CN107292462A - A kind of short-term load forecasting method, apparatus and system - Google Patents
A kind of short-term load forecasting method, apparatus and system Download PDFInfo
- Publication number
- CN107292462A CN107292462A CN201710741552.1A CN201710741552A CN107292462A CN 107292462 A CN107292462 A CN 107292462A CN 201710741552 A CN201710741552 A CN 201710741552A CN 107292462 A CN107292462 A CN 107292462A
- Authority
- CN
- China
- Prior art keywords
- population
- data
- particle
- learning machine
- extreme learning
- 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
- 238000013277 forecasting method Methods 0.000 title claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 37
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 20
- 239000002245 particle Substances 0.000 claims description 76
- 238000000034 method Methods 0.000 claims description 31
- 238000004364 calculation method Methods 0.000 claims description 15
- 230000008569 process Effects 0.000 claims description 15
- 238000012545 processing Methods 0.000 claims description 7
- 238000005457 optimization Methods 0.000 claims description 6
- 230000006978 adaptation Effects 0.000 claims description 4
- 238000010030 laminating Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 description 6
- 230000006870 function Effects 0.000 description 5
- 230000002123 temporal effect Effects 0.000 description 5
- 241000208340 Araliaceae Species 0.000 description 3
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 3
- 235000003140 Panax quinquefolius Nutrition 0.000 description 3
- 230000008901 benefit Effects 0.000 description 3
- 235000008434 ginseng Nutrition 0.000 description 3
- 101100257981 Arabidopsis thaliana S-ACP-DES3 gene Proteins 0.000 description 2
- 101150057677 DAD3 gene Proteins 0.000 description 2
- 102100034004 Gamma-adducin Human genes 0.000 description 2
- 101000799011 Homo sapiens Gamma-adducin Proteins 0.000 description 2
- 101001080624 Homo sapiens Proline/serine-rich coiled-coil protein 1 Proteins 0.000 description 2
- 101000652684 Homo sapiens Transcriptional adapter 3 Proteins 0.000 description 2
- 102100027427 Proline/serine-rich coiled-coil protein 1 Human genes 0.000 description 2
- 101100433696 Saccharomyces cerevisiae (strain ATCC 204508 / S288c) AAD3 gene Proteins 0.000 description 2
- 102100030836 Transcriptional adapter 3 Human genes 0.000 description 2
- 230000002153 concerted effect Effects 0.000 description 2
- 208000036552 dowling-degos disease 3 Diseases 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 208000002153 familial abdominal 3 aortic aneurysm Diseases 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 240000002853 Nelumbo nucifera Species 0.000 description 1
- 235000006508 Nelumbo nucifera Nutrition 0.000 description 1
- 235000006510 Nelumbo pentapetala Nutrition 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000007796 conventional method Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000012417 linear regression Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 241000894007 species Species 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
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06315—Needs-based resource requirements planning or analysis
-
- 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)
- Human Resources & Organizations (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Quality & Reliability (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Operations Research (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Water Supply & Treatment (AREA)
- Educational Administration (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The embodiment of the invention discloses a kind of short-term load forecasting method, apparatus and system, including obtain historical load data after pretreatment and weather data;Historical load data is decomposed into the component data of multiple different frequency ranges using wavelet packet decomposition;Carry out single branch reconstruct respectively to each component data by wavelet package reconstruction algorithm and obtain each the first subsequence;Weather data is added into each first subsequence respectively, each the second subsequence is obtained;Each second subsequence is inputted into the extreme learning machine Optimized model pre-established and is predicted, obtains predicting the outcome with the one-to-one each height of each the second subsequence;Each height is predicted the outcome and is overlapped, short-term load forecasting result is obtained;Wherein, extreme learning machine Optimized model is optimized to extreme learning machine and obtained according to training sample data and population gravitation search hybrid algorithm.Global convergence precision is improved during use of the embodiment of the present invention, makes to predict the outcome more accurate.
Description
Technical field
The present embodiments relate to technical field of power systems, more particularly to a kind of short-term load forecasting method, device
And system.
Background technology
In power system, accurately short-term load forecasting is of crucial importance in Transaction algorithm and operational plan is arranged.With
Continuing to develop for social economy, load is increasingly complicated, accurate load prediction can ensure the stability and peace of power system
Quan Xing, improves the economic benefit and social benefit of power network.
At present, in the research of short-term load forecasting, conventional method uses such as Fourier expansion method, time series method, many
The methods such as first linear regression method and neural network prediction method are predicted to short term, but due to electric load and day gas bar
The many factors such as part, vacation have close relationship, and load curve often changes in randomness, and traditional Forecasting Methodology is difficult to pre-
The change of load is surveyed, global convergence precision is relatively low, and making to predict the outcome is difficult to the demand for meeting present load prediction precision.
Therefore, how to provide a kind of short-term load forecasting method for solving above-mentioned technical problem, apparatus and system turns into this
The current technical issues that need to address of art personnel.
The content of the invention
The purpose of the embodiment of the present invention is to provide a kind of short-term load forecasting method, apparatus and system, is dropped during use
Influence of the fluctuation of low load to load prediction accuracy, strengthens model local search ability, improves global convergence
Precision, makes to predict the outcome more accurate.
In order to solve the above technical problems, the embodiments of the invention provide a kind of short-term load forecasting method, including:
S11:Obtain historical load data after pretreatment and weather data;
S12:The historical load data is decomposed into the component data of multiple different frequency ranges using wavelet packet decomposition;
S13:Component data each described is carried out respectively by wavelet package reconstruction algorithm the reconstruct of single branch obtain each first
Subsequence;
S14:Respectively to weather data is added in first subsequence each described, each the second subsequence is obtained;
S15:Each described second subsequence is inputted into the extreme learning machine Optimized model pre-established and is predicted,
Obtain predicting the outcome with the one-to-one each height of the second subsequence each described;
S16:Each described son is predicted the outcome and is overlapped, short-term load forecasting result is obtained;
Wherein, the process of setting up of the extreme learning machine Optimized model is:
S21:Training sample data are obtained, and the training sample data are inputted into extreme learning machine;
S22:Hybrid algorithm is searched for using population gravitation optimizing processing is carried out to the parameter of extreme learning machine, trained
The extreme learning machine Optimized model afterwards;The parameter includes input weights and hidden layer biasing.
Optionally, the use training sample data and population gravitation search for parameter of the hybrid algorithm to extreme learning machine
Carry out optimizing processing process be:
S221:According to the stochastic parameter initialization population of extreme learning machine, the parameter includes input weights and hidden layer
Biasing;
S222:Each in the initial population is calculated using training sample data and fitness value calculation relational expression initial
The fitness value of particle, wherein:
The fitness value calculation relational expression isWherein, fitiFor i-th of primary
Fitness value,For predicted value, P is desired output, and N is the sum of output predicted value in the training sample data, i=
1,2 ..., M, M are primary sum;
S223:The position and speed of each primary in the initial population are updated, obtain current
Population, and calculate the fitness value of each current particle in the current population;
S224:If the fitness value of i-th of current particle in current population is less than at the beginning of i-th in the initial population
The fitness value of beginning particle, then regard i-th of current particle as i-th of particle in new population;If i-th in current population
The fitness value of current particle is more than the fitness value of i-th of primary in the initial population, then by i-th of initial grain
Son is as i-th of particle in new population, to obtain new population;
S225:Judge whether current iteration number of times reaches maximum iteration, if it is, being found from the new population
The optimal particle of fitness value is exported as global optimum's particle, and obtains optimal ginseng according to global optimum's particle
Number;Otherwise, next iteration is continued, until reaching maximum iteration.
Optionally, the historical load data includes the load data for prediction 1 day a few days ago;The weather data includes pre-
Survey the weather data of the previous day day and predict the data of weather forecast of day.
Optionally, the temporal resolution of each load data is 1 hour.
Optionally, the weather data includes the daily highest temperature, the lowest temperature, rainfall and humidity.
The embodiment of the present invention provides a kind of short-term load forecasting device accordingly, including:
Acquisition module, for obtaining historical load data after pretreatment and weather data;
Decomposing module, point for the historical load data to be decomposed into multiple different frequency ranges using wavelet packet decomposition
Measure data;
Reconstructed module, is obtained for carrying out single branch reconstruct respectively to component data each described by wavelet package reconstruction algorithm
Each first subsequence;
Add module, for weather data is added in first subsequence each described, obtaining each the second sub- sequence respectively
Row;
Prediction module, for each described second subsequence to be inputted into the extreme learning machine Optimized model pre-established
It is predicted, obtains predicting the outcome with the one-to-one each height of the second subsequence each described;
Laminating module, is overlapped for each described son to be predicted the outcome, obtains short-term load forecasting result;
Wherein, the extreme learning machine Optimized model includes:
Acquisition module, is inputted into extreme learning machine for obtaining training sample data, and by the training sample data;
Optimization module, is carried out at optimizing for searching for hybrid algorithm using population gravitation to the parameter of extreme learning machine
Reason, the extreme learning machine Optimized model after being trained;The parameter includes input weights and hidden layer biasing.
Optionally, the optimization module includes:
Initialization unit, population is initialized for the stochastic parameter according to extreme learning machine, and the parameter includes input and weighed
Value and hidden layer biasing;
First computing unit, for calculating the initial kind using training sample data and fitness value calculation relational expression
The fitness value of each primary in group, wherein:
The fitness value calculation relational expression isWherein, fitiFor i-th of primary
Fitness value,For predicted value, P is desired output, and N is the sum of output predicted value in the training sample data, i=
1,2 ..., M, M are primary sum;
Updating block, is updated for the position to each primary in the initial population and speed,
Obtain current population;
Second computing unit, the fitness value for calculating each current particle in the current population;
Optimizing unit, the fitness value for i-th of current particle in current population is less than in the initial population
I-th of primary fitness value, then regard i-th of current particle as i-th of particle in new population;In current population
In the fitness value of i-th of current particle be more than the fitness value of i-th of primary in the initial population, then by the
I primary is as i-th of particle in new population, to obtain new population;
Judging unit, for judging whether current iteration number of times reaches maximum iteration, if it is, from the novel species
The optimal particle of fitness value is found in group to be exported as global optimum's particle, and is obtained according to global optimum's particle
Optimized parameter;Otherwise, next iteration is continued, until reaching maximum iteration.
The embodiment of the present invention additionally provides a kind of Short Term Load Forecasting System, including short-term load forecasting as described above
Device.
The embodiments of the invention provide a kind of short-term load forecasting method, apparatus and system, including obtain after pretreatment
Historical load data and weather data;Historical load data is decomposed into point of multiple different frequency ranges using wavelet packet decomposition
Measure data;Carry out single branch reconstruct respectively to each component data by wavelet package reconstruction algorithm and obtain each the first subsequence;Point
Weather data is not added into each first subsequence, each the second subsequence is obtained;Each second subsequence is inputted to pre-
It is predicted in the extreme learning machine Optimized model first set up, obtains predicting with the one-to-one each height of each the second subsequence
As a result;Each height is predicted the outcome and is overlapped, short-term load forecasting result is obtained;Wherein, extreme learning machine Optimized model
Process is set up to obtain training sample data, and training sample data are inputted into extreme learning machine;Using population gravitation
Search for hybrid algorithm and optimizing processing, the extreme learning machine Optimized model after being trained are carried out to the parameter of extreme learning machine;Ginseng
Number includes input weights and hidden layer biasing.
The embodiment of the present invention carries out resolution process using wavelet packet decomposition to historical load data, obtains multiple data point
Amount, reduces influence of the fluctuation of load to load prediction accuracy, and the extreme learning machine mould in the embodiment of the present invention
Type is to search for after hybrid algorithm carries out optimizing to the parameter of extreme learning machine to set up by population gravitation, using of the invention real
The extreme learning machine Optimized model for applying example offer is predicted to each subsequence, improves the estimated performance of extreme learning machine,
Model local search ability is strengthened, global convergence precision is improved, makes to predict the outcome more accurate.
Brief description of the drawings
Technical scheme in order to illustrate the embodiments of the present invention more clearly, below will be to institute in prior art and embodiment
The accompanying drawing needed to use is briefly described, it should be apparent that, drawings in the following description are only some implementations of the present invention
Example, for those of ordinary skill in the art, on the premise of not paying creative work, can also be obtained according to these accompanying drawings
Obtain other accompanying drawings.
Fig. 1 is a kind of schematic flow sheet of short-term load forecasting method provided in an embodiment of the present invention;
Fig. 2 is a kind of schematic flow sheet for setting up process of extreme learning machine Optimized model provided in an embodiment of the present invention;
Fig. 3 is a kind of structural representation of short-term load forecasting device provided in an embodiment of the present invention;
Fig. 4 is a kind of structural representation of extreme learning machine Optimized model provided in an embodiment of the present invention.
Embodiment
The embodiments of the invention provide a kind of short-term load forecasting method, apparatus and system, reduced during use negative
Influence of the fluctuation of lotus to load prediction accuracy, strengthens model local search ability, improves global convergence precision, makes
Predict the outcome more accurate.
To make the purpose, technical scheme and advantage of the embodiment of the present invention clearer, below in conjunction with the embodiment of the present invention
In accompanying drawing, the technical scheme in the embodiment of the present invention is clearly and completely described, it is clear that described embodiment is
A part of embodiment of the present invention, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art
The every other embodiment obtained under the premise of creative work is not made, belongs to the scope of protection of the invention.
Fig. 1 is refer to, Fig. 1 is a kind of schematic flow sheet of short-term load forecasting method provided in an embodiment of the present invention.
This method includes:
S11:Obtain historical load data after pretreatment and weather data;
It should be noted that the historical load data of collection includes the load data of prediction 1 day a few days ago;Weather data includes
Predict the weather data of 1 day a few days ago and predict the data of weather forecast of day.The temporal resolution of load data collection can be small for 1
When, naturally it is also possible to for other concrete numerical values.Weather data can include prediction 1 day a few days ago and predict the weather forecast number of day
According to specifically including the daily highest temperature, the lowest temperature, rainfall and humidity, naturally it is also possible to for other weather factor numbers
According to.
S12:Historical load data is decomposed into the component data of multiple different frequency ranges using wavelet packet decomposition;
It should be noted that historical load data (i.e. historical load time series signal) is passed through a low pass filter
H and high-pass filter G is filtered, and one group of low frequency signal and one group of high frequency are obtained after being decomposed by two wave filters
Each frequency band of this layer, is then further broken into next layer of two sub-bands by signal.For example by historical load data point
Solve the low frequency signal A1 and high-frequency signal D1 for first layer, then by low frequency signal A1 be decomposed into the second layer low frequency signal AA2 and
High-frequency signal AD2, high-frequency signal D1 is decomposed into the low frequency signal DA2 and high-frequency signal DD2 of the second layer, to each of the second layer
Individual signal, which decompose, carries out each component of signal that a step obtains third layer decomposition, namely third layer has 8 data components, number
It is respectively AAA3, DAA3, ADA3, DDA3, AAD3, DAD3, ADD3, DDD3 according to component, wherein, AAA3, AAD3 are decomposed by AA2
Arrive, ADA3, ADD3 are decomposed by AD2 and obtained, DAA3, DAD3 are decomposed by DA2 and obtained, DDA3, DDD3 are decomposed by DD2 and obtained.
Specifically, each corresponding wavelet packet coefficient (low-frequency wavelet coefficients can be obtained by WAVELET PACKET DECOMPOSITION relational expression
Or high-frequency wavelet coefficient), so that each corresponding component data (i.e. component signal) is obtained, wherein, WAVELET PACKET DECOMPOSITION relational expression
For:
Wherein,For+1 layer of low-frequency wavelet coefficients of jth,For+1 layer of jth
High-frequency wavelet coefficient,For the wavelet coefficient of jth layer;hk-2lAnd gk-2lThe respectively coefficient of WAVELET PACKET DECOMPOSITION conjugate filter;j
For the number of plies of decomposition, n >=1.
S13:Carry out single branch reconstruct respectively to each component data by wavelet package reconstruction algorithm and obtain each the first sub- sequence
Row;
Reconstructed specifically, single branch can be carried out to each component data according to wavelet package reconstruction relational expression, obtain each the
One subsequence, wherein, wavelet package reconstruction relational expression is:
Wherein, pl-2kAnd ql-2kRespectively wavelet reconstruction conjugate filter
Coefficient.
In actual applications, WAVELET PACKET DECOMPOSITION and reconstruct can be realized by MATLAB platforms, can specifically be passed through respectively
Function " wavedec " and " wrcoef " realize, and load data can also be decomposed into 3 layers, that is, obtains 8 component datas,
And corresponding 8 the first subsequences, certain Decomposition order can also be other positive integers.
S14:Weather data is added into each first subsequence respectively, each the second subsequence is obtained;
S15:Each second subsequence is inputted into the extreme learning machine Optimized model pre-established and is predicted, is obtained
Predicted the outcome with the one-to-one each height of each second subsequence;
S16:Each height is predicted the outcome and is overlapped, short-term load forecasting result is obtained;
First pass through the extreme learning machine Optimized model pre-established and analysis is predicted respectively to each second subsequence, it is right
Each second subsequence, which is predicted, to be obtained a corresponding son and predicts the outcome, and further each height predicts the outcome
It is overlapped and can obtain final short-term load forecasting result.
Wherein, Fig. 2 is refer to, Fig. 2 is a kind of foundation of extreme learning machine Optimized model provided in an embodiment of the present invention
The schematic flow sheet of journey, its detailed process is:
S21:Training sample data are obtained, and training sample data are inputted into extreme learning machine;
S22:Hybrid algorithm is searched for using population gravitation optimizing processing is carried out to the parameter of extreme learning machine, obtain optimal
Parameter, with the extreme learning machine Optimized model after being trained, wherein, parameter can include input weights and hidden layer biasing;
Further, the parameter of extreme learning machine is entered using training sample data and population gravitation search hybrid algorithm
Row optimizing is handled, and the process for obtaining optimized parameter is specifically as follows:
S221:According to the stochastic parameter initialization population of extreme learning machine, parameter includes input weights and hidden layer biasing;
Need explanation, according to embodiments of the present invention employed in training sample data can determine the limit learn
The nodes of the network topology structure of machine and each layer, if input layer is n, hidden layer node is l, and determines population scale M,
The dimension of particle is D=n*l+l, parameter to be optimized can be determined according to the dimension of population, parameter to be optimized can be
Input weights and the hidden layer biasing of extreme learning machine, and maximum iteration maxgen, for example, 500 can be pre-set
It is inferior.
Specifically, after optimal input weights and the biasing of optimal hidden layer are obtained, can further pass through following meter
Calculate relational expression and obtain optimal output weight matrix, be specially:
The output weight matrix of extreme learning machine is calculated according to Pseudoinverse algorithm, wherein, corresponding calculation relational expression be β=
H + In T, formula, β is the output weight matrix of extreme learning machine, and T is the output objective matrix of training sample data, and H is hidden layer
Output matrix, hidden layer output matrix can be represented by the formula:
Wherein, g is hidden layer activation primitive, is specifically as follows Sigmoid functions.
It should also be noted that, randomly generating initial population for X=[X1,X2,...,XM]T, wherein, i-th of particle is:
Xi=[w11,w12,…,w1l,w21,w22,…,w2l,…,wn1,wn2,…,wnl,b1,b2,…,bl]
Wherein, i=1,2 ..., M, wnlFor n-th of node of input layer to the input weights of l-th of node of hidden layer, blFor
The biasing of l-th of node of hidden layer.
S222:The primary of each in initial population is calculated using training sample data and fitness value calculation relational expression
Fitness value, wherein:
Fitness value calculation relational expression isWherein, fitiFor the suitable of i-th primary
Answer angle value,For predicted value, P is desired output, and N is the sum that predicted value is exported in training sample data, i=1,2 ...,
M, M are primary sum;
Specifically, training sample data are divided into the input data and desired output data of extreme learning machine model, number is inputted
Gone through according to for the time series of the 1st day, the 2nd day, the 3rd day ..., the n-th -3 days, the n-th -2 days before prediction day (being designated as n-th day)
History load data and the weather data of the 1st day to n-th day;The desired output of extreme learning machine is the time of the 2nd day to (n-1)th day
Sequence history load data.For example:1st input data of the training sample data of model for the 1st day historical load data,
The weather data of the 1st day and the weather data of the 2nd day, the 1st desired output data are the historical load data of the 2nd day;2nd
Individual input data is historical load data, the weather data of the 2nd day and the weather data of the 3rd day of the 2nd day, the 2nd expectation
Output data is the historical load number of the 3rd day, by that analogy.
S223:The position and speed of each primary in initial population are updated, current population is obtained, and count
Calculate the fitness value of each current particle in current population;
Specifically, the position of each particle in population and speed can be updated by following calculation relational expression,
And the fitness value of each current particle in the current population after updating is calculated, it is specific as follows:
In formula:Represent i-th of particle at the t times respectively
In the home position of d dimensions, raw velocity, acceleration, the suffered size made a concerted effort, inertia mass, the speed updated, renewal during iteration
Position, by adjusting c1、c2Value equilibrium particle between gravitation and global information between exchange, in embodiments of the present invention
Choose c1=0.5, c2=1.5.
Wherein, the calculation formula made a concerted effort is as follows:
Wherein, in formula:N is total number of particles;Represent gravitation of the particle j to particle i;randjFor the random of [0,1]
Number;Rij(t) it is particle i and particle j Euclidean distance;ε is the constant of a value very little;
G (t) is gravitational constant;T is iterations;Maxgent is maximum iteration.
The inertia mass of particle can be obtained by following formula:
In formula:fiti(t) fitness of i-th of particle in the t times iteration is represented,
S224:If the fitness value of i-th of current particle in current population is less than i-th of initial grain in initial population
The fitness value of son, then regard i-th of current particle as i-th of particle in new population;If i-th in current population is current
The fitness value of particle is more than the fitness value of i-th of primary in initial population, then using i-th of primary as new
I-th of particle in population, to obtain new population;
Specifically, by population each renewal before parent particle fitness value it is corresponding with each, renewal after son
It is compared for the fitness value of particle, new population is obtained according to the above method.
S225:Judge whether current iteration number of times reaches maximum iteration, if it is, finding adaptation from new population
The optimal particle of angle value is exported as global optimum's particle, and obtains optimized parameter according to global optimum's particle;Otherwise, after
Continuous iteration next time, until reaching maximum iteration.
It should be noted that after new population is obtained, judge whether current iterations reaches maximum iteration,
If it is not, iterations adds 1 (t=t+1), and position to each current particle in current population and speed are carried out more
Newly, the new population to obtain, until iterations reaches maximum iteration (i.e. t > max gen).When iterations reaches most
After big iterations, the optimal particle of fitness value is obtained from the new population finally obtained as global optimum particle gbest,
So as to obtain optimized parameter (such as best initial weights and optimum threshold) corresponding with optimal particle gbest, extreme learning machine.
Optionally, the temporal resolution of each load data is 1 hour.
It should be noted that the temporal resolution of the wind speed historical data in the embodiment of the present invention was not limited only to as 1 hour,
Temporal resolution can also be other concrete numerical values, and such as 0.5 hour, the application was not specifically limited to this.
Optionally, weather data includes the daily highest temperature, the lowest temperature, rainfall and humidity.
Certainly, weather data is not limited only to include above-mentioned several data, can also include other kinds of data, the application
Special restriction is not done to this.
The embodiments of the invention provide a kind of short-term load forecasting method, including obtain historical load number after pretreatment
According to and weather data;Historical load data is decomposed into the component data of multiple different frequency ranges using wavelet packet decomposition;Pass through
Wavelet package reconstruction algorithm carries out single branch reconstruct to each component data and obtains each the first subsequence respectively;Respectively to each first
Weather data is added in subsequence, each the second subsequence is obtained;Each second subsequence is inputted to the limit pre-established
It is predicted in learning machine Optimized model, obtains predicting the outcome with the one-to-one each height of each the second subsequence;By each
Son, which predicts the outcome, to be overlapped, and obtains short-term load forecasting result;Wherein, the process of setting up of extreme learning machine Optimized model is to obtain
Training sample data are taken, and the training sample data are inputted into extreme learning machine;Searched for and mixed using population gravitation
Algorithm carries out optimizing processing, the extreme learning machine Optimized model after being trained to the parameter of extreme learning machine;Parameter includes defeated
Enter weights and hidden layer biasing.The embodiment of the present invention carries out resolution process using wavelet packet decomposition to historical load data, obtains
To multiple data components, influence of the fluctuation of load to load prediction accuracy is reduced, and in the embodiment of the present invention
Extreme learning machine model is to search for after hybrid algorithm carries out optimizing to the parameter of extreme learning machine to set up by population gravitation,
Each subsequence is predicted using extreme learning machine Optimized model provided in an embodiment of the present invention, extreme learning machine is improved
Estimated performance, strengthen model local search ability, improve global convergence precision, make to predict the outcome more accurate.
Accordingly the embodiment of the invention also discloses a kind of short-term load forecasting device, Fig. 3 is specifically refer to, Fig. 3 is this
A kind of structural representation for short-term load forecasting device that inventive embodiments are provided.On the basis of above-described embodiment:
The device, including:
Acquisition module 11, for obtaining historical load data after pretreatment and weather data;
Decomposing module 12, the component for historical load data to be decomposed into multiple different frequency ranges using wavelet packet decomposition
Data;
Reconstructed module 13, obtains each for carrying out single branch reconstruct respectively to each component data by wavelet package reconstruction algorithm
Individual first subsequence;
Add module 14, for adding weather data into each first subsequence respectively, obtains each the second subsequence;
Prediction module 15, for inputting into the extreme learning machine Optimized model pre-established each second subsequence
Row prediction, obtains predicting the outcome with the one-to-one each height of each the second subsequence;
Laminating module 16, is overlapped for each height to be predicted the outcome, and obtains short-term load forecasting result;
Wherein, extreme learning machine Optimized model is included with lower module, specifically refer to Fig. 4:
Acquisition module 21, is inputted to extreme learning machine for obtaining training sample data, and by the training sample data
In;
Optimization module 22, is carried out at optimizing for searching for hybrid algorithm using population gravitation to the parameter of extreme learning machine
Reason, the extreme learning machine Optimized model after being trained;Parameter includes input weights and hidden layer biasing.
Optionally, optimization module 22 includes:
Initialization unit, for according to extreme learning machine stochastic parameter initialize population, parameter include input weights and
Hidden layer is biased;
First computing unit, for being calculated using training sample data and fitness value calculation relational expression in initial population
The fitness value of each primary, wherein:
Fitness value calculation relational expression isWherein, fitiFor the suitable of i-th primary
Answer angle value,For predicted value, P is desired output, and N is the sum that predicted value is exported in training sample data, i=1,2 ...,
M, M are primary sum;
Updating block, is updated for the position to each primary in initial population and speed, obtains current
Population;
Second computing unit, the fitness value for calculating each current particle in current population;
Optimizing unit, the fitness value for i-th of current particle in current population is less than i-th in initial population
The fitness value of individual primary, then regard i-th of current particle as i-th of particle in new population;In current population
The fitness value of i-th of current particle is more than the fitness value of i-th of primary in initial population, then initial by i-th
Particle is as i-th of particle in new population, to obtain new population;
Judging unit, for judging whether current iteration number of times reaches maximum iteration, if it is, from new population
Find the optimal particle of fitness value to be exported as global optimum's particle, and optimal ginseng is obtained according to global optimum's particle
Number;Otherwise, next iteration is continued, until reaching maximum iteration.
It should be noted that the embodiment of the present invention carries out resolution process using wavelet packet decomposition to historical load data,
Multiple data components are obtained, influence of the fluctuation of load to load prediction accuracy is reduced, and in the embodiment of the present invention
Extreme learning machine model be to search for hybrid algorithm by population gravitation the parameter of extreme learning machine is carried out to set up after optimizing
, each subsequence is predicted using extreme learning machine Optimized model provided in an embodiment of the present invention, the limit is improved
The estimated performance of habit machine, strengthens model local search ability, improves global convergence precision, makes to predict the outcome more accurate.
In addition, for short-term load forecasting method involved in the embodiment of the present invention specific introduction refer to it is above-mentioned
Embodiment of the method, the application will not be repeated here.
The embodiment of the present invention additionally provides a kind of Short Term Load Forecasting System, including short-term load forecasting described above dress
Put.
The embodiment of the present invention carries out resolution process using wavelet packet decomposition to historical load data, obtains multiple data point
Amount, reduces influence of the fluctuation of load to load prediction accuracy, and the extreme learning machine mould in the embodiment of the present invention
Type is to search for after hybrid algorithm carries out optimizing to the parameter of extreme learning machine to set up by population gravitation, using of the invention real
The extreme learning machine Optimized model for applying example offer is predicted to each subsequence, improves the estimated performance of extreme learning machine,
Model local search ability is strengthened, global convergence precision is improved, makes to predict the outcome more accurate.In addition, for the present invention
The specific introduction of involved short-term load forecasting method refer to above method embodiment in embodiment, and the application is herein not
Repeat again.
The embodiment of each in this specification is described by the way of progressive, and what each embodiment was stressed is and other
Between the difference of embodiment, each embodiment identical similar portion mutually referring to.For device disclosed in embodiment
For, because it is corresponded to the method disclosed in Example, so description is fairly simple, related part is said referring to method part
It is bright.
It should also be noted that, in this manual, such as first and second or the like relational terms be used merely to by
One entity or operation make a distinction with another entity or operation, and not necessarily require or imply these entities or operation
Between there is any this actual relation or order.Moreover, term " comprising ", "comprising" or its any other variant meaning
Covering including for nonexcludability, so that process, method, article or equipment including a series of key elements not only include that
A little key elements, but also other key elements including being not expressly set out, or also include be this process, method, article or
The intrinsic key element of equipment.In the absence of more restrictions, the key element limited by sentence "including a ...", is not arranged
Except also there is other identical element in the process including the key element, method, article or equipment.
Professional further appreciates that, with reference to the unit of each example of the embodiments described herein description
And algorithm steps, can be realized with electronic hardware, computer software or the combination of the two, in order to clearly demonstrate hardware and
The interchangeability of software, generally describes the composition and step of each example according to function in the above description.These
Function is performed with hardware or software mode actually, depending on the application-specific and design constraint of technical scheme.Specialty
Technical staff can realize described function to each specific application using distinct methods, but this realization should not
Think beyond the scope of this invention.
Directly it can be held with reference to the step of the method or algorithm that the embodiments described herein is described with hardware, processor
Capable software module, or the two combination are implemented.Software module can be placed in random access memory (RAM), internal memory, read-only deposit
Reservoir (ROM), electrically programmable ROM, electrically erasable ROM, register, hard disk, moveable magnetic disc, CD-ROM or technology
In the storage medium of any other forms well known in field.
The foregoing description of the disclosed embodiments, enables professional and technical personnel in the field to realize or using the present invention.
A variety of modifications to these embodiments will be apparent for those skilled in the art, as defined herein
General Principle can be realized in other embodiments without departing from the spirit or scope of the present invention.Therefore, it is of the invention
The embodiments shown herein is not intended to be limited to, and is to fit to and principles disclosed herein and features of novelty phase one
The most wide scope caused.
Claims (8)
1. a kind of short-term load forecasting method, it is characterised in that including:
S11:Obtain historical load data after pretreatment and weather data;
S12:The historical load data is decomposed into the component data of multiple different frequency ranges using wavelet packet decomposition;
S13:Carry out single branch reconstruct respectively to component data each described by wavelet package reconstruction algorithm and obtain each the first sub- sequence
Row;
S14:Respectively to weather data is added in first subsequence each described, each the second subsequence is obtained;
S15:Each described second subsequence is inputted into the extreme learning machine Optimized model pre-established and is predicted, is obtained
Predicted the outcome with the one-to-one each height of the second subsequence each described;
S16:Each described son is predicted the outcome and is overlapped, short-term load forecasting result is obtained;
Wherein, the process of setting up of the extreme learning machine Optimized model is:
S21:Training sample data are obtained, and the training sample data are inputted into extreme learning machine;
S22:Hybrid algorithm is searched for using population gravitation optimizing processing is carried out to the parameter of extreme learning machine, after being trained
The extreme learning machine Optimized model;The parameter includes input weights and hidden layer biasing.
2. short-term load forecasting method according to claim 1, it is characterised in that the use training sample data and grain
Subgroup gravitation searches for hybrid algorithm:
S221:According to the stochastic parameter initialization population of extreme learning machine, the parameter includes input weights and hidden layer biasing;
S222:Each primary in the initial population is calculated using training sample data and fitness value calculation relational expression
Fitness value, wherein:
The fitness value calculation relational expression isWherein, fitiFor the adaptation of i-th of primary
Angle value,For predicted value, P is desired output, and N is the sum that predicted value is exported in the training sample data, i=1,2 ...,
M, M are primary sum;
S223:The position and speed of each primary in the initial population are updated, current population is obtained,
And calculate the fitness value of each current particle in the current population;
S224:If the fitness value of i-th of current particle in current population is less than i-th of initial grain in the initial population
The fitness value of son, then regard i-th of current particle as i-th of particle in new population;If i-th in current population is current
The fitness value of particle is more than the fitness value of i-th of primary in the initial population, then makees i-th of primary
For i-th of particle in new population, to obtain new population;
S225:Judge whether current iteration number of times reaches maximum iteration, if it is, finding adaptation from the new population
The optimal particle of angle value is exported as global optimum's particle, and obtains optimized parameter according to global optimum's particle;It is no
Then, next iteration is continued, until reaching maximum iteration.
3. short-term load forecasting method according to claim 1, it is characterised in that the historical load data is prediction day
The load data of first 1 day;The weather data includes the weather data of prediction the previous day day and predicts the data of weather forecast of day.
4. short-term load forecasting method according to claim 3, it is characterised in that the time of each load data point
Resolution is 1 hour.
5. short-term load forecasting method according to claim 3, it is characterised in that the weather data is included daily most
High temperature, the lowest temperature, rainfall and humidity.
6. a kind of short-term load forecasting device, it is characterised in that including:
Acquisition module, for obtaining historical load data after pretreatment and weather data;
Decomposing module, the number of components for the historical load data to be decomposed into multiple different frequency ranges using wavelet packet decomposition
According to;
Reconstructed module, each is obtained for carrying out single branch reconstruct respectively to component data each described by wavelet package reconstruction algorithm
First subsequence;
Add module, for weather data is added in first subsequence each described, obtaining each the second subsequence respectively;
Prediction module, for each described second subsequence to be inputted into the extreme learning machine Optimized model pre-established into progress
Prediction, obtains predicting the outcome with the one-to-one each height of the second subsequence each described;
Laminating module, is overlapped for each described son to be predicted the outcome, obtains short-term load forecasting result;
Wherein, the extreme learning machine Optimized model includes:
Acquisition module, is inputted into extreme learning machine for obtaining training sample data, and by the training sample data;
Optimization module, carries out optimizing processing to the parameter of extreme learning machine for searching for hybrid algorithm using population gravitation, obtains
The extreme learning machine Optimized model after to training;The parameter includes input weights and hidden layer biasing.
7. short-term load forecasting device according to claim 6, it is characterised in that the optimization module includes:
Initialization unit, for according to extreme learning machine stochastic parameter initialize population, the parameter include input weights and
Hidden layer is biased;
First computing unit, for being calculated using training sample data and fitness value calculation relational expression in the initial population
The fitness value of each primary, wherein:
The fitness value calculation relational expression isWherein, fitiFor the adaptation of i-th of primary
Angle value,For predicted value, P is desired output, and N is the sum that predicted value is exported in the training sample data, i=1,2 ...,
M, M are primary sum;
Updating block, is updated for the position to each primary in the initial population and speed, obtains
Current population;
Second computing unit, the fitness value for calculating each current particle in the current population;
Optimizing unit, the fitness value for i-th of current particle in current population is less than i-th in the initial population
The fitness value of individual primary, then regard i-th of current particle as i-th of particle in new population;In current population
The fitness value of i-th of current particle is more than the fitness value of i-th of primary in the initial population, then by i-th
Primary is as i-th of particle in new population, to obtain new population;
Judging unit, for judging whether current iteration number of times reaches maximum iteration, if it is, from the new population
Find the optimal particle of fitness value to be exported as global optimum's particle, and obtain optimal according to global optimum's particle
Parameter;Otherwise, next iteration is continued, until reaching maximum iteration.
8. a kind of Short Term Load Forecasting System, it is characterised in that including short-term load forecasting dress as claimed in claims 6 or 7
Put.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710741552.1A CN107292462A (en) | 2017-08-25 | 2017-08-25 | A kind of short-term load forecasting method, apparatus and system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710741552.1A CN107292462A (en) | 2017-08-25 | 2017-08-25 | A kind of short-term load forecasting method, apparatus and system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107292462A true CN107292462A (en) | 2017-10-24 |
Family
ID=60106629
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710741552.1A Pending CN107292462A (en) | 2017-08-25 | 2017-08-25 | A kind of short-term load forecasting method, apparatus and system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107292462A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108256676A (en) * | 2017-12-29 | 2018-07-06 | 国网江苏省电力有限公司检修分公司 | A kind of Methods of electric load forecasting of meter and load fluctuation asymmetrical characteristic |
CN108259099A (en) * | 2018-01-25 | 2018-07-06 | 湘潭大学 | A kind of TD-SCDMA base station electromagnetic radiation Forecasting Methodology |
CN108847674A (en) * | 2018-07-27 | 2018-11-20 | 广东电网有限责任公司 | A kind of customer charge based on WAVELET PACKET DECOMPOSITION is adjustable potentiality calculation method and device |
CN109146183A (en) * | 2018-08-24 | 2019-01-04 | 广东工业大学 | Short-term impact load forecasting model method for building up based on signal decomposition and intelligent optimization algorithm |
CN112330078A (en) * | 2021-01-04 | 2021-02-05 | 南方电网数字电网研究院有限公司 | Power consumption prediction method and device, computer equipment and storage medium |
CN114548845A (en) * | 2022-04-27 | 2022-05-27 | 北京智芯微电子科技有限公司 | Distribution network management method, device and system |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105976051A (en) * | 2016-04-29 | 2016-09-28 | 武汉大学 | Wavelet transformation and improved firefly-optimized extreme learning machine-based short-term load prediction method |
-
2017
- 2017-08-25 CN CN201710741552.1A patent/CN107292462A/en active Pending
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105976051A (en) * | 2016-04-29 | 2016-09-28 | 武汉大学 | Wavelet transformation and improved firefly-optimized extreme learning machine-based short-term load prediction method |
Non-Patent Citations (3)
Title |
---|
YOGESH MANJHI: "Forecasting Energy Consumption using Particle", 《FORECASTING ENERGY CONSUMPTION USING PARTICLE》 * |
孟安波等: "基于虚拟预测与小波包变换的风电功率组合预测", 《电力系统保护与控制》 * |
程宇也: "基于人工神经网络的短期电力负荷预测研究", 《中国优秀硕士学位论文全文数据库 工程科技Ⅱ辑》 * |
Cited By (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108256676A (en) * | 2017-12-29 | 2018-07-06 | 国网江苏省电力有限公司检修分公司 | A kind of Methods of electric load forecasting of meter and load fluctuation asymmetrical characteristic |
CN108256676B (en) * | 2017-12-29 | 2021-11-16 | 国网江苏省电力有限公司检修分公司 | Power load prediction method considering load fluctuation asymmetry characteristic |
CN108259099A (en) * | 2018-01-25 | 2018-07-06 | 湘潭大学 | A kind of TD-SCDMA base station electromagnetic radiation Forecasting Methodology |
CN108847674A (en) * | 2018-07-27 | 2018-11-20 | 广东电网有限责任公司 | A kind of customer charge based on WAVELET PACKET DECOMPOSITION is adjustable potentiality calculation method and device |
CN108847674B (en) * | 2018-07-27 | 2021-01-26 | 广东电网有限责任公司 | User load adjustable potential calculation method and device based on wavelet packet decomposition |
CN109146183A (en) * | 2018-08-24 | 2019-01-04 | 广东工业大学 | Short-term impact load forecasting model method for building up based on signal decomposition and intelligent optimization algorithm |
CN112330078A (en) * | 2021-01-04 | 2021-02-05 | 南方电网数字电网研究院有限公司 | Power consumption prediction method and device, computer equipment and storage medium |
CN114548845A (en) * | 2022-04-27 | 2022-05-27 | 北京智芯微电子科技有限公司 | Distribution network management method, device and system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107292462A (en) | A kind of short-term load forecasting method, apparatus and system | |
CN111161535B (en) | Attention mechanism-based graph neural network traffic flow prediction method and system | |
Kang et al. | Short-term traffic flow prediction with LSTM recurrent neural network | |
Luo et al. | Design of a combined wind speed forecasting system based on decomposition-ensemble and multi-objective optimization approach | |
CN107622329A (en) | The Methods of electric load forecasting of Memory Neural Networks in short-term is grown based on Multiple Time Scales | |
CN105389980B (en) | Short-time Traffic Flow Forecasting Methods based on long short-term memory recurrent neural network | |
CN109754113A (en) | Load forecasting method based on dynamic time warping Yu length time memory | |
CN107392397A (en) | A kind of short-term wind speed forecasting method, apparatus and system | |
CN103226741B (en) | Public supply mains tube explosion prediction method | |
CN105096614B (en) | Newly-built crossing traffic flow Forecasting Methodology based on generation moldeed depth belief network | |
CN110942637B (en) | SCATS system road traffic flow prediction method based on airspace map convolutional neural network | |
CN107102969A (en) | The Forecasting Methodology and system of a kind of time series data | |
CN108091135A (en) | Parking position multistep forecasting method based on Optimization of Wavelet neutral net | |
CN107516150A (en) | A kind of Forecasting Methodology of short-term wind-electricity power, apparatus and system | |
CN106781489A (en) | A kind of road network trend prediction method based on recurrent neural network | |
CN105046953A (en) | Short-time traffic-flow combination prediction method | |
Han et al. | Network traffic prediction using variational mode decomposition and multi-reservoirs echo state network | |
CN106803135A (en) | The Forecasting Methodology and device of a kind of photovoltaic power generation system output power | |
Li et al. | A k-nearest neighbor locally weighted regression method for short-term traffic flow forecasting | |
CN106600959A (en) | Traffic congestion index-based prediction method | |
CN107506861A (en) | A kind of short-term wind speed forecasting method, apparatus and system | |
CN111027662A (en) | SD-LSSVR short-time traffic flow prediction method based on chaos quantum particle swarm optimization | |
CN110119838A (en) | A kind of shared bicycle demand forecast system, method and device | |
CN105184398A (en) | Power maximum load small-sample prediction method | |
CN115496257A (en) | Short-term vehicle speed prediction based on space-time fusion |
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: 20171024 |