CN110348630A - A kind of isolated island region Methods of electric load forecasting and system - Google Patents
A kind of isolated island region Methods of electric load forecasting and system Download PDFInfo
- Publication number
- CN110348630A CN110348630A CN201910616995.7A CN201910616995A CN110348630A CN 110348630 A CN110348630 A CN 110348630A CN 201910616995 A CN201910616995 A CN 201910616995A CN 110348630 A CN110348630 A CN 110348630A
- Authority
- CN
- China
- Prior art keywords
- neural network
- network model
- historical data
- isolated island
- 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
- 238000000034 method Methods 0.000 title claims abstract description 37
- 238000003062 neural network model Methods 0.000 claims abstract description 60
- 239000000284 extract Substances 0.000 claims abstract description 10
- 238000000605 extraction Methods 0.000 claims abstract description 5
- 238000013528 artificial neural network Methods 0.000 claims description 13
- 102000008297 Nuclear Matrix-Associated Proteins Human genes 0.000 claims description 9
- 108010035916 Nuclear Matrix-Associated Proteins Proteins 0.000 claims description 9
- 210000000299 nuclear matrix Anatomy 0.000 claims description 9
- 230000007704 transition Effects 0.000 claims description 7
- 210000004218 nerve net Anatomy 0.000 claims description 5
- 238000009825 accumulation Methods 0.000 claims description 3
- 230000005540 biological transmission Effects 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 230000010152 pollination Effects 0.000 claims description 3
- 238000001556 precipitation Methods 0.000 claims description 3
- 238000000513 principal component analysis Methods 0.000 claims description 2
- 238000011084 recovery Methods 0.000 claims description 2
- 230000005611 electricity Effects 0.000 claims 2
- 238000005457 optimization Methods 0.000 abstract description 7
- 238000013439 planning Methods 0.000 abstract description 3
- 230000002596 correlated effect Effects 0.000 description 3
- 240000002853 Nelumbo nucifera Species 0.000 description 2
- 235000006508 Nelumbo nucifera Nutrition 0.000 description 2
- 235000006510 Nelumbo pentapetala Nutrition 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 125000002015 acyclic group Chemical group 0.000 description 1
- 239000000470 constituent Substances 0.000 description 1
- 238000009313 farming Methods 0.000 description 1
- 238000013277 forecasting method Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
Classifications
-
- 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
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- 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
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Biophysics (AREA)
- Data Mining & Analysis (AREA)
- Software Systems (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- Evolutionary Computation (AREA)
- Tourism & Hospitality (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Public Health (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Water Supply & Treatment (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to a kind of isolated island region Methods of electric load forecasting, this method comprises: step 1, obtains the historical data for needing the electric power in the isolated island region predicted, extract core principle component analysis to historical data；Step 2, it is optimized using parameter of the pollen algorithm to neural network model, training obtains the neural network model of isolated island region load forecast；Step 3, the historical data input neural network model after extraction core principle component is subjected to load forecast.By constructing the power load forecasting module based on pollen algorithm and BP neural network model, the calculating and prediction of isolated island type region electric load can be achieved, the accuracy of load forecast is improved, can be isolated island type regional generation planning optimization and adjustment, reasonable employment electric power resource provides effectively reference.
Description
Technical field
The present invention relates to electric power project engineering field more particularly to a kind of isolated island region Methods of electric load forecasting and it is
System.
Background technique
Power system load is a part important in electric system, and accurate load prediction helps to ensure electric system
Safe and stable operation, electric load analysis prediction with control management is realize electric power ordering management important link,
It is the component part of energy source optimization in energy internet system.
Since power system load is under certain conditions there is obvious trend, such as farming power, have
Certain regularity, this regularity can be linear or nonlinear, periodical or acyclic therefore existing
Methods of electric load forecasting is compared by the way that passing data are carried out with the algorithm based on experience mostly, and prediction result is not accurate enough.
In addition, there is various factors for electric load, therefore the result of existing Methods of electric load forecasting prediction is unsatisfactory.
Summary of the invention
The present invention for the technical problems in the prior art, provide a kind of isolated island region Methods of electric load forecasting and
System.
The technical scheme to solve the above technical problems is that a kind of isolated island region Methods of electric load forecasting, packet
It includes:
Step 1, the historical data for needing the electric power in the isolated island region predicted is obtained, core is extracted to the historical data
Principal component analysis；
Step 2, it is optimized using parameter of the pollen algorithm to the neural network model, training obtains the isolated island area
The neural network model of domain load forecast；
Step 3, the historical data after extraction core principle component is inputted into the neural network model and carries out electric load
Prediction.
A kind of isolated island region Electric Load Prediction System, the system comprises: historical data obtains module, neural network mould
Type training module and load forecast module；
The historical data obtains module, for obtaining the historical data for needing the electric power in the isolated island region predicted, to institute
It states historical data and extracts core principle component analysis；
The neural network model training module, for being carried out using parameter of the pollen algorithm to the neural network model
Optimization, training obtain the neural network model of isolated island region load forecast；
The load forecast module inputs the nerve net for that will extract the historical data after core principle component
Network model carries out load forecast.
The beneficial effects of the present invention are: passing through load forecast of the building based on pollen algorithm and BP neural network model
Model, it can be achieved that isolated island type region electric load calculating and prediction, improve the accuracy of load forecast, can be isolated island type
Regional generation planning optimization and adjustment, reasonable employment electric power resource provide effectively reference.
Based on the above technical solution, the present invention can also be improved as follows.
Further, it is obtained in the step 1 and needs the historical data of isolated island region electric power predicted to include:
The maximum temperature in the isolated island region for needing to predict, mean temperature, minimum temperature, season type, month, precipitation shape
Condition, day type, wind speed, humidity, on the day of prediction day before the load value at 3 moment and the load of prediction 5 days synchronizations a few days ago
Value.
It further include to the historical data before extracting core principle component analysis to the historical data in the step 1
It is normalized:
Wherein, xiFor the current value of i-th of load data in the historical data, xminAnd xmaxThe respectively described history number
The minimum value and maximum value in load data in.
After the historical data is normalized in the step 1, core master is extracted to the historical data
The process of constituent analysis includes:
Step 101, by n index of m sample in the historical data, write as the data matrix A of m × n dimension；
Step 102, kernel function and relevant parameter are determined, nuclear matrix K is calculateduv:
Kuv≡(Φ(xu)·Φ(xv))；
Step 103, the nuclear matrix K is correcteduv:
M is the size of sample space；
Step 104, the nuclear matrix K is calculateduvEigenvalue λ1,λ2,…,λnAnd feature vector v1,v2,…,vn；
Step 105, to the eigenvalue λ1,λ2,…,λnWith described eigenvector v1,v2,…,vnRespectively according to descending
Mode sorts to obtain λ '1> λ '2> ... > λ 'nWith v '1,v'2,…,v'n；
Step 106, described eigenvector unit orthogonalization, α is obtained1,α2,…,αn；
Step 107, the eigenvalue λ is calculated1,λ2,…,λnAccumulation contribution rate B1,B2,…,Bn, according to the extraction of setting
Rate p, works as Bt>=p then chooses t and principal component α1,α2,…,αt；
Step 108, sample X is calculated in α1,α2,…,αtOn projection Y, Y=Kuv·α。
The neural network model of the isolated island region load forecast created in the step 1 is BP neural network model；
Extracting main component and obtaining the parameter of the BP neural network model includes: the neural network number of plies, each node layer
Number, hidden layer, the transmission function of output layer and training function.
Include: using the process that parameter of the pollen algorithm to the neural network model optimizes in the step 2
Step 201, by the parameter of the BP neural network model, parameter is encoded as a whole, determines that FPA is calculated
Transition probability p and maximum number of iterations T is arranged in the search space dimension D of method；
Step 202, the fitness function of the pollen algorithm is determined:
For i-th of sample reality output；yiFor i-th of sample desired output；I=1,2,3 ..., n, n are sample number
Amount；
Step 203, the position of random initializtion individual calculates the fitness letter of each individual according to the fitness function
Numerical value retains the smallest individual of fitness value；
Step 204, transition probability P:p=0.8+0.2 × rand is calculated, rand ∈ [0,1] is randomly generated, judges that conversion is general
When rate P > rand, global search is carried out；Otherwise local pollination is carried out；
Step 205, the fitness function value that each pollen is calculated according to the fitness function finds out current optimal solution；
Step 206, terminate when judgement meets the condition that pollen algorithm terminates, pollen individual is decoded as the BP nerve net
The weight of network model and the initial value of threshold value.
Code length L in the step 201 are as follows:
L=n × m+m × s+m+s；
N is input layer number, and m is the number of hidden nodes, and s is output layer number of nodes.
The formula of global search is carried out in the step 204 are as follows:The public affairs locally pollinated
Formula are as follows:
Wherein, L is the random real number for obeying the L é vy distribution of index 1.5；g*It is current globally optimal solution；γ is [0,1]
Between obey equally distributed random real number；J, k is the random number between [1, n], and meets between i, j, k mutual not phase two-by-two
Deng,It is i-th of sample in the location of t moment.
After the step 206 further include:
According to the initial value of the weight of the BP neural network model and threshold value, the BP neural network model is instructed
Practice；When judgement meets the condition that the BP neural network model terminates, judgement meets the training item of the BP neural network model
The BP neural network model is exported when part, it is no to then follow the steps 206.
Beneficial effect using above-mentioned further scheme is: being acquired and pre-processes to initial data first, obtains original
The correlated characteristic of beginning data.Then it needs to be determined the neural network topology structure in prediction model according to prediction.?
On the basis of this, neural network and prediction model are trained using initial data, when frequency of training and precision of prediction reach
When required target, the load data in certain following a period of time predicted.So as to realize isolated island type
The calculating and prediction of region electric load, improve the accuracy of load forecast, can optimize and adjust for generation schedule, rationally
Effective Technical Reference is provided using electric power resource.
Detailed description of the invention
Fig. 1 is a kind of flow chart of isolated island region Methods of electric load forecasting provided by the invention；
Fig. 2 is a kind of flow chart of the embodiment of isolated island region Methods of electric load forecasting provided by the invention；
Fig. 3 is a kind of structural block diagram of isolated island region Electric Load Prediction System provided by the invention.
In attached drawing, parts list represented by the reference numerals are as follows:
1, historical data obtains module, 2, neural network model training module, 3, load forecast module.
Specific embodiment
The principle and features of the present invention will be described below with reference to the accompanying drawings, and the given examples are served only to explain the present invention, and
It is non-to be used to limit the scope of the invention.
It is certain to consider that the load in isolated island type region has for a kind of isolated island region Methods of electric load forecasting provided by the invention
Periodicity, and the type of load and capacity all compare fixation, are as shown in Figure 1 a kind of isolated island region electric power provided by the invention
The flow chart of load forecasting method, as shown in Figure 1, this method comprises:
Step 1, historical data is obtained, core principle component analysis is carried out to the historical data, main component is extracted and is created
Build the parameter of the neural network model of isolated island region load forecast.
Step 2, it is optimized using parameter of the pollen algorithm to the neural network model.
Step 3, neural network model progress load forecast is utilized after updating the parameter of the neural network model.
The present invention provides a kind of isolated island region Methods of electric load forecasting, is based on pollen algorithm and BP nerve net by building
The power load forecasting module of network model, it can be achieved that isolated island type region electric load calculating and prediction, improve Electric Load Forecasting
The accuracy of survey, can be isolated island type regional generation planning optimization and adjustment, and reasonable employment electric power resource provides effectively reference.
Embodiment 1
Embodiment 1 provided by the invention is a kind of implementation of isolated island region Methods of electric load forecasting provided by the invention
Example, is illustrated in figure 2 a kind of flow chart of the embodiment of isolated island region Methods of electric load forecasting provided by the invention, by Fig. 2
It is found that a kind of embodiment of isolated island region Methods of electric load forecasting provided by the invention includes:
Step 1, obtain the historical data for needing the electric power in isolated island region predicted, to historical data extract core it is main at
Analysis.
It obtains and needs the historical data of isolated island region electric power predicted to include:
The maximum temperature in the isolated island region for needing to predict, mean temperature, minimum temperature, season type, month, precipitation shape
Condition, day type, wind speed, humidity, on the day of prediction day before the load value at 3 moment and the load of prediction 5 days synchronizations a few days ago
Value.
Further include that historical data is normalized before extracting core principle component analysis to historical data:
Wherein, xiFor the current value of i-th of load data in historical data, xminAnd xmaxIt is negative respectively in historical data
The minimum value and maximum value of lotus data.
After historical data is normalized in step 1, the mistake of core principle component analysis is extracted to historical data
Journey includes:
Step 101, by n index of m sample in historical data, write as the data matrix A of m × n dimension.
Step 102, kernel function and relevant parameter are determined, nuclear matrix K is calculateduv:
Kuv≡(Φ(xu)·Φ(xv))。
Step 103, nuclear matrix K is correcteduv:
M is the size of sample space, the i.e. maximum quantity of sample.
Step 104, nuclear matrix K is calculateduvEigenvalue λ1,λ2,…,λnAnd feature vector v1,v2,…,vn。
Step 105, to eigenvalue λ1,λ2,…,λnWith feature vector v1,v2,…,vnIt sorts in the way of descending respectively
Obtain λ '1> λ '2> ... > λ 'nWith v '1,v'2,…,v'n。
Step 106, feature vector unit orthogonalization, α is obtained1,α2,…,αn。
Step 107, eigenvalue λ is calculated1,λ2,…,λnAccumulation contribution rate B1,B2,…,Bn, according to the recovery rate p of setting,
Work as Bt>=p then chooses t and principal component α1,α2,…,αt。
Step 108, sample X is calculated in α1,α2,…,αtOn projection Y, Y=Kuv·α。
The neural network model of the isolated island region load forecast created in step 1 is BP neural network model.
Extract main component obtain the parameter of BP neural network model include: the neural network number of plies, it is each node layer number, hidden
Layer, the transmission function of output layer and training function.
Step 2, it is optimized using parameter of the pollen algorithm to neural network model, training obtains isolated island region power load
The neural network model of lotus prediction.
Include: using the process that parameter of the pollen algorithm to neural network model optimizes in step 2
Step 201, by the parameter of BP neural network model, parameter is encoded as a whole, determines FPA algorithm
Transition probability p and maximum number of iterations T is arranged in search space dimension D.
Code length L in step 201 are as follows:
L=n × m+m × s+m+s.
N is input layer number, and m is the number of hidden nodes, and s is output layer number of nodes.
Pollen algorithm coding method is real coding, and each particle position contains whole power of BP neural network model
Value and threshold value.
Step 202, the fitness function of pollen algorithm is determined:
Select least mean-square error as fitness function herein,It isiA sample reality output；yiIt isiA sample
Desired output；I=1,2,3 ..., n, n are sample size.
Step 203, the position of random initializtion individual calculates the fitness function value of each individual according to fitness function,
Retain the smallest individual of fitness value.
Step 204, transition probability P:p=0.8+0.2 × rand is calculated, rand ∈ [0,1] is randomly generated, judges that conversion is general
When rate P > rand, global search is carried out；Otherwise local pollination is carried out.
The formula of global search is carried out in step 204 are as follows:The formula locally pollinated are as follows:
Wherein, L is the random real number for obeying the L é vy distribution of index 1.5；g*It is current globally optimal solution；γ is [0,1]
Between obey equally distributed random real number；J, k is a random number between [1, n], and meets between i, j, k mutual not phase two-by-two
Deng,It is i-th of sample in the location of t moment.
Step 205, the fitness function value that each pollen is calculated according to fitness function finds out current optimal solution.
Step 206, terminate when judgement meets the condition that pollen algorithm terminates, pollen individual is decoded as BP neural network mould
The weight of type and the initial value of threshold value.
After step 206 further include:
According to the initial value of the weight of BP neural network model and threshold value, BP neural network model is trained；Judgement
When meeting the condition that BP neural network model terminates, judgement exports BP nerve net when meeting the training condition of BP neural network model
Network model, it is no to then follow the steps 206.
Step 3, the historical data input neural network model after extraction core principle component is subjected to load forecast.
A kind of isolated island region Methods of electric load forecasting provided by the invention is first acquired initial data and locates in advance
Reason, obtains the correlated characteristic of initial data.Then according to prediction need to the neural network topology structure in prediction model into
Row determines.On this basis, neural network and prediction model are trained using initial data, when frequency of training and prediction
When precision reaches required target, the load data in certain following a period of time predicted.So as to reality
The calculating and prediction of existing isolated island type region electric load, improve the accuracy of load forecast, can for generation schedule optimization and
Adjustment, reasonable employment electric power resource provide effective Technical Reference.
Embodiment 2
Embodiment 2 provided by the invention is a kind of implementation of isolated island region Electric Load Prediction System provided by the invention
Example, is illustrated in figure 3 a kind of structural block diagram of the embodiment of isolated island region Electric Load Prediction System provided by the invention, by scheming
3 it is found that a kind of embodiment of isolated island region Electric Load Prediction System provided by the invention include: historical data obtain module 1,
Neural network model training module 2 and load forecast module 3.
Historical data obtains module 1, for obtaining the historical data for needing the electric power in the isolated island region predicted, to history number
According to extracting core principle component analysis.
Neural network model training module 2 is instructed for being optimized using parameter of the pollen algorithm to neural network model
Get the neural network model of isolated island region load forecast.
Load forecast module 3 is carried out for that will extract the input neural network model of the historical data after core principle component
Load forecast.
A kind of isolated island region Electric Load Prediction System provided by the invention is first acquired initial data and locates in advance
Reason, obtains the correlated characteristic of initial data.Then according to prediction need to the neural network topology structure in prediction model into
Row determines.On this basis, neural network and prediction model are trained using initial data, when frequency of training and prediction
When precision reaches required target, the load data in certain following a period of time predicted.So as to reality
The calculating and prediction of existing isolated island type region electric load, improve the accuracy of load forecast, can for generation schedule optimization and
Adjustment, reasonable employment electric power resource provide effective Technical Reference.
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all in spirit of the invention and
Within principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of isolated island region Methods of electric load forecasting, which is characterized in that the described method includes:
Step 1, obtain the historical data for needing the electric power in isolated island region predicted, to the historical data extract core it is main at
Analysis；
Step 2, it is optimized using parameter of the pollen algorithm to the neural network model, training obtains the isolated island region electricity
The neural network model of power load prediction；
Step 3, the historical data after extraction core principle component is inputted into the neural network model and carries out load forecast.
2. the method according to claim 1, wherein obtaining the isolated island region electricity for needing to predict in the step 1
The historical data of power includes:
Maximum temperature, mean temperature, minimum temperature, season type, month, the precipitation situation, day in the isolated island region for needing to predict
Type, wind speed, humidity, on the day of prediction day before the load value at 3 moment and the load value of prediction 5 days synchronizations a few days ago.
3. the method according to claim 1, wherein extracting core to the historical data in the step 1
Further include that the historical data is normalized before principal component analysis:
Wherein, xiFor the current value of i-th of load data in the historical data, xminAnd xmaxIn the respectively described historical data
Load data in minimum value and maximum value.
4. according to the method described in claim 3, it is characterized in that, the historical data is normalized in the step 1
After processing, the process for extracting core principle component analysis to the historical data includes:
Step 101, by n index of m sample in the historical data, write as the data matrix A of m × n dimension；
Step 102, kernel function and relevant parameter are determined, nuclear matrix K is calculateduv:
Kuv≡(Φ(xu)·Φ(xv))；
Step 103, the nuclear matrix K is correcteduv:
M is the size of sample space；
Step 104, the nuclear matrix K is calculateduvEigenvalue λ1,λ2,…,λnAnd feature vector v1,v2,…,vn；
Step 105, to the eigenvalue λ1,λ2,…,λnWith described eigenvector v1,v2,…,vnRespectively in the way of descending
Sequence obtains λ '1> λ '2> ... > λ 'nWith v '1,v'2,…,v'n；
Step 106, described eigenvector unit orthogonalization, α is obtained1,α2,…,αn；
Step 107, the eigenvalue λ is calculated1,λ2,…,λnAccumulation contribution rate B1,B2,…,Bn, according to the recovery rate p of setting,
Work as Bt>=p then chooses t and principal component α1,α2,…,αt；
Step 108, sample X is calculated in α1,α2,…,αtOn projection Y, Y=Kuv·α。
5. the method according to claim 1, wherein the isolated island region Electric Load Forecasting created in the step 1
The neural network model of survey is BP neural network model；
Extract main component obtain the parameter of the BP neural network model include: the neural network number of plies, it is each node layer number, hidden
Layer, the transmission function of output layer and training function.
6. according to the method described in claim 5, it is characterized in that, using pollen algorithm to the nerve net in the step 2
The process that the parameter of network model optimizes includes:
Step 201, by the parameter of the BP neural network model, parameter is encoded as a whole, determines FPA algorithm
Transition probability p and maximum number of iterations T is arranged in search space dimension D；
Step 202, the fitness function of the pollen algorithm is determined:
For i-th of sample reality output；yiFor i-th of sample desired output；I=1,2,3 ..., n, n are sample size；
Step 203, the position of random initializtion individual calculates the fitness function value of each individual according to the fitness function,
Retain the smallest individual of fitness value；
Step 204, transition probability P:p=0.8+0.2 × rand is calculated, rand ∈ [0,1] is randomly generated, judges transition probability P
When > rand, global search is carried out；Otherwise local pollination is carried out；
Step 205, the fitness function value that each pollen is calculated according to the fitness function finds out current optimal solution；
Step 206, terminate when judgement meets the condition that pollen algorithm terminates, pollen individual is decoded as the BP neural network mould
The weight of type and the initial value of threshold value.
7. according to the method described in claim 6, it is characterized in that, code length L in the step 201 are as follows:
L=n × m+m × s+m+s；
N is input layer number, and m is the number of hidden nodes, and s is output layer number of nodes.
8. according to the method described in claim 6, it is characterized in that, carrying out the formula of global search in the step 204 are as follows:The formula locally pollinated are as follows:
Wherein, L is the random real number for obeying the L é vy distribution of index 1.5；g*It is current globally optimal solution；γ is taken between [0,1]
From equally distributed random real number；J, k is a random number between [1, n], is not mutually equal two-by-two between i, j, k,For i-th of sample
This is in the location of t moment.
9. according to the method described in claim 6, it is characterized in that, after the step 206 further include:
According to the initial value of the weight of the BP neural network model and threshold value, the BP neural network model is trained；
When judgement meets the condition that the BP neural network model terminates, when judgement meets the training condition of the BP neural network model
The BP neural network model is exported, it is no to then follow the steps 206.
10. a kind of isolated island region Electric Load Prediction System, which is characterized in that the system comprises: historical data acquisition module,
Neural network model training module and load forecast module；
The historical data obtains module, for obtaining the historical data for needing the electric power in the isolated island region predicted, goes through to described
History data extract core principle component analysis；
The neural network model training module, it is excellent for being carried out using parameter of the pollen algorithm to the neural network model
Change, training obtains the neural network model of isolated island region load forecast；
The load forecast module inputs the neural network mould for that will extract the historical data after core principle component
Type carries out load forecast.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910616995.7A CN110348630A (en) | 2019-07-09 | 2019-07-09 | A kind of isolated island region Methods of electric load forecasting and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910616995.7A CN110348630A (en) | 2019-07-09 | 2019-07-09 | A kind of isolated island region Methods of electric load forecasting and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110348630A true CN110348630A (en) | 2019-10-18 |
Family
ID=68178011
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910616995.7A Pending CN110348630A (en) | 2019-07-09 | 2019-07-09 | A kind of isolated island region Methods of electric load forecasting and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110348630A (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105303262A (en) * | 2015-11-12 | 2016-02-03 | 河海大学 | Short period load prediction method based on kernel principle component analysis and random forest |
| CN108320046A (en) * | 2017-12-27 | 2018-07-24 | 安徽机电职业技术学院 | Short-term electric load prediction modeling method |
| CN109426889A (en) * | 2017-09-01 | 2019-03-05 | 南京理工大学 | Short-term load forecasting method based on KPCA in conjunction with improvement neural network |
| US20190081476A1 (en) * | 2017-09-12 | 2019-03-14 | Sas Institute Inc. | Electric power grid supply and load prediction |
| CN109726849A (en) * | 2018-11-26 | 2019-05-07 | 湘潭大学 | A kind of building microgrid load forecasting method of colored pollination algorithm optimization neural network |
-
2019
- 2019-07-09 CN CN201910616995.7A patent/CN110348630A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105303262A (en) * | 2015-11-12 | 2016-02-03 | 河海大学 | Short period load prediction method based on kernel principle component analysis and random forest |
| CN109426889A (en) * | 2017-09-01 | 2019-03-05 | 南京理工大学 | Short-term load forecasting method based on KPCA in conjunction with improvement neural network |
| US20190081476A1 (en) * | 2017-09-12 | 2019-03-14 | Sas Institute Inc. | Electric power grid supply and load prediction |
| CN108320046A (en) * | 2017-12-27 | 2018-07-24 | 安徽机电职业技术学院 | Short-term electric load prediction modeling method |
| CN109726849A (en) * | 2018-11-26 | 2019-05-07 | 湘潭大学 | A kind of building microgrid load forecasting method of colored pollination algorithm optimization neural network |
Non-Patent Citations (1)
| Title |
|---|
| 孙景文等: "基于核主成分分析和CSM-PSCO优化LSSVR的短期负荷预测", 《广东电力》 * |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111353653B (en) | Photovoltaic output short-term interval prediction method | |
| Yu et al. | Short-term hydro-thermal scheduling using particle swarm optimization method | |
| CN110084424A (en) | A kind of Methods of electric load forecasting based on LSTM and LGBM | |
| CN108365608B (en) | Uncertain optimization scheduling method and system for regional energy Internet | |
| Piltan et al. | Energy demand forecasting in Iranian metal industry using linear and nonlinear models based on evolutionary algorithms | |
| CN104978611A (en) | Neural network photovoltaic power generation output prediction method based on grey correlation analysis | |
| CN108711847A (en) | A kind of short-term wind power forecast method based on coding and decoding shot and long term memory network | |
| CN109255726A (en) | A kind of ultra-short term wind power prediction method of Hybrid Intelligent Technology | |
| CN107909221A (en) | Power-system short-term load forecasting method based on combination neural net | |
| CN110070228A (en) | BP neural network wind speed prediction method for neuron branch evolution | |
| Ge et al. | Short-term load forecasting of regional distribution network based on generalized regression neural network optimized by grey wolf optimization algorithm | |
| CN109242136A (en) | A kind of micro-capacitance sensor wind power Chaos-Genetic-BP neural network prediction technique | |
| CN109242200B (en) | Wind power interval prediction method of Bayesian network prediction model | |
| CN112330487B (en) | Photovoltaic power generation short-term power prediction method | |
| Tian et al. | An adaptive ensemble predictive strategy for multiple scale electrical energy usages forecasting | |
| CN113554466A (en) | Short-term power consumption prediction model construction method, prediction method and device | |
| CN111415049A (en) | Power failure sensitivity analysis method based on neural network and clustering | |
| CN109858665A (en) | Photovoltaic short term power prediction technique based on Feature Selection and ANFIS-PSO | |
| CN111008790A (en) | Hydropower station group power generation electric scheduling rule extraction method | |
| Xiao et al. | Online sequential extreme learning machine algorithm for better predispatch electricity price forecasting grids | |
| Zhang et al. | Scheduling electric power production at a wind farm | |
| Pandu et al. | Artificial Intelligence Based Solar Radiation Predictive Model Using Weather Forecasts. | |
| CN107330538A (en) | A kind of method of climate lower storage reservoir adaptability scheduling rule establishment | |
| CN110348630A (en) | A kind of isolated island region Methods of electric load forecasting and system | |
| CN116485582A (en) | Heat supply optimization regulation and control method and device based on deep learning |
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: 20191018 |
|
| RJ01 | Rejection of invention patent application after publication |