: random component is predicted using random forest prediction model, as shown in figure 3, specific steps are such as
Under:
(1) the record number for assuming initial data is N, and the trees number of random forest is k, the training set sample of each tree
Number is n, the training sample set for using Bootstrap resampling technique to randomly select k scale from raw data set N as n.
(2) assume there be T input variable, each node can randomly choose t (t < T) a specified variable, then transport
Optimal split point is determined with this t variable.In the entire generating process of decision tree, t value is to maintain invariable.
(3) each decision tree grows as much as possible, without carrying out beta pruning, and then constitute k decision tree form with
Machine forest.
(4) one new sample of every input, each decision tree inside forest can all be judged, judge that the sample is
Which kind of belongs to.For sorting algorithm, decision is eventually carried out according to the method that the minority is subordinate to the majority, and then for regression algorithm
Final prediction result is obtained using average weighted.
Step S3-4: it is overlapped to obtain final prediction result to by step S3-1 to the obtained result of step S3-3.
The invention will be further described with embodiment with reference to the accompanying drawing.
Step S1, which is collected, predicts regional power load, the production output value, economic conditions, the historical data of weather condition, goes forward side by side
The pretreatment of row data.Missing data accounting 2.13% before handling, exceptional value accounting 0.57% mentions significantly by data prediction
High data utilizabilitys.
Step S2 carries out wavelet decomposition to the historical data of power load, obtained after decomposition trend component, periodic component and
Random component.User year load data and year load data wavelet decomposition as a result, be shown in Fig. 5 and Fig. 6 respectively.It can by decomposition result
Know, load data has isolated trend component (a after wavelet decomposition6), periodic component (d6、d5、d4) and random component (d3、
d2、d1).The Decomposition order of this example is 6, and the reconstruction formula of entire signal is s=a6+d6+d5+d4+d3+d2+d1, i.e., this is several layers of
Signal is added.
Step S3: based on specific steps in above-mentioned steps S3, the development law useable linear model table of long-term trend component
The development law of sign, seasonal component can be characterized with ARIMA model characterization (Fig. 2), periodic component with nonparametric model, and random
Component then corrected by available random forest model.Obtained daily load prediction result is as shown in Figure 4 a- shown in Figure 4 d.Wherein for
The prediction of user 1 is relatively large in 10 point prediction error of the morning, mainly due to electricity consumption situation different from the past on the day of the user
Cause；It is more accurate for the prediction of user 2, the electricity consumption rule of user can be embodied, there are certain mistakes at the moment in night
Difference, this is mainly due to user's night load level is lower, and load variations influence prediction result big, it is difficult to Accurate Prediction；For
The prediction of user 3 15 up to 19 when there are certain error, mainly due to user power utilization regular fluctuation in the period it is larger and
Cause；Relative Error to each user is far below for the Relative Error of garden total load, this is because entirely
The load level of garden is relatively stable, fluctuates smaller.Obtained daily load prediction precision and all load prediction precision meters
The results are shown in Table 1 for calculation:
Power load day prediction result and weekly forecasting result of the table 1 based on multiple timings
In conclusion the present invention is to predict that regional power load as research object, is decomposed by timing, makes full use of load
Historical data has sufficiently excavated the self-law of power load, can preferably carry out load prediction, with traditional single load
Prediction technique is compared, and model prediction accuracy proposed by the present invention is higher, and the variation of power load can be held from multiple angles
The flexibility of Electric Power Network Planning can be enhanced in rule.
It is emphasized that example of the present invention be it is illustrative, without being restrictive, therefore the present invention is not
It is limited to example described in specific embodiment, its all obtained according to the technique and scheme of the present invention by those skilled in the art
His embodiment, also belongs to the scope of protection of the invention.
It is discussed in detail although the contents of the present invention have passed through above preferred embodiment, but it should be appreciated that above-mentioned
Description is not considered as limitation of the present invention.After those skilled in the art have read above content, for of the invention
A variety of modifications and substitutions all will be apparent.Therefore, protection scope of the present invention should be limited to the appended claims.
Claims (8)
1. a kind of electro-load forecast method based on multiple timings, which is characterized in that comprise the steps of:
S1, it collects and predicts regional power load, the production output value, economic conditions, the historical data of weather condition, and carry out data
Pretreatment；
S2, wavelet decomposition is carried out to the historical data of power load, trend component, periodic component are obtained after decomposition and divided at random
Amount；
The different load forecasting method of S3, selection carries out load prediction to trend component, periodic component and random component respectively, most
The result of electro-load forecast is obtained by superposition afterwards.
2. a kind of electro-load forecast method based on multiple timings as described in claim 1, which is characterized in that
In the step S1, the replacement to missing data includes following procedure:
Periodicity based on load data was weighted for one day of missing data using its adjacent several days normal data
Processing, is then filled, formula is as follows:
X (d, t)=ω1x(d1,t)+ω2x(d2,t)
In formula, x (d, t) indicates t hours the d days corresponding load values；ω1x(d1, t) indicate t hours the d-1 days it is corresponding
Load value；ω2x(d2, t) and indicate t hours the d+1 days corresponding load values；Wherein ω1=ω2=0.5 respectively indicate it is previous
It and the weight of load one day after.
3. a kind of electro-load forecast method based on multiple timings as described in claim 1, which is characterized in that
In the step S1, data outliers are handled, directly delete comprising exceptional value record or to exceptional value into
Row smoothing processing；The formula of smoothing processing is as follows:
X (d, t)=ω1x(d1,t)+ω2x(d2,t)
In formula, x (d, t) indicates t hours the d days corresponding load values；ω1x(d1, t) indicate t hours the d-1 days it is corresponding
Load value；ω2x(d2, t) and indicate t hours the d+1 days corresponding load values；Wherein ω1=ω2=0.5 respectively indicate it is previous
It and the weight of load one day after.
4. a kind of electro-load forecast method based on multiple timings as described in claim 1, which is characterized in that
In the step S1, the data of separate sources carry out characteristic criterion, by the boundary value in the section of characteristic value by numerical value
[0,1] section is normalized to, calculation formula is as follows:
In formula, Xmin、XmaxIt is characterized the minimum value and maximum value of X, x respectivelyiIt is characterized the value of X, x'iFor the value after standardization.
5. a kind of electro-load forecast method based on multiple timings as described in claim 1, which is characterized in that
In the step S2, wavelet decomposition is carried out to the historical data of power load, trend component, periodic component are obtained after decomposition
And random component, include following procedure:
The continuous wavelet transform of signal x (t) is defined as:
In formula, a is scale factor, and small echo Φ is required to meet admissibility condition:
For the wavelet of selection；
Original signal can be replied out by the wavelet transformation of signal at this time, restores publicity are as follows:
In formula,For complex conjugate；
Wavelet function Ψ (t) through integer scaling and whole node translation generate T/F space in family of functions, constitute from
Dissipate small echo；f(t)∈L2It (R) is signal to be decomposed, L2(R) finite energy space is indicated；And Ψ0,n={ Ψ
(t-n)}n∈ZIt is V respectively0And W0Orthonormal basis, then according to wavelet series expansion have:
wj∈Wj,0≤J≤N-1
fNFor all scale frequency ingredients of signal to be decomposed；VjIt is VNIn a closed subspace；VNIt has been filled with L2(R) entire
The closed subspace sequence in space；Relative to fNFor, fJIn containing the frequency content lower than scale J, but without containing between scale J
Frequency content between N；WjIt is the frequency content that scale j is contained only in f (t)；To the regional historical load sequence small echo of prediction
Transformation and Mallat algorithm carry out multi-resolution decomposition, using near symmetrical, smooth compact schemes biorthogonal wavelet Daubechies
Function is as morther wavelet；Trend component, periodic component and random component are obtained after decomposing.
6. a kind of electro-load forecast method based on multiple timings as described in claim 1, which is characterized in that
In the step S3, to the method that trend component uses linear regression, using the Generalized Multivariate linear regression containing error term
Model carries out load prediction:
yi=β0+β1xi1+β2xi2+…+βp-1xi,p-1+ei
In formula, i indicates the sequential value of load；β0…βp-1For regression coefficient；xi1…xi,p-1It is more in broad sense for influence factor constant
In first linear regression model (LRM), influence factor variable can be continuous quantity, discrete magnitude or indicator variable；eiTo obey N (0, σ2) point
The stochastic variable of cloth；Both sides are same to take expectation that can obtain:
E (y)=β0+β1x1+β2x2+…+βp-1xp-1
In formula, x1…xp-1For p-1 influence factor variable, observation yiObedience desired value is E (y), variance σ2Normal state point
Cloth；Wherein y is prediction load value, x1…xp-1For the correlative factor for influencing load, regression coefficient β1…βp-1By the side of linear regression
Method is found out.
7. a kind of electro-load forecast method based on multiple timings as described in claim 1, which is characterized in that
In the step S3, periodic component is predicted using Time Series Analysis Model, further includes following procedure:
The stationarity of checking sequence carries out one or many difference to it, is translated into steady sequence for non-stationary series
Column:
In formula,For difference operator；ytIndicate the load value of t moment；yt-1Indicate the load value at t-1 moment
yt-2Indicate the load value at t-2 moment；
For d rank homogeneous nonstationary time series yt,It is stationary time series, using from ARMA (p, q) model, i.e.,
In formula, λ (B)=1- λ1B-λ2B2-…-λpBp, θ (B)=1- θ1B-θ2B2-…-θqBqRespectively autoregressive coefficient multinomial
With sliding average coefficient polynomial；εtFor zero-mean white noise sequence.
8. a kind of electro-load forecast method based on multiple timings as described in claim 1, which is characterized in that
In step S3, random component is predicted using random forest prediction model, further includes following procedure:
(1) the record number of initial data is N, and the trees number of random forest is k, and the training set number of samples of each tree is n, is adopted
The training sample set that k scale is n is randomly selected from raw data set N with Bootstrap resampling technique；
(2) there is T input variable, each node randomly chooses t (t < T) a specified variable, then uses this t variable
To determine optimal split point；In the entire generating process of decision tree, t value is to maintain invariable；
(3) each decision tree grows as much as possible, and without carrying out beta pruning, and then it is random gloomy to constitute k decision tree composition
Woods；
(4) one new sample of every input, each decision tree inside forest can all be judged, judge that the sample is to belong to
Which kind of；For sorting algorithm, decision is carried out according to the method that the minority is subordinate to the majority, and regression algorithm is then added using average
Power obtains final prediction result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910534470.9A CN110264001A (en) | 2019-06-20 | 2019-06-20 | Electro-load forecast method based on multiple timings |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910534470.9A CN110264001A (en) | 2019-06-20 | 2019-06-20 | Electro-load forecast method based on multiple timings |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110264001A true CN110264001A (en) | 2019-09-20 |
Family
ID=67919590
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910534470.9A Pending CN110264001A (en) | 2019-06-20 | 2019-06-20 | Electro-load forecast method based on multiple timings |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110264001A (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111080486A (en) * | 2019-12-27 | 2020-04-28 | 新奥数能科技有限公司 | Electric load prediction method and device |
| CN111461496A (en) * | 2020-03-10 | 2020-07-28 | 国网山西省电力公司经济技术研究院 | Seasonal adjustment method for medium and long-term load data |
| CN111506635A (en) * | 2020-05-11 | 2020-08-07 | 上海积成能源科技有限公司 | System and method for analyzing residential electricity consumption behavior based on autoregressive naive Bayes algorithm |
| CN111583065A (en) * | 2020-05-12 | 2020-08-25 | 广东电网有限责任公司计量中心 | Power load data prediction method and device |
| CN111667377A (en) * | 2020-04-24 | 2020-09-15 | 南方电网能源发展研究院有限责任公司 | Full-automatic power consumption prediction method and system |
| CN111754037A (en) * | 2020-06-19 | 2020-10-09 | 国网河南省电力公司经济技术研究院 | Long-term load hybrid prediction method for regional terminal integrated energy supply system |
| CN112232575A (en) * | 2020-10-21 | 2021-01-15 | 国网山西省电力公司经济技术研究院 | Comprehensive energy system regulation and control method and device based on multivariate load prediction |
| CN112269059A (en) * | 2020-09-27 | 2021-01-26 | 深圳供电局有限公司 | Electric mesh belt load testing method and device, computer equipment and storage medium |
| CN112381266A (en) * | 2020-10-22 | 2021-02-19 | 国网湖北省电力有限公司武汉供电公司 | System and method for predicting future power supply amount based on historical power supply and weather data |
| CN112651545A (en) * | 2020-12-07 | 2021-04-13 | 华电郑州机械设计研究院有限公司 | Industrial heat load prediction method based on characteristic decomposition |
| CN112989266A (en) * | 2019-12-02 | 2021-06-18 | 阿里巴巴集团控股有限公司 | Periodicity detection and cycle length estimation in a time series |
| CN113743665A (en) * | 2021-09-03 | 2021-12-03 | 国网湖南省电力有限公司 | Power load prediction method based on typical user group of power supply grid |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105404939A (en) * | 2015-12-04 | 2016-03-16 | 河南许继仪表有限公司 | Short-term power load prediction method |
| CN109376911A (en) * | 2018-09-28 | 2019-02-22 | 上海电力学院 | A kind of micro-capacitance sensor short-term load forecasting method based on EMD-KELMs-SDPSO |
| CN109583635A (en) * | 2018-11-16 | 2019-04-05 | 贵州电网有限责任公司 | A kind of short-term load forecasting modeling method towards operational reliability |
| CN109583621A (en) * | 2018-10-12 | 2019-04-05 | 中国农业大学 | A kind of PSO-LSSVM short-term load forecasting method based on improvement variation mode decomposition |
| US20190171494A1 (en) * | 2017-12-04 | 2019-06-06 | Cisco Technology, Inc. | Cost-optimal cluster configuration analytics package |
-
2019
- 2019-06-20 CN CN201910534470.9A patent/CN110264001A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105404939A (en) * | 2015-12-04 | 2016-03-16 | 河南许继仪表有限公司 | Short-term power load prediction method |
| US20190171494A1 (en) * | 2017-12-04 | 2019-06-06 | Cisco Technology, Inc. | Cost-optimal cluster configuration analytics package |
| CN109376911A (en) * | 2018-09-28 | 2019-02-22 | 上海电力学院 | A kind of micro-capacitance sensor short-term load forecasting method based on EMD-KELMs-SDPSO |
| CN109583621A (en) * | 2018-10-12 | 2019-04-05 | 中国农业大学 | A kind of PSO-LSSVM short-term load forecasting method based on improvement variation mode decomposition |
| CN109583635A (en) * | 2018-11-16 | 2019-04-05 | 贵州电网有限责任公司 | A kind of short-term load forecasting modeling method towards operational reliability |
Non-Patent Citations (8)
| Title |
|---|
| JV HANSEN 等: "Forecasting and recombining time-series components by using neural networks", 《JOOURNAL OF THE OPERATIONAL RESEARCH SOCIETY》 * |
| LEO BREIMAN: "Random Forests", 《MACHINE LEARNING》 * |
| 夏丽: "基于ARIMA模型及回归分析的区域用电量预测方法研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
| 曹正凤: "随机森林算法优化研究", 《中国博士学位论文全文数据库 社会科学Ⅱ辑》 * |
| 王凌谊 等: "融合日期类型的改进线性回归短期负荷预测模型", 《广东电力》 * |
| 董师师 等: "随机森林理论浅析", 《集成技术》 * |
| 邰能灵 等: "基于小波分析的电力系统短期负荷预测方法", 《中国电机工程学报》 * |
| 马哲 等: "基于ESPRIT分解算法的短期电力负荷预测", 《电力系统保护与控制》 * |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112989266A (en) * | 2019-12-02 | 2021-06-18 | 阿里巴巴集团控股有限公司 | Periodicity detection and cycle length estimation in a time series |
| CN111080486A (en) * | 2019-12-27 | 2020-04-28 | 新奥数能科技有限公司 | Electric load prediction method and device |
| CN111461496A (en) * | 2020-03-10 | 2020-07-28 | 国网山西省电力公司经济技术研究院 | Seasonal adjustment method for medium and long-term load data |
| CN111667377A (en) * | 2020-04-24 | 2020-09-15 | 南方电网能源发展研究院有限责任公司 | Full-automatic power consumption prediction method and system |
| CN111506635A (en) * | 2020-05-11 | 2020-08-07 | 上海积成能源科技有限公司 | System and method for analyzing residential electricity consumption behavior based on autoregressive naive Bayes algorithm |
| CN111583065A (en) * | 2020-05-12 | 2020-08-25 | 广东电网有限责任公司计量中心 | Power load data prediction method and device |
| CN111583065B (en) * | 2020-05-12 | 2023-08-22 | 广东电网有限责任公司计量中心 | Power load data prediction method and device |
| CN111754037B (en) * | 2020-06-19 | 2023-01-20 | 国网河南省电力公司经济技术研究院 | Long-term load hybrid prediction method for regional terminal integrated energy supply system |
| CN111754037A (en) * | 2020-06-19 | 2020-10-09 | 国网河南省电力公司经济技术研究院 | Long-term load hybrid prediction method for regional terminal integrated energy supply system |
| CN112269059A (en) * | 2020-09-27 | 2021-01-26 | 深圳供电局有限公司 | Electric mesh belt load testing method and device, computer equipment and storage medium |
| CN112269059B (en) * | 2020-09-27 | 2023-06-06 | 深圳供电局有限公司 | Power grid load test method and device, computer equipment and storage medium |
| CN112232575A (en) * | 2020-10-21 | 2021-01-15 | 国网山西省电力公司经济技术研究院 | Comprehensive energy system regulation and control method and device based on multivariate load prediction |
| CN112232575B (en) * | 2020-10-21 | 2023-12-19 | 国网山西省电力公司经济技术研究院 | Comprehensive energy system regulation and control method and device based on multi-element load prediction |
| CN112381266A (en) * | 2020-10-22 | 2021-02-19 | 国网湖北省电力有限公司武汉供电公司 | System and method for predicting future power supply amount based on historical power supply and weather data |
| CN112381266B (en) * | 2020-10-22 | 2024-01-09 | 国网湖北省电力有限公司武汉供电公司 | System and method for predicting future power supply quantity based on historical power supply and weather data |
| CN112651545A (en) * | 2020-12-07 | 2021-04-13 | 华电郑州机械设计研究院有限公司 | Industrial heat load prediction method based on characteristic decomposition |
| CN113743665A (en) * | 2021-09-03 | 2021-12-03 | 国网湖南省电力有限公司 | Power load prediction method based on typical user group of power supply grid |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110264001A (en) | Electro-load forecast method based on multiple timings | |
| CN108846530B (en) | Short-term load prediction method based on clustering-regression model | |
| CN110503256B (en) | Short-term load prediction method and system based on big data technology | |
| CN106503851A (en) | A kind of improved Short-Term Load Forecasting Method based on wavelet analysises | |
| CN110991700A (en) | Weather and electricity utilization correlation prediction method and device based on deep learning improvement | |
| CN114676883A (en) | Power grid operation management method, device and equipment based on big data and storage medium | |
| Sigauke | Forecasting medium-term electricity demand in a South African electric power supply system | |
| CN111680841B (en) | Short-term load prediction method, system and terminal equipment based on principal component analysis | |
| CN114372360A (en) | Method, terminal and storage medium for power load prediction | |
| CN115688993A (en) | Short-term power load prediction method suitable for power distribution station area | |
| CN111600300A (en) | Robust optimization scheduling method considering wind power multivariate correlation ellipsoid set | |
| CN114742302A (en) | Method for inverting festival and holiday domestic garbage yield based on LSTM multivariable time sequence prediction | |
| Zhang et al. | Load prediction based on hybrid model of VMD-mRMR-BPNN-LSSVM | |
| He et al. | High frequency short-term demand forecasting model for distribution power grid based on ARIMA | |
| Aghajani et al. | Short-term prediction of market-clearing price of electricity in the presence of wind power plants by a hybrid intelligent system | |
| CN113837434A (en) | Solar photovoltaic power generation prediction method and device, electronic equipment and storage medium | |
| Sun et al. | Short-term power load prediction based on VMD-SG-LSTM | |
| Qin et al. | Forecasting of China consumer price index based on EEMD and SVR method | |
| CN116258269A (en) | Ultra-short-term load dynamic prediction method based on load characteristic decomposition | |
| CN115860797A (en) | Electric quantity demand prediction method suitable for new electricity price reform situation | |
| CN113780655A (en) | Steel multi-variety demand prediction method based on intelligent supply chain | |
| Zhao et al. | Internet-of-thing based real-time electrical market monitoring system design | |
| CN110991748A (en) | Short-term load prediction method for urban power grid | |
| CN117578465B (en) | Multi-scale interpretable micro-grid power load prediction method and system | |
| CN111368257A (en) | Method and device for analyzing and predicting coal-to-electricity load characteristics |
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: 20190920 |