CN109359778A - Short-term Forecast of Natural Gas Load method based on Optimization Experience mode decomposition - Google Patents
Short-term Forecast of Natural Gas Load method based on Optimization Experience mode decomposition Download PDFInfo
- Publication number
- CN109359778A CN109359778A CN201811346703.4A CN201811346703A CN109359778A CN 109359778 A CN109359778 A CN 109359778A CN 201811346703 A CN201811346703 A CN 201811346703A CN 109359778 A CN109359778 A CN 109359778A
- Authority
- CN
- China
- Prior art keywords
- natural gas
- short
- term
- mode decomposition
- gas load
- 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
- VNWKTOKETHGBQD-UHFFFAOYSA-N methane Chemical compound C VNWKTOKETHGBQD-UHFFFAOYSA-N 0.000 title claims abstract description 132
- 238000000034 method Methods 0.000 title claims abstract description 68
- 239000003345 natural gas Substances 0.000 title claims abstract description 66
- 238000000354 decomposition reaction Methods 0.000 title claims abstract description 65
- 238000005457 optimization Methods 0.000 title claims abstract description 36
- 239000002245 particle Substances 0.000 claims abstract description 20
- 238000012549 training Methods 0.000 claims description 17
- 238000009499 grossing Methods 0.000 claims description 13
- 238000004458 analytical method Methods 0.000 claims description 5
- 230000000694 effects Effects 0.000 claims description 4
- 238000011156 evaluation Methods 0.000 claims description 3
- 238000013507 mapping Methods 0.000 claims description 3
- 238000011160 research Methods 0.000 claims description 3
- 230000009977 dual effect Effects 0.000 claims description 2
- 238000010606 normalization Methods 0.000 claims description 2
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 claims 1
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 claims 1
- 235000006508 Nelumbo nucifera Nutrition 0.000 claims 1
- 240000002853 Nelumbo nucifera Species 0.000 claims 1
- 235000006510 Nelumbo pentapetala Nutrition 0.000 claims 1
- 239000007789 gas Substances 0.000 abstract description 10
- 230000006870 function Effects 0.000 abstract description 5
- 238000010586 diagram Methods 0.000 description 39
- 238000012360 testing method Methods 0.000 description 15
- 238000013459 approach Methods 0.000 description 2
- 239000002131 composite material Substances 0.000 description 2
- 230000002068 genetic effect Effects 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 238000003672 processing method Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 239000003814 drug Substances 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000013277 forecasting method Methods 0.000 description 1
- 238000010353 genetic engineering Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012706 support-vector machine Methods 0.000 description 1
- 230000004083 survival effect Effects 0.000 description 1
- 238000000714 time series forecasting Methods 0.000 description 1
- 239000002699 waste material Substances 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
- 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
- G06Q50/02—Agriculture; Fishing; Forestry; Mining
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Theoretical Computer Science (AREA)
- Marketing (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Tourism & Hospitality (AREA)
- Animal Husbandry (AREA)
- Life Sciences & Earth Sciences (AREA)
- Agronomy & Crop Science (AREA)
- Mining & Mineral Resources (AREA)
- General Health & Medical Sciences (AREA)
- Marine Sciences & Fisheries (AREA)
- Health & Medical Sciences (AREA)
- Primary Health Care (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)
Abstract
The invention belongs to Forecast of Natural Gas Load method and technology fields, in particular to the short-term Forecast of Natural Gas Load technology based on Optimization Experience mode decomposition.Data are acquired from city gas door station, first, it is smoothed using data of the Gauss weighted moving average filter to acquisition, optimizing application empirical mode decomposition algorithm, i.e. integrated empirical mode decomposition algorithm, to treated, data are decomposed, and several IMF (Intrinsic mode functions) and trend term for decompositing next are normalized;Using particle group optimizing least square method supporting vector machine, to treated, component is predicted, and prediction result is reconstructed, and obtains final prediction result, and prediction result is compared with the result of least square method supporting vector machine prediction is individually used.The present invention has higher precision of prediction to short-term Forecast of Natural Gas Load, is a kind of efficient natural gas Short Term load Forecasting Technique.
Description
Technical field
The invention belongs to Forecast of Natural Gas Load method and technology fields, in particular to a kind of to be based on Optimization Experience mode decomposition
Short-term Forecast of Natural Gas Load technology.
Background technique
As the acceleration of Chinese Industrialization and urbanization process promotes, demand of all trades and professions to natural gas is growing day by day,
Due to Gas Productivity wretched insufficiency and consumption figure rapid growth, " gas is waste ", specific manifestation are as follows: peak are caused to urban
Period, demand is excessive, and air pressure is relatively low in pipe network, causes many users that cannot just commonly use gas;Low-valley interval, demand compared with
Few, gas ductwork air pressure inside is excessively high, equipment safety is threatened, so that pipe network operation inefficiency.Therefore, short-term natural gas load
The research of prediction, for guaranteeing that gas distributing system gas consumption, the scheduling for optimizing pipe network and maintenance of equipment have extremely important meaning
Justice.
However, natural gas load except have the characteristics that with week, day mechanical periodicity in addition to, also by weather, season, festivals or holidays etc.
Factors influence, and cause natural gas load fluctuation very frequent, in nonlinearity, time variation, dispersibility and randomness etc.
Feature, Accurate Prediction difficulty are big.Traditional Forecast of Natural Gas Load method includes linear regression analysis, time series method and ash
Color Systems Theory, but these methods are the model based on linear data prediction mostly, therefore are not suitable for complicated natural gas load
Prediction.There is BP neural network Nonlinear Mapping, arbitrary accuracy to approach, have the advantages such as very strong generalization ability and self study, but
There is also Local Minimum, overfitting and hidden layer number of network node to choose the defects of lacking theoretical direction, weakens it and predicts energy
Power.
Since empirical mode decomposition has adaptive capacity of decomposition, it both can handle stationary signal or can handle non-flat
Therefore steady signal finds it in the multiple fields such as electric power, medicine, traffic, machinery application meaning all with higher so far
Justice.The principle of integrated empirical mode decomposition is then averaged, is mutually supported to reach by repeating to add white noise sequence
Disappear, so that empirical mode decomposition method is really known as binary filter group, to a certain extent, to empirical mode decomposition
It is improved.
Least square method supporting vector machine is to improve to obtain on the basis of standard support vector machines, can be by time-consuming
Quadratic programming problem is transformed into Solving Linear problem, significantly reduces computation complexity, accelerates arithmetic speed.It should
The key of prediction model is to choose suitable input vector and model parameter.And genetic algorithm genetic manipulation and the survival of the fittest
Natural selection carrys out guidance learning and determines the direction of search, and ability of searching optimum is stronger.By the genetic algorithm of optimization and minimum two
Multiply SVM prediction model combination, provides new approaches for short-term Forecast of Natural Gas Load.
As described above, existing short-term Forecast of Natural Gas Load method, directly predicts it frequently with prediction model,
Precision of prediction is not high, and in order to obtain the more information of short-term natural gas load time series, forefathers often adopt when studying
With wavelet analysis, but there are two aspects in wavelet analysis, meanwhile, there are end effects when decomposing for empirical mode decomposition
The problem of, therefore propose that a kind of effect is obvious, short-term Forecast of Natural Gas Load technical meaning with high accuracy is great.
Summary of the invention
It is an object of the invention to be provided aiming at the problems existing in the prior art based on the short of Optimization Experience mode decomposition
Phase natural gas load composite prediction technology, this method can be improved the precision of prediction of short-term natural gas load time series forecasting.
The technical scheme is that
Short-term Forecast of Natural Gas Load method based on Optimization Experience mode decomposition, includes the following steps:
(1) acquire certain city's natural gas station when Load Time Series initial data, it is pre-processed, use
Method is Gauss weighted moving average filter, filters out noise item, i.e., is had with short-term natural gas load time series itself
The unrelated information of some rules;Optimizing application empirical mode decomposition algorithm, i.e., integrated empirical mode decomposition algorithm is to short-term natural
Gas Load Time Series are decomposed, and the purpose of decomposition is to obtain the short-term more information of natural gas load time series, together
When, the IMF component and trend term that decomposite are normalized, building is short-term natural based on Optimization Experience mode decomposition
Gas load forecasting model;
(2) component come is decomposited to integrated empirical mode decomposition, using particle group optimizing least square method supporting vector machine
It is predicted respectively, using regularization parameter and nuclear parameter in particle group optimizing least square method supporting vector machine;
(3) using the prediction model based on particle group optimizing least square method supporting vector machine to the prediction knot of each component
Fruit is reconstructed, and obtains final prediction result, and by the prediction result and individually using least square method supporting vector machine prediction
Prediction result compares, to determine the precision of prediction and validity of the prediction model of building.
Specifically, the detailed process of the step (1) are as follows:
1. set short-term natural gas load time series as x (i), i=1,2 ..., N;
2. being smoothed using Gauss weighted moving average filter to short-term natural gas load time series, also
It is to filter gaussian noise, improves precision of prediction;
3. using Optimization Experience mode decomposition algorithm, i.e., integrated empirical mode decomposition algorithm, to processed short-term natural
Gas Load Time Series are decomposed, several IMF components and trend term are decomposited;
4. using normalization method to decomposite come IMF component and trend term be normalized.
Specifically, the specific decomposable process of empirical mode decomposition of the step (1) are as follows:
It is the opposite lesser sequence of root mean square 1. white noise sequence is added in original short-term natural gas load time series
Column, can indicate are as follows:
xm(t)=x (t)+nm(t)
In formula, x (t) is the short-term natural gas load time series of actual acquisition, nmIt (t) is the sequence of white noise, xm(t)
To be mixed with the short-term natural gas load time series for needing to decompose after white noise sequence;
2. the short-term natural gas load time series x of white noise will be mixed withm(t) it is decomposed using integrated empirical modal,
It can be analyzed to multiple IMF component cim, i=1,2 ..., k;
1. and 2. 3. the equal different white noise sequences of root mean square are added every time, step is repeated, n group available in this way is not
With IMF component m=1,2 ..., n;
4. IMF component mean values different to n group calculate, using each IMF mean value as the final of integrated empirical mode decomposition
Decomposition result;
The principle of integrated empirical mode decomposition is then to be averaged, by repeating to add white noise sequence to reach
It cancels out each other, so that empirical mode decomposition method is really known as binary filter group, to a certain extent, to Empirical Mode
State decomposition is improved.
5. it is as follows to normalize formula:
In formula: ymax=1;ymin=-1;xiFor each single item inside short-term natural gas load time series;xmaxIt is short-term
Maximum value inside natural gas load time series;xminFor the minimum value inside short-term natural gas load time series.
Specifically, the detailed process of the step (2) are as follows: given training sample set (xi,yi), i=1 ..., m, xi∈
Rn,yi∈ R selects a Nonlinear MappingIt is then empty herein the input vector space reflection of sample set to higher dimensional space F
Between upper construct optimum linearity decision function.
According to structural risk minimization (structure risk minimization, SRM) principle, the general of interval is introduced
It reads, and considers the complexity of error of fitting and function, obtain optimization problem are as follows:
Constraint condition are as follows:
In formula: eiFor error;e∈Rl×1For error vector;γ is regularization parameter, controls the punishment degree to error.
Introduce Lagrange multiplier αi, αi∈Rl×1The then Lagrange multinomial of its dual problem are as follows:
Above formula can be obtained according to KKT (Karush-Kuhn-Tucker) condition:
To which the optimization problem solved can be converted into the system of linear equations problem solved in following formula.
In formula: I=[1,1 ..., 1]T, α=(α1,α2…,αl)T, y=[y1,y2,…,yl]T, A=ZZT+γ-1I,
In formula: K (xi, x) be least square method supporting vector machine kernel function, research shows that effect is preferable in regression forecasting
Be Radial basis kernel function, it is as follows:
In formula: σ is core width, reflects the radius that closing of the frontier includes.
Specifically, the detailed process of the step (3) are as follows: to optimized empirical mode decomposition algorithm, i.e., integrated experience
Each IMF and trend term of mode decomposition algorithm and particle group optimizing least square method supporting vector machine carry out renormalization, into
Row reconstruct obtains final prediction result, and by prediction result and individually using the result of least square method supporting vector machine prediction
It compares, to determine the precision of prediction and validity of the prediction model of building, error is carried out using following three evaluation criterion
Evaluation analysis, respectively RSME (root-mean-square error), MAE (mean absolute error), MAPE (average absolute percentage error), specifically
Formula are as follows:
In formula: N is the total quantity of error comparison;Lactual is the natural gas load of actual acquisition;Lforecast is pre-
The natural gas load of survey.
The beneficial effects of the present invention are: directly it is predicted when being directed to existing short-term Forecast of Natural Gas Load, and not
Noise Processing for removing is carried out, while not excavating the problem that short-term natural gas load time series itself contains rule, the present invention
The short-term natural gas load composite prediction technology based on Optimization Experience mode decomposition is provided, this method can be improved short-term natural gas
The precision of prediction of load time series prediction.
Detailed description of the invention
Fig. 1 is processing method technology path schematic diagram of the present invention to data.
Fig. 2 is the schematic diagram that the initial data of acquisition is formed;
Fig. 3 is the schematic diagram that data are formed after carrying out Gaussian smoothing to initial data;
Fig. 4 is the schematic diagram individually formed using least square method supporting vector machine training set;
Fig. 5 is the schematic diagram individually formed using least square method supporting vector machine test set;
Fig. 6 is the schematic diagram that 1 component of imf of integrated empirical mode decomposition is formed;
Fig. 7 is the schematic diagram that 2 component of imf of integrated empirical mode decomposition is formed;
Fig. 8 is the schematic diagram that 3 component of imf of integrated empirical mode decomposition is formed;
Fig. 9 is the schematic diagram that 4 component of imf of integrated empirical mode decomposition is formed;
Figure 10 is the schematic diagram that 5 component of imf of integrated empirical mode decomposition is formed;
Figure 11 is the schematic diagram that 6 component of imf of integrated empirical mode decomposition is formed;
Figure 12 is the schematic diagram that 7 component of imf of integrated empirical mode decomposition is formed;
Figure 13 is the schematic diagram that 8 component of imf of integrated empirical mode decomposition is formed;
Figure 14 is the schematic diagram that the Residue component of integrated empirical mode decomposition is formed;
Figure 15 is to carry out the schematic diagram that smooth rear result is formed to 1 component of imf using Gaussian smoothing;
Figure 16 is to carry out the schematic diagram that smooth rear result is formed to 2 component of imf using Gaussian smoothing;
Figure 17 is to carry out the schematic diagram that smooth rear result is formed to 3 component of imf using Gaussian smoothing;
Figure 18 is to carry out the schematic diagram that smooth rear result is formed to 4 component of imf using Gaussian smoothing;
Figure 19 is to carry out the schematic diagram that smooth rear result is formed to 5 component of imf using Gaussian smoothing;
Figure 20 is to carry out the schematic diagram that smooth rear result is formed to 6 component of imf using Gaussian smoothing;
Figure 21 is to carry out the schematic diagram that smooth rear result is formed to 7 component of imf using Gaussian smoothing;
Figure 22 is to carry out the schematic diagram that smooth rear result is formed to 8 component of imf using Gaussian smoothing;
Figure 23 is to carry out the schematic diagram that smooth rear result is formed to Residue component using Gaussian smoothing;
Figure 24 is 1 component of particle group optimizing least square method supporting vector machine imf and training test result schematic diagram;
Figure 25 is 2 component of particle group optimizing least square method supporting vector machine imf and training test result schematic diagram;
Figure 26 is 3 component of particle group optimizing least square method supporting vector machine imf and training test result schematic diagram;
Figure 27 is 4 component of particle group optimizing least square method supporting vector machine imf and training test result schematic diagram;
Figure 28 is 5 component of particle group optimizing least square method supporting vector machine imf and training test result schematic diagram;
Figure 29 is 6 component of particle group optimizing least square method supporting vector machine imf and training test result schematic diagram;
Figure 30 is 7 component of particle group optimizing least square method supporting vector machine imf and training test result schematic diagram;
Figure 31 is 8 component of particle group optimizing least square method supporting vector machine imf and training test result schematic diagram;
Figure 32 is particle group optimizing least square method supporting vector machine Residue component and training test result schematic diagram;
Figure 33 is reconstructed prediction result.
Specific embodiment
Below with reference to specific example data and attached drawing to the short-term day provided by the invention based on Optimization Experience mode decomposition
Right gas load combination forecasting method makes detailed explanation.
Fig. 1 is processing method technology path of the present invention to data, acquires the daily load number at the natural valve station in certain city first
According to the sample size of acquisition is 762 groups, and the time of acquisition is on October 30,1 day to 2017 October in 2016, the daily morning
It is acquired when 10 primary.In order to improve precision of prediction, using Optimization Experience mode decomposition method, i.e., integrated empirical mode decomposition is to adopting
The natural gas daily load time series of collection is decomposed, several imf components and trend term are decomposited, and is weighted and is moved using Gauss
Method is smoothed each component, constructs the short-term Forecast of Natural Gas Load mould based on Optimization Experience mode decomposition
Type;Secondly, being predicted most using particle group optimizing least square method supporting vector machine each component Jing Guo smoothing processing
After be reconstructed, and by algorithm of the invention and individually compared and analyzed using the algorithm of least square method supporting vector machine.
Fig. 2 is the schematic diagram that the initial data of acquisition is formed, and Fig. 3 is that moving method is weighted using Gauss to the natural of acquisition
Gas daily load time series data carries out smoothed out schematic diagram.
Fig. 4 gives the individually natural gas daily load time series training set using least square method supporting vector machine to acquisition
The schematic diagram of formation, Fig. 5 are the schematic diagram individually formed using least square method supporting vector machine test set, initial parameter c in Fig. 4
It is set as 0.1 and 1 with g, the error of RMSE, MAE and MAPE value of training set is respectively 65.5436,0.0163 and 2.4142,
Fig. 5 initial parameter c and g are set as 0.1 and 1, the error of RMSE, MAE and MAPE value of test set is respectively 220.6973,
0.0275 and 4.3015, it can be seen that, three errors of test set are larger compared with for training set.
Fig. 6-Figure 14 gives using Optimization Experience mode decomposition, i.e., natural gas day of the integrated empirical mode decomposition to acquisition
The schematic diagram that the component that Load Time Series are decomposed is formed, amplitude coefficient are set as 0.1.
Figure 15-Figure 23 gives to be smoothed respectively using component of the Gaussian smoothing to optimized empirical mode decomposition
Schematic diagram afterwards.
Figure 24-Figure 32 is that different component training sets and test set compare, major parameter setting are as follows: the length of sliding window is
10, greatest iteration number 20, population number 50, Studying factors c1 is 1, and Studying factors c2 is 1, weight limit 0.8, parameter c
Search range be (0,10), the search range of parameter g is (0,1000), using particle group optimizing least square method supporting vector machine
In parameter, after optimization imf1 component parameter c and g optimization after be 0.6496 and 493.2469, the parameter c and g of imf2 component
It is 2.2000 and 991.0764 after optimization, is 3.0801 and 991.8896 after parameter c and the g optimization of imf3 component, imf4 component
It is 3.4444 and 995.9978 after the optimization of parameter c and g, is 2.4765 Hes after parameter c and the g optimization of imf5 component
It is 3.8963 and 985.7387 after parameter c and the g optimization of 985.4218, imf6 components, after parameter c and the g optimization of imf7 component
It is 2.6331 and 983.7516 after parameter c and the g optimization of imf8 component for 3.3704 and 972.2381, the ginseng of residue component
It is 9.9181 and 973.9277 after number c and g optimization.The three error assessment indexs and test set of each component of training set are each divided
Three error assessment indexs of amount are as shown in table 1 below:
Table 1
Figure 33 is prediction result display schematic diagram after reconstruct, it can be seen from the figure that optimized empirical mode decomposition and grain
Subgroup optimizes the prediction result of least square supporting vector and actual value is coincide preferably.
In conclusion being individually respectively using the error of RMSE, MAE and MAPE value of least square method supporting vector machine
220.6973,0.0275 and 4.3015, the error using RMSE, MAE and MAPE value of Optimization Experience mode decomposition is respectively
11.7345,0.0093 and 0.5846, it follows that innovatory algorithm is more individually predicted using least square method supporting vector machine,
For three error assessment standards, 208.9628,0.0182 and 3.7169 have been respectively increased, precision of prediction is obviously improved,
Demonstrate the validity of inventive algorithm.
Finally it should be noted that: the above embodiments are merely illustrative of the technical scheme of the present invention and are not intended to be limiting thereof;To the greatest extent
The present invention is described in detail with reference to preferred embodiments for pipe, it should be understood by those ordinary skilled in the art that: still
It can modify to a specific embodiment of the invention or some technical features can be equivalently replaced;Without departing from this hair
The spirit of bright technical solution should all cover within the scope of the technical scheme claimed by the invention.
Claims (5)
1. the short-term Forecast of Natural Gas Load method based on Optimization Experience mode decomposition, which comprises the steps of:
(1) the when Load Time Series data for acquiring certain city's natural gas station, when to the short-term natural gas of actual acquisition when load
Between sequence carry out Gaussian smoothing, using Optimization Experience mode decomposition algorithm, i.e., integrated empirical mode decomposition algorithm, to short-term
Natural gas load time series is decomposed, and decomposites IMF and trend term, and carry out normalizing to the IMF and trend term decomposited
Change, constructs the short-term Forecast of Natural Gas Load model based on Optimization Experience mode decomposition;
(2) component come is decomposited to integrated empirical mode decomposition, is distinguished using particle group optimizing least square method supporting vector machine
It is predicted, using regularization parameter and nuclear parameter in particle group optimizing least square method supporting vector machine;
(3) using the prediction model based on particle group optimizing least square method supporting vector machine to the prediction result of each component into
Row reconstruct obtains final prediction result, and by the prediction result and individually using the prediction of least square method supporting vector machine prediction
As a result it compares, to determine the precision of prediction and validity of the prediction model of building.
2. the short-term Forecast of Natural Gas Load method based on Optimization Experience mode decomposition, feature exist according to claim 1
In the detailed process of the step (1) are as follows:
1. set short-term natural gas load time series as x (i), i=1,2 ..., N;
2. being smoothed using Gauss weighted moving average filter to short-term natural gas load time series, that is, filter
Fall gaussian noise, improves precision of prediction;
3. using Optimization Experience mode decomposition algorithm, i.e., integrated empirical mode decomposition algorithm is negative to processed short-term natural gas
Lotus time series is decomposed, several IMF components and trend term are decomposited;
4. using normalization method to decomposite come IMF component and trend term be normalized.
3. the short-term Forecast of Natural Gas Load method based on Optimization Experience mode decomposition, feature exist according to claim 1
In the detailed process of the step (2) are as follows:
Given training sample set (xi,yi), i=1 ..., m, xi∈Rn,yi∈ R selects a Nonlinear MappingSample set
Then input vector space reflection spatially constructs optimum linearity decision function to higher dimensional space F herein;
According to structural risk minimization principle, the concept at interval is introduced, and considers the complexity of error of fitting and function, is obtained excellent
Change problem are as follows:
Constraint condition are as follows:
In formula: eiFor error;e∈Rl×1For error vector;γ is regularization parameter, controls the punishment degree to error;
Introduce Lagrange multiplier αi, αi∈Rl×1The then Lagrange multinomial of its dual problem are as follows:
Above formula can be obtained according to KKT condition:
To which the optimization problem solved can be converted into the system of linear equations problem solved in following formula;
In formula: I=[1,1 ..., 1]T, α=(α1,α2…,αl)T, y=[y1,y2,…,yl]T, A=ZZT+γ-1I,I=1,2 ..., l
In formula: K (xi, x) be least square method supporting vector machine kernel function, research shows that effect is preferably in regression forecasting
Radial basis kernel function, as follows:
In formula: σ is core width, reflects the radius that closing of the frontier includes.
4. the short-term Forecast of Natural Gas Load method based on Optimization Experience mode decomposition, feature exist according to claim 1
In the detailed process of the step (3) are as follows:
To optimized empirical mode decomposition algorithm, i.e., integrated empirical mode decomposition algorithm and particle group optimizing least square support to
Each IMF and trend term of amount machine carry out renormalization, are reconstructed, and obtain final prediction result, and by prediction result
It is compared with the result of least square method supporting vector machine prediction is individually used, to determine the precision of prediction of the prediction model of building
And validity, error assessment analysis, respectively RSME, MAE, MAPE, specific formula are carried out using following three evaluation criterion are as follows:
In formula: N is the total quantity of error comparison;Lactual is the natural gas load of actual acquisition;Lforecast is prediction
Natural gas load.
5. the short-term Forecast of Natural Gas Load method based on Optimization Experience mode decomposition, feature exist according to claim 2
In the detailed process of the step (1) are as follows:
It is the opposite lesser sequence of root mean square 1. white noise sequence is added in original short-term natural gas load time series,
It can indicate are as follows:
xm(t)=x (t)+nm(t)
In formula, x (t) is the short-term natural gas load time series of actual acquisition, nmIt (t) is the sequence of white noise, xmIt (t) is mixed
There is the short-term natural gas load time series for needing to decompose after white noise sequence;
2. the short-term natural gas load time series x of white noise will be mixed withm(t) it is decomposed, be can decompose using integrated empirical modal
For multiple IMF component cim, i=1,2 ..., k;
1. and 2. 3. the equal different white noise sequences of root mean square are added every time, step is repeated, n group difference available in this way
IMF component m=1,2 ..., n;
4. IMF component mean values different to n group calculate, using each IMF mean value as the final decomposition of integrated empirical mode decomposition
As a result;
The principle of integrated empirical mode decomposition is then to be averaged, by repeating to add white noise sequence to reach mutual
It offsets, empirical mode decomposition method is allow really to be known as binary filter group, to a certain extent, to empirical modal point
Solution is improved.
5. it is as follows to normalize formula:
In formula: ymax=1;ymin=-1;xiFor each single item inside short-term natural gas load time series;xmaxFor short-term natural gas
Maximum value inside Load Time Series;xminFor the minimum value inside short-term natural gas load time series.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811346703.4A CN109359778A (en) | 2018-11-13 | 2018-11-13 | Short-term Forecast of Natural Gas Load method based on Optimization Experience mode decomposition |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811346703.4A CN109359778A (en) | 2018-11-13 | 2018-11-13 | Short-term Forecast of Natural Gas Load method based on Optimization Experience mode decomposition |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109359778A true CN109359778A (en) | 2019-02-19 |
Family
ID=65345128
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811346703.4A Pending CN109359778A (en) | 2018-11-13 | 2018-11-13 | Short-term Forecast of Natural Gas Load method based on Optimization Experience mode decomposition |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109359778A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109948861A (en) * | 2019-03-26 | 2019-06-28 | 西南交通大学 | A kind of urban rail based on mode decomposition and deep learning passenger flow forecasting in short-term |
CN111166322A (en) * | 2020-01-17 | 2020-05-19 | 哈尔滨工业大学 | Electrocardiosignal characteristic wave extraction method |
CN111429180A (en) * | 2020-03-23 | 2020-07-17 | 上海中旖能源科技有限公司 | Liquefied natural gas demand prediction method and device |
CN111652424A (en) * | 2020-05-28 | 2020-09-11 | 国网甘肃省电力公司经济技术研究院 | Hydrogen load prediction method |
CN113222234A (en) * | 2021-04-30 | 2021-08-06 | 中国石油天然气股份有限公司 | Gas demand prediction method and system based on integrated modal decomposition |
CN113496315A (en) * | 2021-09-08 | 2021-10-12 | 广东电网有限责任公司湛江供电局 | Load interval prediction method and system based on quantile gradient lifting decision tree |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101846753A (en) * | 2010-04-29 | 2010-09-29 | 南京信息工程大学 | Climate time sequence forecasting method based on empirical mode decomposition and support vector machine |
RU2563323C1 (en) * | 2014-04-02 | 2015-09-20 | Открытое акционерное общество "Нефтяная компания "Роснефть" | Method of reconstruction of fine structure of geological object and forecast of its fluid saturation |
CN105373856A (en) * | 2015-11-23 | 2016-03-02 | 江苏省电力公司南通供电公司 | Wind electricity power short-term combined prediction method considering run detection method reconstruction |
CN107766968A (en) * | 2017-09-26 | 2018-03-06 | 河海大学 | Short-term wind speed forecasting method based on CAPSO RVM built-up patterns |
-
2018
- 2018-11-13 CN CN201811346703.4A patent/CN109359778A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101846753A (en) * | 2010-04-29 | 2010-09-29 | 南京信息工程大学 | Climate time sequence forecasting method based on empirical mode decomposition and support vector machine |
RU2563323C1 (en) * | 2014-04-02 | 2015-09-20 | Открытое акционерное общество "Нефтяная компания "Роснефть" | Method of reconstruction of fine structure of geological object and forecast of its fluid saturation |
CN105373856A (en) * | 2015-11-23 | 2016-03-02 | 江苏省电力公司南通供电公司 | Wind electricity power short-term combined prediction method considering run detection method reconstruction |
CN107766968A (en) * | 2017-09-26 | 2018-03-06 | 河海大学 | Short-term wind speed forecasting method based on CAPSO RVM built-up patterns |
Non-Patent Citations (2)
Title |
---|
李霄 等: "基于改进最小二乘支持向量机和预测误差校正的短期风电负荷预测", 《电力系统保护与控制》 * |
王贺 等: "基于聚类经验模态分解和最小二乘支持向量机的短期风速组合预测", 《电工技术学报》 * |
Cited By (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109948861A (en) * | 2019-03-26 | 2019-06-28 | 西南交通大学 | A kind of urban rail based on mode decomposition and deep learning passenger flow forecasting in short-term |
CN111166322A (en) * | 2020-01-17 | 2020-05-19 | 哈尔滨工业大学 | Electrocardiosignal characteristic wave extraction method |
CN111429180A (en) * | 2020-03-23 | 2020-07-17 | 上海中旖能源科技有限公司 | Liquefied natural gas demand prediction method and device |
CN111652424A (en) * | 2020-05-28 | 2020-09-11 | 国网甘肃省电力公司经济技术研究院 | Hydrogen load prediction method |
CN111652424B (en) * | 2020-05-28 | 2022-06-10 | 国网甘肃省电力公司经济技术研究院 | Hydrogen load prediction method |
CN113222234A (en) * | 2021-04-30 | 2021-08-06 | 中国石油天然气股份有限公司 | Gas demand prediction method and system based on integrated modal decomposition |
CN113222234B (en) * | 2021-04-30 | 2024-04-09 | 中国石油天然气股份有限公司 | Gas demand prediction method and system based on integrated modal decomposition |
CN113496315A (en) * | 2021-09-08 | 2021-10-12 | 广东电网有限责任公司湛江供电局 | Load interval prediction method and system based on quantile gradient lifting decision tree |
WO2023035564A1 (en) * | 2021-09-08 | 2023-03-16 | 广东电网有限责任公司湛江供电局 | Load interval prediction method and system based on quantile gradient boosting decision tree |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109359778A (en) | Short-term Forecast of Natural Gas Load method based on Optimization Experience mode decomposition | |
Liang et al. | A novel wind speed prediction strategy based on Bi-LSTM, MOOFADA and transfer learning for centralized control centers | |
Sun et al. | Prediction and analysis of the three major industries and residential consumption CO2 emissions based on least squares support vector machine in China | |
Xu et al. | House price forecasting with neural networks | |
CN101414366B (en) | Method for forecasting electric power system short-term load based on method for improving uttermost learning machine | |
Song et al. | Hourly heat load prediction model based on temporal convolutional neural network | |
Al-Qaness et al. | Oil consumption forecasting using optimized adaptive neuro-fuzzy inference system based on sine cosine algorithm | |
CN109345027B (en) | Micro-grid short-term load prediction method based on independent component analysis and support vector machine | |
CN109871977A (en) | Load forecasting method based on wavelet transform and the minimum support vector machines of optimization | |
Prakash et al. | Modified immune algorithm for job selection and operation allocation problem in flexible manufacturing systems | |
CN113159361A (en) | Short-term load prediction method and system based on VDM and Stacking model fusion | |
CN103295075A (en) | Ultra-short-term power load forecasting and early warning method | |
CN102663513A (en) | Combination forecast modeling method of wind farm power by using gray correlation analysis | |
CN106971238A (en) | The Short-Term Load Forecasting Method of Elman neutral nets is obscured based on T S | |
CN110689162B (en) | Bus load prediction method, device and system based on user side classification | |
CN115689008A (en) | CNN-BilSTM short-term photovoltaic power prediction method and system based on ensemble empirical mode decomposition | |
CN115329682B (en) | LSTM-SVR subway station temperature prediction method based on multi-cycle characteristics | |
CN113554466A (en) | Short-term power consumption prediction model construction method, prediction method and device | |
CN115034430A (en) | Carbon emission prediction method, device, terminal and storage medium | |
CN110796303B (en) | Short-term power load prediction method based on EWT and ODBSCAN | |
Sheng et al. | Data Mining in census data with CART | |
CN112215410A (en) | Power load prediction method based on improved deep learning | |
CN115936184A (en) | Load prediction matching method suitable for multi-user types | |
Sala et al. | Smart multi-model approach based on adaptive Neuro-Fuzzy Inference Systems and Genetic Algorithms | |
CN113947432A (en) | Industrial product price short-term prediction method based on periodic decomposition |
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: 20190219 |
|
RJ01 | Rejection of invention patent application after publication |