CN107038502A - Consider the improvement wavelet packet electricity demand forecasting method of Seasonal Characteristics - Google Patents
Consider the improvement wavelet packet electricity demand forecasting method of Seasonal Characteristics Download PDFInfo
- Publication number
- CN107038502A CN107038502A CN201710251850.2A CN201710251850A CN107038502A CN 107038502 A CN107038502 A CN 107038502A CN 201710251850 A CN201710251850 A CN 201710251850A CN 107038502 A CN107038502 A CN 107038502A
- Authority
- CN
- China
- Prior art keywords
- seasonal
- power consumption
- season
- index number
- demand forecasting
- 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
- 230000001932 seasonal effect Effects 0.000 title claims abstract description 101
- 230000005611 electricity Effects 0.000 title claims abstract description 51
- 238000013277 forecasting method Methods 0.000 title claims abstract description 17
- 230000006872 improvement Effects 0.000 title claims abstract description 17
- 239000011159 matrix material Substances 0.000 claims abstract description 22
- 230000015572 biosynthetic process Effects 0.000 claims abstract description 18
- 238000003786 synthesis reaction Methods 0.000 claims abstract description 18
- 238000003062 neural network model Methods 0.000 claims abstract description 6
- 238000000034 method Methods 0.000 claims description 29
- 238000000354 decomposition reaction Methods 0.000 claims description 11
- 230000007935 neutral effect Effects 0.000 claims description 9
- 238000004458 analytical method Methods 0.000 claims description 7
- PEDCQBHIVMGVHV-UHFFFAOYSA-N Glycerine Chemical group OCC(O)CO PEDCQBHIVMGVHV-UHFFFAOYSA-N 0.000 claims description 5
- 230000004913 activation Effects 0.000 claims description 4
- 230000001186 cumulative effect Effects 0.000 claims description 4
- 230000000694 effects Effects 0.000 claims description 4
- 230000009466 transformation Effects 0.000 claims description 3
- 238000010276 construction Methods 0.000 claims 1
- 230000008859 change Effects 0.000 description 4
- 210000002569 neuron Anatomy 0.000 description 4
- 230000009467 reduction Effects 0.000 description 3
- 238000012937 correction Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000004378 air conditioning Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000013213 extrapolation Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000002715 modification method Methods 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000011946 reduction process Methods 0.000 description 1
- 238000002922 simulated annealing Methods 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- 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
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Software Systems (AREA)
- Marketing (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a kind of improvement wavelet packet electricity demand forecasting method for considering Seasonal Characteristics, it is that continuous two-dimentional electricity consumption moment matrix is subjected to denoising, fits regular stronger electric quantity data;Power consumption is predicted based on radial basis function neural network model simultaneously, power consumption Seasonal Characteristics obvious season is recognized by introducing seasonal index number, to seasonal obvious monthly power consumption, is predicted using synthesis in many months, then be divided in portion to correct monthly predicted value.Using above-mentioned technical proposal, contribute to the demand of electricity demand forecasting, electricity demand forecasting accuracy rate precision can be effectively improved;The monthly precision of prediction in crest paddy season can be effectively improved, overall precision of prediction high target is reached.
Description
Technical field
The invention belongs to the technical field of electric power market demand prediction, it is related to a kind of improvement wavelet packet for considering Seasonal Characteristics
Electricity demand forecasting method, can effectively improve the monthly precision of prediction in crest paddy season.
Background technology
As China's electricity market is further reasonable, perfect, power industry need to analyze investment demand trend, so as to adjust company
Development strategy, adapt to electricity change rear complexity electricity transaction market environment thus the quality and accuracy of electricity demand forecasting are proposed
Higher standard, it is desirable to accomplish the electricity demand forecasting to 12 month of next year, could strategically arrange investment and financing plan, be electricity
Force system planning provides strong support.
The meteorologic factors such as temperature, humidity, wind speed can produce randomness interference to the fluctuation of monthly power consumption, this to be referred to as
The interference of noise acts on power consumption in time domain and frequency domain, and the regularity and continuity of electricity fluctuation are found when predicting power consumption
Influence is all produced, error is brought.Noise reduction pretreatment is carried out to initial data, fluctuation, the continuity etc. of power consumption are reduced as far as possible
Rule turns into an important step of power quantity predicting.In addition, seasonal climate.The factors such as festivals or holidays can be produced to monthly power consumption
Scale influences, but noise reduction process can underestimate season Wave crest and wave trough, it is necessary to carry out power consumption amendment.
There is randomness and complexity because middle or short term measures change with list, but have cyclic swing, when horizontal and vertical
Between put it is continuous.Classical forecast and modification method such as linear regression method, sliding time predicted method and simulated annealing etc. can not be adapted to,
The noise in signal can not be accurately distinguished, the amendment work of prediction power consumption is have impact on, so that final predicted value does not reach standard
The requirement of true rate.
The content of the invention
The present invention provides a kind of improvement wavelet packet electricity demand forecasting method for considering Seasonal Characteristics, and the purpose is to effectively improve
The monthly precision of prediction in crest paddy season.
To achieve these goals, the technical scheme taken of the present invention is:
The improvement wavelet packet electricity demand forecasting method of the consideration Seasonal Characteristics of the present invention, described Forecasting Methodology is will be continuous
Two-dimentional electricity consumption moment matrix carry out denoising, fit regular stronger electric quantity data;Simultaneously based on RBF
(RBF) Neural Network model predictive power consumption, recognizes power consumption Seasonal Characteristics obvious season, to season by introducing seasonal index number
The obvious monthly power consumption of section property, was predicted, then be divided in portion to correct monthly predicted value using synthesis in many months.
Wavelet transformation is compared in view of wavelet package transforms, more strongly can be analyzed in time-domain and frequency-domain, and because electricity
Horizontal and vertical continuity, it is laterally upper that same month data line up, on longitudinal direction by adjacent month data beats in column, progress normalizing
Change, form the two-dimensional matrix of similar gray level image matrix.
According to the two-dimensional matrix, introduce the 2-d wavelet bag originating from graphical analysis and decompose, effective letter is distinguished with threshold value
Number and noise, power consumption data are handled using the denoising method based on 2-d wavelet bag threshold value.
The Seasonal Characteristics consider that the power consumption of spring and summer is in paddy peak value, and wavelet packet denoising can eliminate Seasonal Characteristics,
Thus selection RBF- synthesis decomposition method is predicted and corrects power consumption, and pre- by the cumulative carry out of seasonal characteristic month and its adjacent moon
Survey, then be decomposed into each monthly data, so as to correct power consumption season Wave crest and wave trough prediction data.
Described Forecasting Methodology is feed-forward type neutral net using RBF (RBF) neutral net, and structure is comprising hidden
Connection weight between number containing node layer, the center of each activation primitive, width and hidden layer and output layer.
Described Forecasting Methodology includes step in detail below:
1), data are by behavior 12 months 1 year, and be classified as not the same year lines up two-dimensional matrix identical month, and is normalized
Processing;
2), decomposed using 2-d wavelet bag, wavelet basis function takes db4, Decomposition order is 3, and entropy standard takes Shannon, structure
Make optimal tree;
3) denoising, is carried out using hyperbola threshold function table, and is reconstructed, renormalization, and with using hard -threshold, soft threshold
The denoising effect contrast of value function;
4), denoising data are carried out with RBF predictions, and predicts the outcome with the RBF of non-denoising data and to make accuracy rate and definitely by mistake
Difference contrast;
5), the power that seasonal index number judges Seasonal Characteristics is calculated;If seasonal strong, predicted using the amendment of synthesis decomposition method
Data.
In described step 5) in, the amendment prediction data method is comprised the steps of:
A), the sequence after wavelet packet denoising is predicted using based on RBF (RBF);
B) seasonal index number, is calculated, judges whether to need to correct electricity demand forecasting result;
C) synthesis revised law, is used to the seasonal index number season larger with average seasonal index number relative deviation.
Described Seasonal Characteristics are stated using seasonal index number, and seasonal index number is defined as to actual value and the season in the season
The ratio between the average value at place year;Set N to take the time span of recent 3~5 years, realize the renewal with power consumption historical data, season
Section index, which also occurs to roll, to be updated, and will be divided into K sections season, note Sij is the season power consumption actual value in 1 year jth season, Dan Nianji
The season power consumption in section 1 year jth season of exponential representation accounts for the ratio of 1 year Urban Annual Electrical Power Consumption amount, but annual seasons index is:
In formula:I=1,2 ..., N;J=1,2 ..., K;
Seasonal index number Fj is averaged for N single annual seasons index, i.e.,:
In formula:J=1,2 ..., K;
Average seasonal index numberFor the average value in K season, i.e.,:
In formula:J=1,2 ..., K;
The relative deviation of seasonal index number and average seasonal index number is:
The present invention uses above-mentioned technical proposal, is carried out continuous two-dimentional electricity consumption moment matrix at denoising using wavelet packet analysis
Reason, fits regular stronger electric quantity data, contributes to the demand of electricity demand forecasting, can effectively improve electricity demand forecasting accurate
True rate precision;Simultaneously based on RBF (RBF) Neural Network model predictive power consumption, by introducing seasonal index number identification use
In electricity Seasonal Characteristics obvious season, to seasonal obvious monthly power consumption, predicted using synthesis in many months, then be divided in portion
To correct monthly predicted value, the monthly precision of prediction in crest paddy season can be effectively improved, overall precision of prediction high mesh is reached
Mark.
Brief description of the drawings
Fig. 1 predicts flow chart for the wavelet packet electricity demand forecasting method of the consideration Seasonal Characteristics of the present invention;
Fig. 2 predicts the outcome for the power consumption combined value of the 1-3 months in 2015;
Fig. 3 is the prediction power consumption and actual power consumption curve map in January, 2011 in June, 2015.
Embodiment
Below against accompanying drawing, by the description to embodiment, the embodiment to the present invention makees further details of
Illustrate, to help those skilled in the art to have more complete, accurate and deep reason to inventive concept of the invention, technical scheme
Solution.
Technical scheme as expressed by Fig. 1, is that a kind of improvement wavelet packet power consumption for considering Seasonal Characteristics is pre-
The pre- flow gauge of survey method.For the deficiency being previously mentioned in background technology, the present invention proposes a kind of the small of consideration Seasonal Characteristics
Ripple bag electricity demand forecasting method, introduces wavelet packet analysis and continuous two-dimentional electricity consumption moment matrix is carried out into denoising, introduce simultaneously
Seasonal index number recognizes power consumption Seasonal Characteristics obvious season, effectively improves the monthly precision of prediction in crest paddy season, solves to pass
The problem of Forecasting Methodology precision of prediction of uniting is low.
In order to solve the problem of prior art is present and overcome its defect, realization effectively improves the monthly pre- of crest paddy season
The goal of the invention of precision is surveyed, the technical scheme that the present invention takes is:
As shown in Figure 1 to Figure 3, the improvement wavelet packet electricity demand forecasting method of consideration Seasonal Characteristics of the invention, described
Forecasting Methodology is that continuous two-dimentional electricity consumption moment matrix is carried out into denoising, fits regular stronger electric quantity data;Simultaneously
Based on RBF (RBF) Neural Network model predictive power consumption, power consumption Seasonal Characteristics are recognized by introducing seasonal index number
In obvious season, to seasonal obvious monthly power consumption, predicted using synthesis in many months, then be divided in portion monthly pre- to correct
Measured value.
The present invention solves the problem of Traditional Wavelet bag denoising forecast model can remove Seasonal Characteristics in historical data, the party
Continuous two-dimentional electricity consumption moment matrix is carried out denoising by method using wavelet packet analysis, fits regular stronger electricity number
According to contributing to the demand of electricity demand forecasting, electricity demand forecasting accuracy rate precision can be effectively improved;Simultaneously based on RBF
(RBF) Neural Network model predictive power consumption, recognizes power consumption Seasonal Characteristics obvious season, to season by introducing seasonal index number
The obvious monthly power consumption of section property, was predicted using synthesis in many months, then was divided in portion to correct monthly predicted value, can be effectively improved
The monthly precision of prediction in crest paddy season, reaches overall precision of prediction high target.
Signal is carried out with wavelet package transforms same month power consumption data lining up in denoising, transverse direction, on longitudinal direction
Adjacent month data are arranged in columns, it is normalized, forms the two-dimensional matrix of similar gray level image matrix, selection RBF- synthesis point
Solution is predicted and corrects power consumption, i.e., predicted using RBF, and is predicted by seasonal characteristic month and its cumulative of the adjacent moon,
Be decomposed into again it is each monthly, so as to correct power consumption season Wave crest and wave trough prediction data.
Wavelet transformation is compared in view of wavelet package transforms, more strongly can be analyzed in time-domain and frequency-domain, and because electricity
Horizontal and vertical continuity, it is laterally upper that same month data line up, on longitudinal direction by adjacent month data beats in column, progress normalizing
Change, form the two-dimensional matrix of similar gray level image matrix.
According to the two-dimensional matrix, introduce the 2-d wavelet bag originating from graphical analysis and decompose, effective letter is distinguished with threshold value
Number and noise, power consumption data are handled using the denoising method based on 2-d wavelet bag threshold value.
The Seasonal Characteristics consider that the power consumption of spring and summer is in paddy peak value, and wavelet packet denoising can eliminate Seasonal Characteristics,
Thus selection RBF- synthesis decomposition method is predicted and corrects power consumption, and pre- by the cumulative carry out of seasonal characteristic month and its adjacent moon
Survey, then be decomposed into each monthly data, so as to correct power consumption season Wave crest and wave trough prediction data.
Described Forecasting Methodology is feed-forward type neutral net using RBF (RBF) neutral net, and structure is comprising hidden
Connection weight between number containing node layer, the center of each activation primitive, width and hidden layer and output layer.
Described Forecasting Methodology includes step in detail below:
1), data are by behavior 12 months 1 year, and be classified as not the same year lines up two-dimensional matrix identical month, and is normalized
Processing;
2), decomposed using 2-d wavelet bag, wavelet basis function takes db4, Decomposition order is 3, and entropy standard takes Shannon, structure
Make optimal tree;
3) denoising, is carried out using hyperbola threshold function table, and is reconstructed, renormalization, and with using hard -threshold, soft threshold
The denoising effect contrast of value function;
4), denoising data are carried out with RBF predictions, and predicts the outcome with the RBF of non-denoising data and to make accuracy rate and definitely by mistake
Difference contrast;
5), the power that seasonal index number judges Seasonal Characteristics is calculated;If seasonal strong, predicted using the amendment of synthesis decomposition method
Data.
In described step 5) in, the amendment prediction data method is comprised the steps of:
A), the sequence after wavelet packet denoising is predicted using based on RBF (RBF);
B) seasonal index number, is calculated, judges whether to need to correct electricity demand forecasting result;
C) synthesis revised law, is used to the seasonal index number season larger with average seasonal index number relative deviation.
Step 5) specific practice:
(1) add up and synthesize the monthly power consumption in the season, form season power consumption, deviate it to eliminate this stage power consumption
Remaining month larger feature;
(2) gray prediction is used to the season power consumption over the years, obtains predicting the predicted value of annual seasons power consumption;
(3) proportion, progress historical trend extrapolation in the season power consumption according to deviation over the years month, obtains predicting year
The prediction ratio value of deviation month;
(4) in season power consumption, prediction year, the monthly seasonal forecasting value for adding sum subtracted the first month predicted value, current to reflect
Season electricity consumption measure feature;Then remaining each month proportional assignment prediction power consumption, obtains the monthly electricity demand forecasting in the season
Value, as correction value.
The Seasonal Characteristics of the power consumption are stated using seasonal index number, and seasonal index number is defined as to the actual value in the season
With the ratio between the average value in year where season;
Provided with N power consumption data, N takes the time span of recent 3~5 years, realizes the renewal with power consumption historical data,
Seasonal index number, which also occurs to roll, to be updated;K sections will be divided into season, note Sij is the season power consumption actual value in 1 year jth season, Dan Nian
Seasonal index number represents that the season power consumption in 1 year jth season accounts for the ratio of 1 year Urban Annual Electrical Power Consumption amount, but annual seasons index is:
In formula:I=1,2 ..., N;J=1,2 ..., K;
Seasonal index number Fj is averaged for N single annual seasons index, i.e.,:
In formula:J=1,2 ..., K;
Average seasonal index numberFor the average value in K season, i.e.,:
In formula:J=1,2 ..., K;
The relative deviation of seasonal index number and average seasonal index number is:
Step 5) prediction uses RBF (RBF) neutral net to be feed-forward type neutral net, if being wrapped in neutral net
Containing h hidden layer neuron, then m is that the corresponding networks of input vector x=(x1, x2 ..., xm) T are output as:
Wherein, Si is the connection weight between i-th of hidden layer neuron and output layer neuron;Gi (x) is hidden i-th
The activation primitive of the neuron containing layer.
Continuous two-dimentional electricity consumption moment matrix is subjected to denoising present invention introduces wavelet packet analysis, referred to while introducing season
In number identification power consumption Seasonal Characteristics obvious season, effectively improve the monthly precision of prediction in crest paddy season.
Here is to use the above method, and saving Analyzing Total Electricity Consumption in monthly, 2011~2014 historical data to Z goes
The example made an uproar, predicted and correct, further illustrates the specific implementation method of the present invention.
Step (1), takes totally 4 × 12 data in January, 2011 in December, 2015, data are by behavior 12 months 1 year, row
Two-dimensional matrix is lined up identical month for not the same year, and is normalized;
Step (2), is decomposed using 2-d wavelet bag, and wavelet basis function takes db4, and Decomposition order is 3, and entropy standard takes
Shannon, constructs optimal tree;
Step (3), denoising is carried out using hyperbola threshold function table, and is reconstructed, renormalization, and with using hard threshold
The denoising effect contrast of value, soft-threshold function, using the denoising index such as table 1 of three kinds of threshold function tables;
Table 1:
Denoising data are carried out RBF predictions by step (4), and are predicted the outcome with the RBF of non-denoising data and made accuracy rate and mistake
Difference contrast, using the electricity demand forecasting accuracy comparison such as table 2 of distinct methods;
Table 2:
Note:2* is represented it should be noted that being predicted the power consumption relative error of 2 months 2015 using wavelet packet denoising-RBF in table 2
12.78%, error is far above remaining month, it is necessary to be predicted value amendment.
It is visible in table 2:Monthly power consumption is predicted using RBF neural relative to direct, gone using wavelet packet
The monthly mean accuracy for carrying out RBF predictions after making an uproar again improves 2% or so.
Step (5), calculates the power that seasonal index number judges Seasonal Characteristics, Jiang Genian was divided into for 4 season, by 2011-2014
Year calculate seasonal index number Fj, and with average seasonal index numberRelative deviation Δ F is calculated, if relative deviation is larger, illustrates the season
Substantially, power consumption will have larger fluctuation to the Seasonal Characteristics of section.Result of calculation is as shown in table 3.
Table 3:
By the relative deviation Δ F that calculates for 2011~2014 years as can be seen that the first quarter, the third quarter relative deviation it is remote
Higher than the 2nd, fourth quarter, that is, the 1st, the average seasonal index number of seasonal index number deviation in the third quarter is larger.This explanation the 1st, the season in the third quarter
Characteristic is more apparent, then when predicting 2015, and the electricity demand forecasting value to the 1, third quarter need to use synthesis revised law to be modified.
To the amendment of predicted value by taking the first quarter as an example, for 2015, there are Seasonal Characteristics in the first quarter in power consumption,
On the one hand be due to the first quarter temperature it is relatively low, each industry to air-conditioning rely on reduce, power consumption reduce;On the other hand spring in 2015
Save and 2 months, returned to one's native place by population from other places and form the reduction of population inside the province, cause the reduction of power consumption.
The power consumption combined value of the 1-3 months in 2015 is predicted using grey forecasting model, as a result as shown in Figure 2.
The 1-3 months in 2015 synthesize predicted value error -0.64%, and precision of prediction is higher.
Then power consumption accounted for composite value then and obtained ratio the moon where calculating each Spring Festival in year, is averaged and obtained according to trend
Ratio is 24.62% within 2 months 2015, and March is 38.83%;January was subtracted with the kW.h of 1-3 month predicted values 76,877,930,000 in 2015
The kW.h of RBF predicted values 30,919,750,000, then by 2,2 months kW.h of correction value 17,831,600,000 of prediction pro rate in March, before and after amendment
Compare as shown in table 4.
Table 4:
Correct within 2 months 2015 | Actual value/(ten thousand kW.h) | Predicted value/(ten thousand kW.h) | Relative error/% |
Before amendment | 1765218 | 1990781 | 12.78 |
After amendment | 1765218 | 1783160 | 1.02 |
The prediction power consumption and actual power consumption in January, 2011 in June, 2015 are as shown in Figure 3.
From the figure 3, it may be seen that bright in the seasonal index number identification Seasonal Characteristics first quarter in 2015 calculated according to 2011~2014 years
After aobvious, using synthesis decomposition method amendment predicted value, the relative error of predicted value can be efficiently reduced, the essence of monthly prediction is improved
Exactness.
The present invention is exemplarily described above in conjunction with accompanying drawing, it is clear that the present invention is implemented not by aforesaid way
Limitation, as long as the improvement of the various unsubstantialities of inventive concept and technical scheme of the present invention progress is employed, or without changing
Enter and the design of the present invention and technical scheme are directly applied into other occasions, within protection scope of the present invention.
Claims (8)
1. consider the improvement wavelet packet electricity demand forecasting method of Seasonal Characteristics, it is characterised in that:Described Forecasting Methodology is by even
Continuous two-dimentional electricity consumption moment matrix carries out denoising, fits regular stronger electric quantity data;Simultaneously based on RBF
Neural Network model predictive power consumption, recognizes power consumption Seasonal Characteristics obvious season, to seasonality by introducing seasonal index number
Obvious monthly power consumption, was predicted, then be divided in portion to correct monthly predicted value using synthesis in many months.
2. according to the improvement wavelet packet electricity demand forecasting method of the consideration Seasonal Characteristics described in claim 1, it is characterised in that:Have
In view of wavelet package transforms compare wavelet transformation, more strongly can be analyzed in time-domain and frequency-domain, and because electricity transverse direction and indulge
To continuity, same month data are lined up in transverse direction, on longitudinal direction by adjacent month data beats in column, are normalized, class is formed
The two-dimensional matrix of ashy degree image array.
3. according to the improvement wavelet packet electricity demand forecasting method of the consideration Seasonal Characteristics described in claim 2, it is characterised in that:Root
According to the two-dimensional matrix, introduce the 2-d wavelet bag originating from graphical analysis and decompose, useful signal and noise are distinguished with threshold value, adopted
Power consumption data are handled with the denoising method based on 2-d wavelet bag threshold value.
4. according to the improvement wavelet packet electricity demand forecasting method of the consideration Seasonal Characteristics described in claim 1, it is characterised in that:Institute
State Seasonal Characteristics and be in paddy peak value in view of the power consumption of spring and summer, wavelet packet denoising can eliminate Seasonal Characteristics, thus selection
RBF- synthesis decomposition method is predicted and corrects power consumption, and is predicted by seasonal characteristic month and its cumulative of the adjacent moon, then divides
Solve as each monthly data, so as to correct power consumption season Wave crest and wave trough prediction data.
5. according to the improvement wavelet packet electricity demand forecasting method of the consideration Seasonal Characteristics described in claim 1, it is characterised in that:Institute
The Forecasting Methodology stated using RBF (RBF) neutral net be feed-forward type neutral net, structure comprising node in hidden layer,
Connection weight between the center of each activation primitive, width and hidden layer and output layer.
6. according to the improvement wavelet packet electricity demand forecasting method of the consideration Seasonal Characteristics described in claim 1, it is characterised in that:Institute
The Forecasting Methodology stated includes step in detail below:
1), data are by behavior 12 months 1 year, and be classified as not the same year lines up two-dimensional matrix identical month, and place is normalized
Reason;
2), decomposed using 2-d wavelet bag, wavelet basis function takes db4, Decomposition order is 3, entropy standard takes Shannon, and construction is most
Select tree;
3) denoising, is carried out using hyperbola threshold function table, and is reconstructed, renormalization, and with using hard -threshold, soft-threshold letter
Several denoising effect contrasts;
4), denoising data are carried out with RBF predictions, and is predicted the outcome with the RBF of non-denoising data and makees accuracy rate and absolute error pair
Than;
5), the power that seasonal index number judges Seasonal Characteristics is calculated;If seasonal strong, synthesis decomposition method amendment prediction number is used
According to.
7. according to the improvement wavelet packet electricity demand forecasting method of the consideration Seasonal Characteristics described in claim 6, it is characterised in that:
Described step 5) in, the amendment prediction data method is comprised the steps of:
A), the sequence after wavelet packet denoising is predicted using based on RBF;
B) seasonal index number, is calculated, judges whether to need to correct electricity demand forecasting result;
C) synthesis revised law, is used to the seasonal index number season larger with average seasonal index number relative deviation.
8. according to the improvement wavelet packet electricity demand forecasting method of the consideration Seasonal Characteristics described in claim 1, it is characterised in that:Institute
The Seasonal Characteristics stated are stated using seasonal index number, and seasonal index number is defined as into the actual value in the season and putting down for place year in season
The ratio between average;Set N to take the time span of recent 3~5 years, realize the renewal with power consumption historical data, seasonal index number is also sent out
Raw roll updates, and will be divided into K sections season, note Sij is the season power consumption actual value in 1 year jth season, single annual seasons exponential representation
The season power consumption in 1 year jth season accounts for the ratio of 1 year Urban Annual Electrical Power Consumption amount, but annual seasons index is:
In formula:I=1,2 ..., N;J=1,2 ..., K;
Seasonal index number Fj is averaged for N single annual seasons index, i.e.,:
In formula:J=1,2 ..., K;
Average seasonal index numberFor the average value in K season, i.e.,:
In formula:J=1,2 ..., K;
The relative deviation of seasonal index number and average seasonal index number is:
2
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710251850.2A CN107038502A (en) | 2017-04-18 | 2017-04-18 | Consider the improvement wavelet packet electricity demand forecasting method of Seasonal Characteristics |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710251850.2A CN107038502A (en) | 2017-04-18 | 2017-04-18 | Consider the improvement wavelet packet electricity demand forecasting method of Seasonal Characteristics |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107038502A true CN107038502A (en) | 2017-08-11 |
Family
ID=59535513
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710251850.2A Pending CN107038502A (en) | 2017-04-18 | 2017-04-18 | Consider the improvement wavelet packet electricity demand forecasting method of Seasonal Characteristics |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107038502A (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111275247A (en) * | 2020-01-14 | 2020-06-12 | 西安理工大学 | Extreme learning machine monthly electric quantity prediction method based on multiple influence factors |
CN112365039A (en) * | 2020-11-03 | 2021-02-12 | 国网冀北电力有限公司计量中心 | Method and system for monthly power consumption prediction based on longitudinal and transverse correlation |
WO2021137759A1 (en) * | 2019-12-31 | 2021-07-08 | Envision Digital International Pte. Ltd. | Method and apparatus for predicting power consumption, device and readiable storage medium |
CN113610296A (en) * | 2021-08-05 | 2021-11-05 | 王莹 | Method for predicting peak value of electrical load in region and method for planning investment of power grid |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102663420A (en) * | 2012-03-22 | 2012-09-12 | 北京航空航天大学 | Hyperspectral image classification method based on wavelet packet transformation and grey prediction model |
CN104636822A (en) * | 2015-01-21 | 2015-05-20 | 广州市香港科大霍英东研究院 | Residential load prediction method of elman-based neural network |
CN105184679A (en) * | 2015-09-22 | 2015-12-23 | 国网山东省电力公司电力科学研究院 | Wind power plant power multi-time scale joint prediction method |
CN105184391A (en) * | 2015-08-19 | 2015-12-23 | 国网山东省电力公司电力科学研究院 | Method for predicting wind speed and power of wind farm based on wavelet decomposition and support vector machine |
CN106845714A (en) * | 2017-01-24 | 2017-06-13 | 东南大学 | A kind of monthly passenger flow method of ARIMA model prediction urban track traffics based on seasonal index number |
-
2017
- 2017-04-18 CN CN201710251850.2A patent/CN107038502A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102663420A (en) * | 2012-03-22 | 2012-09-12 | 北京航空航天大学 | Hyperspectral image classification method based on wavelet packet transformation and grey prediction model |
CN104636822A (en) * | 2015-01-21 | 2015-05-20 | 广州市香港科大霍英东研究院 | Residential load prediction method of elman-based neural network |
CN105184391A (en) * | 2015-08-19 | 2015-12-23 | 国网山东省电力公司电力科学研究院 | Method for predicting wind speed and power of wind farm based on wavelet decomposition and support vector machine |
CN105184679A (en) * | 2015-09-22 | 2015-12-23 | 国网山东省电力公司电力科学研究院 | Wind power plant power multi-time scale joint prediction method |
CN106845714A (en) * | 2017-01-24 | 2017-06-13 | 东南大学 | A kind of monthly passenger flow method of ARIMA model prediction urban track traffics based on seasonal index number |
Non-Patent Citations (1)
Title |
---|
覃泓皓等: "考虑季节特性的改进小波包-RBF用电量预测方法", 《万方数据库》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2021137759A1 (en) * | 2019-12-31 | 2021-07-08 | Envision Digital International Pte. Ltd. | Method and apparatus for predicting power consumption, device and readiable storage medium |
CN111275247A (en) * | 2020-01-14 | 2020-06-12 | 西安理工大学 | Extreme learning machine monthly electric quantity prediction method based on multiple influence factors |
CN111275247B (en) * | 2020-01-14 | 2024-02-23 | 西安理工大学 | Extreme learning machine month electric quantity prediction method based on multiple influence factors |
CN112365039A (en) * | 2020-11-03 | 2021-02-12 | 国网冀北电力有限公司计量中心 | Method and system for monthly power consumption prediction based on longitudinal and transverse correlation |
CN112365039B (en) * | 2020-11-03 | 2023-11-24 | 国网冀北电力有限公司计量中心 | Method and system for predicting monthly electricity consumption based on longitudinal and transverse correlation |
CN113610296A (en) * | 2021-08-05 | 2021-11-05 | 王莹 | Method for predicting peak value of electrical load in region and method for planning investment of power grid |
CN113610296B (en) * | 2021-08-05 | 2022-03-11 | 国网黑龙江省电力有限公司经济技术研究院 | Method for predicting peak value of electrical load in region and method for planning investment of power grid |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111860979B (en) | Short-term load prediction method based on TCN and IPSO-LSSVM combined model | |
Wei et al. | Daily natural gas consumption forecasting via the application of a novel hybrid model | |
CN112488395B (en) | Method and system for predicting line loss of power distribution network | |
CN107038502A (en) | Consider the improvement wavelet packet electricity demand forecasting method of Seasonal Characteristics | |
Tang et al. | Oil price shocks and their short-and long-term effects on the Chinese economy | |
CN110705743B (en) | New energy consumption electric quantity prediction method based on long-term and short-term memory neural network | |
Costantini et al. | A hierarchical procedure for the combination of forecasts | |
Geem | Transport energy demand modeling of South Korea using artificial neural network | |
CN109583621A (en) | A kind of PSO-LSSVM short-term load forecasting method based on improvement variation mode decomposition | |
CN101480143B (en) | Method for predicating single yield of crops in irrigated area | |
CN106874581A (en) | A kind of energy consumption of air conditioning system in buildings Forecasting Methodology based on BP neural network model | |
CN112288164B (en) | Wind power combined prediction method considering spatial correlation and correcting numerical weather forecast | |
CN112149879B (en) | New energy medium-and-long-term electric quantity prediction method considering macroscopic volatility classification | |
CN110210993B (en) | Urban short-term gas load prediction method based on cyclic neural network model | |
CN101825622A (en) | Water quality prediction method and device | |
CN107248013A (en) | A kind of monthly electricity demand forecasting method for considering diversified economy factor | |
CN105678404A (en) | Micro-grid load prediction system and method based on electricity purchased on-line and dynamic correlation factor | |
CN103514491A (en) | Power load forecasting method | |
CN111539844B (en) | Moon electricity quantity prediction method based on K-L information quantity and ARIMA error correction | |
CN104573857A (en) | Power grid load rate prediction method based on intelligent algorithm optimization and combination | |
CN105447646A (en) | Health index assessment method for power distribution system | |
CN105956716A (en) | Total social electricity consumption prediction method based on industry economy and electricity relationship | |
CN106875104A (en) | A kind of cold-strip steel quality overall evaluation method | |
CN104834975A (en) | Power network load factor prediction method based on intelligent algorithm optimization combination | |
CN110880044B (en) | Markov chain-based load prediction method |
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 |
Application publication date: 20170811 |
|
RJ01 | Rejection of invention patent application after publication |