CN108960491A - Method for forecasting photovoltaic power generation quantity based on RBF neural - Google Patents
Method for forecasting photovoltaic power generation quantity based on RBF neural Download PDFInfo
- Publication number
- CN108960491A CN108960491A CN201810620653.8A CN201810620653A CN108960491A CN 108960491 A CN108960491 A CN 108960491A CN 201810620653 A CN201810620653 A CN 201810620653A CN 108960491 A CN108960491 A CN 108960491A
- Authority
- CN
- China
- Prior art keywords
- power generation
- rbf neural
- photovoltaic power
- individual
- value
- 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
- 238000010248 power generation Methods 0.000 title claims abstract description 82
- 230000001537 neural effect Effects 0.000 title claims abstract description 81
- 238000000034 method Methods 0.000 title claims abstract description 37
- 230000003044 adaptive effect Effects 0.000 claims abstract description 10
- 210000000349 chromosome Anatomy 0.000 claims description 26
- 238000010606 normalization Methods 0.000 claims description 18
- 230000006978 adaptation Effects 0.000 claims description 12
- 210000002569 neuron Anatomy 0.000 claims description 12
- 230000035772 mutation Effects 0.000 claims description 9
- 238000013528 artificial neural network Methods 0.000 claims description 8
- 230000004927 fusion Effects 0.000 claims description 6
- 230000005855 radiation Effects 0.000 claims description 4
- 238000007499 fusion processing Methods 0.000 claims description 3
- 238000003064 k means clustering Methods 0.000 claims description 3
- 241001269238 Data Species 0.000 claims description 2
- 210000004218 nerve net Anatomy 0.000 claims description 2
- 210000004027 cell Anatomy 0.000 claims 1
- 108090000623 proteins and genes Proteins 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000000926 separation method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/048—Activation functions
-
- 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
- G06N3/086—Learning methods using evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Health & Medical Sciences (AREA)
- Marketing (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Molecular Biology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Computational Linguistics (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Water Supply & Treatment (AREA)
- Public Health (AREA)
- Primary Health Care (AREA)
- Physiology (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Development Economics (AREA)
- Game Theory and Decision Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of method for forecasting photovoltaic power generation quantities based on RBF neural, comprising steps of constructing training sample according to photovoltaic power generation quantity and its quasi- historical data for choosing influence factor；Based on the training sample constructed, photovoltaic power generation quantity influence factor is chosen using improved adaptive GA-IAGA, and be trained to RBF neural, obtain photovoltaic power generation quantity influence factor and trained RBF neural；The day data to be predicted of photovoltaic power generation quantity influence factor are inputted into trained RBF neural, obtain photovoltaic power generation quantity predicted value.The present invention can preferably solve the evolvement problem of RBF neural, improve the accuracy of photovoltaic power generation prediction result.
Description
Technical field
The present invention relates to technical field of photovoltaic power generation, more particularly to one kind is based on the extensive error model (Localized in part
Generalization Error Model, is abbreviated as L-GEM) radial basis function (Radial Basis Function, letter
It is written as RBF) method for forecasting photovoltaic power generation quantity of neural network.
Background technique
Large-scale photovoltaic power generation is a kind of effective means using solar energy, but solar radiation, atmospheric temperature, weather pattern
It is easy to have an impact photovoltaic power generation with factors such as battery plate temperatures, and in non-linear.Therefore, the prediction pair of photovoltaic power generation quantity
Reasonable arrangement electric appliance usage time and to greatest extent using solar energy resources, reduce electric cost have great significance.And light
The accurate prediction for lying prostrate generated energy depends on the reasonable selection of photovoltaic power generation quantity influence factor.
Radial basis function (Radial Basis Function, RBF) neural network is a kind of feed-forward type mind of function admirable
Through network, arbitrary nonlinear function can be approached with arbitrary accuracy, and there is global approximation capability, and topological structure is compact,
Structural parameters can realize separation study, fast convergence rate.But photovoltaic power generation quantity prediction is carried out using RBF neural, often
There is the error very little that trained network generates the data in training set, but is not fine to the Data Representation in test set
The case where, i.e., the problem of the generalization ability difference of so-called neural network；Moreover, in RBF neural application, the ginseng of neural network
Number chooses whether that rationally, the prediction of neural network can be seriously affected.
Summary of the invention
It is an object of the invention to overcome deficiency in the prior art, a kind of photovoltaic hair based on RBF neural is provided
Power predicating method solves the relatively low technical problem of photovoltaic power generation quantity prediction accuracy.
In order to solve the above technical problems, the technical scheme adopted by the invention is that: the photovoltaic power generation based on RBF neural
Prediction technique is measured, is included the following steps:
Training sample is constructed according to photovoltaic power generation quantity and its quasi- historical data for choosing influence factor；
Based on the training sample constructed, photovoltaic power generation quantity influence factor is chosen using improved adaptive GA-IAGA, and to RBF mind
It is trained through network, obtains photovoltaic power generation quantity influence factor and trained RBF neural；
The day data to be predicted of photovoltaic power generation quantity influence factor are inputted into trained RBF neural, obtain photovoltaic
Generated energy predicted value.
Further, the quasi- selection influence factor includes: intensity of solar radiation, maximum temperature, minimum temperature and photovoltaic
Battery plate temperature, wind speed, relative humidity, weather pattern.
Further, it constructs training sample the specific method is as follows:
The historical data of the quasi- selection influence factor of selection photovoltaic power generation quantity and its corresponding actual power generation；
It, will be corresponding using the historical data of the quasi- selection influence factor of photovoltaic power generation quantity as the input vector of training sample
Photovoltaic actual power generation is normalized as output vector, and to input vector and output vector.
Further, input vector and output vector are normalized that the specific method is as follows:
Weather pattern is divided into fine, cloudy, negative, light rain or snow, moderate rain or snow, heavy rain or snow, corresponds to normalized value
1,0.8,0.7,0.5,0.4,0.3 is taken respectively, and weather pattern is changed, takes the flat of variation front and back weather pattern normalized value
Mean value；
In addition to weather pattern, other quasi- historical datas for choosing influence factor are public using following normalization as input vector
Formula processing:
Actual power generation uses following normalization formula manipulation as output vector:
Wherein, niFor input layer number；xiFor i-th of component in history input vector before normalized；Y is normalizing
Change history output data, x before handlingi,min, xi,maxI-th of component be most in history input vector respectively before normalized
Small value and maximum value, ymin, ymaxMinimum value and maximum value respectively before normalized in history output data,For normalizing
Change i-th of component in treated history input vector,For the history output data after normalized.
Further, for the training sample constructed, photovoltaic power generation quantity influence factor is chosen using improved adaptive GA-IAGA,
And RBF neural is trained, obtain photovoltaic power generation quantity influence factor and trained RBF neural；Specific method
It is as follows:
A, initialize: setting initial population size enables each chromosome by being randomly assigned 0 and 1 creation initial population
Size is equal to the number of the quasi- selection influence factor of photovoltaic power generation quantity；Arrange to form character subset, RBF nerve with 1 feature
Network is based on this feature subset and is trained；Maximum number of iterations is set, and juxtaposition primary iteration number is 1；
B, the fitness of each chromosome is calculated, and calculates optimal adaptation degree；
C, the number of iterations adds 1, if the number of iterations is greater than maximum number of iterations, then goes to step H；
D, duplication operation is carried out by duplication probability；
E, crossover operation is carried out to chromosome in population by crossover probability；
F, mutation operation is carried out to chromosome by mutation probability；
G, circulation step B~F；
H, optimal adaptation degree and its corresponding optimal chromosome are exported, wherein optimal chromosome represents optimal photovoltaic power generation
Amount influence factor, optimal adaptation degree represent the error bounds of the extensive error model in optimal part.
Further, the formula of each chromosome fitness is calculated are as follows:
Wherein, fitnessiFor the fitness of i-th of chromosome,It is to be calculated using the extensive error model in part
The error bounds of i-th of the RBF neural arrived；SQRepresent the union of the Q neighborhood of whole sample points；
RBF neural is single outputting radial basis function neural network, and expression-form is as follows:
Wherein, f (x) is the true output of RBF neural, and M represents the number of hidden nodes, by formula
It acquires；N is the characteristic in training sample input vector；A is the constant between 1 to 10；wjJ-th of hidden neuron is represented to connect
It is connected to the weight of output layer neuron；ujFor the center vector value of j-th of hidden neuron；vjFor the height of j-th of hidden neuron
The width of this basic function；X is the input of RBF neural；
It is acquired using following formula:
Wherein:
Wherein, F (x) is the desired output of RBF neural；L is number of training；WithRespectively represent i-th of spy
The expectation and variance of sign；C is a normal number；Q is a given positive real number；ED[] is mathematic expectaion, ujiFor ujI-th point
Amount, RempFor the mean square deviation of RBF neural output；γj、ξj、sjIt is intermediate variable, no physical meaning.
Further, the center of the RBF neural is acquired by K-means clustering algorithm, and the specific method is as follows:
1-a, arbitrarily select kk data object as initial cluster center from nn data object；
1-b, each data object is calculated at a distance from cluster centre, and respective data object is carried out according to minimum range
It divides；
1-c, each mean value for changing cluster is recalculated, using the mean value acquired as cluster centre；
1-d, step 1-b, 1-c is repeated, until the variation of all mean values is less than given threshold value；
In the case where the center of RBF neural has determined, the width of RBF neural is sought using following methods
Degree, the specific method is as follows:
2-a, the distance for finding out all cluster centres pairm1=1,2 ..., M, n1=m1+1,m1+
2 ..., M, and m1≠n1；Wherein,Represent m1A cluster centre；Represent n-th1A cluster centre；M expression needs to solve
Weight number；
2-b, to the distance-taxis of gained cluster centre pair, each RBF neural width vjIdentical value v is taken, and just
Than in the average value of preceding p minimum range.
Further, the weight of RBF neural is acquired using improvement brainstorming algorithm, the specific method is as follows:
3-a, n is randomly generated2Individual, each individual have M dimension, represent a RBF neural weight per one-dimensional；
3-b, with k-means by this n2Individual is divided into m2Class；
3-c, this n is assessed2Individual；
3-d, the individual in every one kind is ranked up, is selected in cluster of the individual optimal in every one kind as such
The heart；
3-e, be randomly generated one 0 to 1 between numerical value r1；
(3-e1) is if r1< probability parameter p1,
I. a cluster centre is randomly choosed；
Ii. an individual is randomly generated instead of the cluster centre；
3-f, the update for carrying out individual:
(3-f1) be randomly generated one 0 to 1 between numerical value r2；
(3-f2) is if r2Less than probability parameter p2；
I. a cluster is randomly choosed by probability；
Ii. the numerical value r between being randomly generated one 0 to 13；
Iii. if r3Less than probability parameter p3, cluster centre is selected to generate new individual plus random perturbation；
Iv. otherwise, the numerical value r between being randomly generated one 0 to 14；
V. if r4Less than probability parameter p4, from an individual is randomly choosed in such, in addition random perturbation generates new
Body；
Vi. otherwise, 2 individuals are randomly choosed from such, are merged into each other and are generated a new individual plus random perturbation；
(3-f3) otherwise, randomly chooses two classes and generates new individual
I. the numerical value r between being randomly generated one 0 to 15；
Ii. if r5Less than probability parameter p5, the cluster centre of two classes is merged into each other and generates one plus random perturbation
New individual；
Iii. otherwise, the numerical value r between being randomly generated one 0 to 16；
Iv. if r6Less than probability parameter p6, by the cluster centre of first class and randomly selected one from second class
Individual, which is merged into each other, generates a new individual plus random perturbation；
V. otherwise, select an individual to merge into each other from two classes respectively and generate a new individual plus random perturbation；
(3-f4) newly generated individual is compared with current individual, new of the good individual of adaptive value as next iteration
Body；
If 3-g, having generated n new individuals, 3-h is gone to step, 3-f is otherwise gone to step；
3-h, stop if reaching maximum number of iterations, otherwise go to step 3-b；
3-i, output optimum individual and optimal adaptation degree；
It is related to 2 individual fusions, fusion process carries out as the following formula:
Inew=α1I1+(1-α1)I2 (12)
In formula: InewThe offspring individual generated for 2 individual fusions；I1And I2Expression receives 2 individuals of mixing operation；α1
For the random number between one 0 to 1, for adjusting 2 individual weights；
Random perturbation is added during new individual generates, mode is as follows:
In formula:To choose value of the individual in d dimension；Newly to generate value of the individual in d dimension；N (μ, σ) is
Gaussian random function, mean value μ, variance σ；ξ1For step-length, for adjusting the value range of random perturbation；
ξ1Value is calculated as follows to obtain:
ξ1=logsig ((0.5*max_iteration-current_iteration)/k1)*rand() (14)
In formula: logsig () is a logarithm S-transformation function, and max_iteration is maximum number of iterations, current_
Iteration is current iteration number, k1For the slope gradient for controlling logsig (), rand () is generated between one 0 to 1
Random number.
9. the method for forecasting photovoltaic power generation quantity according to claim 1 based on RBF neural, which is characterized in that
The day data to be predicted of photovoltaic power generation quantity influence factor should carry out normalizing before being input to RBF neural as input vector
Change processing predicts that output data obtains the photovoltaic power generation quantity predicted value of day to be predicted, anti-normalization processing through anti-normalization processing
Formula are as follows:
Wherein:For the photovoltaic power generation quantity prediction data before the anti-normalization processing predicted through RBF neural, y*
For the photovoltaic power generation quantity predicted value after anti-normalization processing, ymin、ymaxRespectively before normalized in history output data
Minimum value and maximum value.
Compared with prior art, the beneficial effects obtained by the present invention are as follows being:
By the quasi- selection influence factor of photovoltaic power generation quantity, photovoltaic power generation is realized using the RBF neural based on L-GEM
Amount prediction, preferably solves the evolvement problem of RBF neural；Improve the accuracy of system prediction result；
It is inverted the new mutation operation that variation combines with sequence based on single-point variation, realizes that global parameter is optimal；
The weight of RBF neural is acquired using brainstorming algorithm is improved, and is acquired using the extensive error model in part
The error bounds of RBF neural improve the anti-precocious ability of algorithm, and improve the solving precision of algorithm.
Detailed description of the invention
Fig. 1 is flow chart of the invention；
Fig. 2 is the method flow diagram using improved adaptive GA-IAGA selection photovoltaic power generation quantity influence factor；
Fig. 3 is using the method flow diagram for improving brainstorming algorithm calculating RBF neural weight.
Specific embodiment
The invention will be further described below in conjunction with the accompanying drawings.Following embodiment is only used for clearly illustrating the present invention
Technical solution, and not intended to limit the protection scope of the present invention.
As shown in Figures 1 to 3, the present invention the following steps are included:
Step (1a): the historical data and the photovoltaic power generation corresponding to it for acquiring the quasi- selection influence factor of photovoltaic power generation quantity
Historical data is measured, training sample set is obtained；
The quasi- selection influence factor of the photovoltaic power generation quantity includes: the intensity of solar radiation of day to be predicted, maximum temperature, most
Low temperature and battery plate temperature, wind speed, relative humidity, weather pattern；Photovoltaic power generation quantity historical data corresponding to it refers to: quasi- choosing
Take the actual power generation of day to be predicted corresponding to the historical data of influence factor；
Step (1b): input is generated according to the historical data of the quasi- selection influence factor of photovoltaic power generation quantity obtained by step (1a)
Vector using the historical data of the photovoltaic power generation quantity corresponding to it as output vector, and carries out input vector and output vector
Normalized obtains training sample；The specific steps are that:
(1b-i) utilizes the historical data of the quasi- selection influence factor of gained photovoltaic power generation quantity to generate input vector, with its institute
The historical data of corresponding photovoltaic power generation quantity is as output vector；
Input vector obtained by step (1b-i) and output data is normalized in (1b-ii) respectively, the steps include:
(iia) weather pattern is divided into fine, cloudy, negative, light rain (or snow), moderate rain (or snow), heavy rain (or snow), corresponds to
Normalized value takes 1,0.8,0.7,0.5,0.4,0.3 respectively, and weather pattern is changed, takes variation front and back weather pattern normalizing
The average value of change value；
(iib) in addition to weather pattern, other influences factor uses following normalization formula manipulation as input vector:
(iic) output vector uses following normalization formula manipulation:
Wherein, niFor input layer number；xiFor i-th of component in history input vector before normalized；Y is normalizing
Change history output data, x before handlingi,min, xi,maxI-th of component be most in history input vector respectively before normalized
Small value and maximum value, ymin, ymaxMinimum value and maximum value respectively before normalized in history output data,For normalizing
Change i-th of component in treated history input vector,For the history output data after normalized.
(1b-iii) saves the minimum value x of each component in history input vector before normalizedi,minAnd maximum value
xi,max, minimum value y in history output dataminWith maximum value ymax。
Step (1c): using improved adaptive GA-IAGA on training sample selection photovoltaic power generation quantity obtained by step (1b) influence because
Element, the RBF neural after obtaining photovoltaic power generation quantity influence factor and training, the specific steps are that:
(2a) initialization: setting Population Size generates initial population by Population Size, and initial population is by being randomly assigned
0 and 1 creates.Each chromosome is a binary digit string, and size is equal to the feature in training sample input vector
Number (i.e. the number of the quasi- selection influence factor of photovoltaic power generation quantity).Feature with 1 forms character subset, RBF neural base
It is trained in this feature subset；Maximum number of iterations is set, and juxtaposition the number of iterations is 1；
(2b) calculates the fitness of each chromosome, and calculates optimal adaptation degree；
(2c) the number of iterations adds 1, if the number of iterations is greater than maximum number of iterations, then turns (2h)；
(2d) carries out duplication operation by duplication probability；
(2e) carries out crossover operation to chromosome in population by crossover probability.It, which is comprised the concrete steps that, is randomly generated one 0 to 1
Between numerical value, if the value be less than crossover probability, then from population randomly choose two chromosomes, carry out crossover operation.Intersect behaviour
Make to use two-point crossover, two crosspoints are randomly generated, the gene in the friendship of two chromosomes in two crunodes is exchanged.
(2f) carries out mutation operation to chromosome by mutation probability.It comprises the concrete steps that the number being randomly generated between one 0 to 1
Value then carries out mutation operation to current chromosome if the value is less than mutation probability.Its method is randomly generated between one 0 to 1
Numerical value, if the value be less than variation mode select probability, then using single-point make a variation, random selection one variation position, to the gene position into
Row inversion operation；Otherwise variation is then inverted using sequence, two change points is randomly generated, the gene order backward in two change points
Arrangement.
(2g) turns (2b)；
(2h) exports optimal chromosome and optimal adaptation degree.Wherein optimal chromosome represents optimal photovoltaic power generation quantity and influences
Factor, optimal adaptation degree represent the error bounds of the extensive error model in optimal part.
Step (1d): day photovoltaic power generation quantity influence factor data to be predicted are acquired and generate prediction input vector, carry out normalizing
Change processing, the prediction input vector after obtaining normalized；Normalized uses step (1b-ii) same procedure；
Step (1e): the prediction input vector after step (1d) described normalized is inputted into corresponding RBF nerve net
Network, obtains photovoltaic power generation quantity prediction output data, and the photovoltaic that prediction output data obtains day to be predicted through anti-normalization processing is sent out
Power quantity predicting value.
The formula of anti-normalization processing are as follows:
For the photovoltaic power generation quantity prediction data before the anti-normalization processing predicted through RBF neural, y*It is anti-
Photovoltaic power generation quantity predicted value after normalized, ymin、ymaxMinimum respectively before normalized in history output data
Value and maximum value.
Step (2b) calculates the formula of the fitness of each chromosome are as follows:
Wherein,It is i-th of RBF neural being calculated using the extensive error model (L_GEM) in part
Error bounds.RBF neural is single outputting radial basis function neural network, can be expressed as following form:
Wherein, f (x) is the true output of RBF neural, and M represents the number of hidden nodes, by formula
It acquires；N is the characteristic in training sample input vector；A is the constant between 1 to 10；wjJ-th of hidden neuron is represented to connect
It is connected to the weight of output layer neuron；ujFor the center vector value of j-th of hidden neuron；vjFor the height of j-th of hidden neuron
The width of this basic function；X is the input of RBF neural；
It can be acquired by following formula:
Wherein:
Wherein, F (x) is the desired output of fallout predictor；L is number of training；WithRespectively represent ith feature
It is expected that and variance；C is a normal number；Q is a given positive real number；ED[] is mathematic expectaion, ujiFor ujI-th of component, Remp
For the mean square deviation of RBF neural output；γj、ξj、sjIt is intermediate variable, no physical meaning.
Wherein, the center of RBF neural can be acquired by K-means clustering algorithm, the specific steps are that:
(3a) arbitrarily selects kk data object as initial cluster center from nn data object；
(3b) calculates each data object at a distance from these cluster centres, and is carried out according to minimum range to corresponding object
It divides；
(3c) recalculates the mean value of each (changing) cluster, and the mean value acquired is as cluster centre；
(3d) repeats step (3b), step (3c), until the variation of all mean values is less than given threshold value.
In the case where the center of RBF neural has determined, following methods can be used, and to seek RBF neural wide
Degree, the steps include:
(3e) finds out the distance of all cluster centres pairm1=1,2 ..., M, n1=m1+1,m1+
2 ..., M, and m1≠n1；Wherein,Represent m1A cluster centre；Represent n-th1A cluster centre；M expression needs to solve
Weight number；
The distance-taxis of (3f) to gained cluster centre pair, each RBF neural width vjIdentical value v is taken, and just
Than in the average value of preceding p minimum range.
In the case where the center of RBF neural, width have determined, RBF neural can be acquired by following methods
Weight, and acquire using the extensive error model in part the error bounds of RBF neural, the steps include:
(3h) acquires the weight of RBF neural using brainstorming algorithm is improved, and utilizes the extensive error model in part
The error bounds of RBF neural are acquired, the specific steps are that:
N is randomly generated in (4a)2Individual, each individual have M dimension, represent a RBF neural weight per one-dimensional；
(4b) is with k-means by this n2Individual is divided into m2Class；
(4c) assesses this n2Individual；
Individual in every one kind is ranked up by (4d), is selected in cluster of the individual optimal in every one kind as such
The heart；
(4e) be randomly generated one 0 to 1 between numerical value r1；
If r1< probability parameter p1,
I. a cluster centre is randomly choosed；
Ii. an individual is randomly generated instead of the cluster centre；
(4f) carries out the update of individual
(4f1) be randomly generated one 0 to 1 between numerical value r2；
(4f2) is if r2Less than probability parameter p2；
I. a cluster is randomly choosed by probability；
Ii. the numerical value r between being randomly generated one 0 to 13；
Iii. if r3Less than probability parameter p3, cluster centre is selected to generate new individual plus random perturbation.
Iv. otherwise, the numerical value r between being randomly generated one 0 to 14；
V. if r4Less than probability parameter p4, from an individual is randomly choosed in such, in addition random perturbation generates new
Body；
Vi. otherwise, 2 individuals are randomly choosed from such, are merged into each other and are generated a new individual plus random perturbation.
(4f3) otherwise, randomly chooses two classes and generates new individual
I. the numerical value r between being randomly generated one 0 to 15；
Ii. if r5Less than probability parameter p5, the cluster centre of two classes is merged into each other and generates one plus random perturbation
New individual；
Iii. otherwise, the numerical value r between being randomly generated one 0 to 16；
Iv. if r6Less than probability parameter p6, by the cluster centre of first class and randomly selected one from second class
Individual, which is merged into each other, generates a new individual plus random perturbation；
V. otherwise, select an individual to merge into each other from two classes respectively and generate a new individual plus random perturbation.
(4f4) newly generated individual is compared with current individual, new individual of the good individual of adaptive value as next iteration.
(4g) goes to step (4h), otherwise goes to step (4f) if having generated n new individuals；
(4h) stops if reaching maximum number of iterations, otherwise goes to step (4b)；
(4i) exports optimum individual and optimal adaptation degree.
Wherein, step (4c), which assesses each individual, is realized according to the fitness in formula (4).
It is related to 2 individual fusions during algorithm is realized, fusion process carries out as the following formula:
Inew=α I1+(1-α)I2 (13)
In formula: InewThe offspring individual generated for 2 individual fusions；I1And I2Expression receives 2 individuals of mixing operation；α
For the random number between one 0 to 1,2 individual weights are adjusted.Random perturbation, side are added during new individual generates
Formula is as follows:
In formula:To choose value of the individual in d dimension；Newly to generate value of the individual in d dimension；N (μ, σ) is
Gaussian random function, mean value μ, variance σ；ξ1For step-length, for adjusting the value range of random perturbation；ξ1Value is as the following formula
It is calculated:
ξ1=logsig ((0.5*max_iteration-current_iteration)/k1)*rand() (15)
In formula: logsig () is a logarithm S-transformation function, and max_iteration is maximum number of iterations, current_
Iteration is current iteration number, k1For the slope gradient for controlling logsig (), rand () is generated between one 0 to 1
Random number.
The above is only a preferred embodiment of the present invention, it is noted that for the ordinary skill people of the art
For member, without departing from the technical principles of the invention, several improvement and deformations can also be made, these improvement and deformations
Also it should be regarded as protection scope of the present invention.
Claims (9)
1. the method for forecasting photovoltaic power generation quantity based on RBF neural, which comprises the steps of:
Training sample is constructed according to photovoltaic power generation quantity and its quasi- historical data for choosing influence factor；
Based on the training sample constructed, photovoltaic power generation quantity influence factor is chosen using improved adaptive GA-IAGA, and to RBF nerve net
Network is trained, and obtains photovoltaic power generation quantity influence factor and trained RBF neural；
The day data to be predicted of photovoltaic power generation quantity influence factor are inputted into trained RBF neural, obtain photovoltaic power generation
Measure predicted value.
2. the method for forecasting photovoltaic power generation quantity according to claim 1 based on RBF neural, which is characterized in that described
It is quasi- choose influence factor include: intensity of solar radiation, maximum temperature, minimum temperature and photovoltaic cell plate temperature, it is wind speed, relatively wet
Degree, weather pattern.
3. the method for forecasting photovoltaic power generation quantity according to claim 2 based on RBF neural, which is characterized in that building
The specific method is as follows for training sample:
The historical data of the quasi- selection influence factor of selection photovoltaic power generation quantity and its corresponding actual power generation；
Using the historical data of the quasi- selection influence factor of photovoltaic power generation quantity as the input vector of training sample, by corresponding photovoltaic
Actual power generation is normalized as output vector, and to input vector and output vector.
4. the method for forecasting photovoltaic power generation quantity according to claim 3 based on RBF neural, which is characterized in that defeated
Incoming vector and output vector are normalized that the specific method is as follows:
Weather pattern is divided into fine, cloudy, negative, light rain or snow, moderate rain or snow, heavy rain or snow, corresponds to normalized value difference
1,0.8,0.7,0.5,0.4,0.3 is taken, weather pattern is changed, takes being averaged for variation front and back weather pattern normalized value
Value；
In addition to weather pattern, other quasi- historical datas for choosing influence factor are as input vector using at following normalization formula
Reason:
Actual power generation uses following normalization formula manipulation as output vector:
Wherein, niFor input layer number；xiFor i-th of component in history input vector before normalized；Y is at normalization
History output data, x before managingi,min, xi,maxRespectively before normalized in history input vector i-th of component minimum value
And maximum value, ymin, ymaxMinimum value and maximum value respectively before normalized in history output data,At normalization
I-th of component in history input vector after reason,For the history output data after normalized.
5. the method for forecasting photovoltaic power generation quantity according to claim 1 based on RBF neural, which is characterized in that be directed to
The training sample constructed is chosen photovoltaic power generation quantity influence factor using improved adaptive GA-IAGA, and is instructed to RBF neural
Practice, obtains photovoltaic power generation quantity influence factor and trained RBF neural；The specific method is as follows:
A, initialize: setting initial population size enables the size of each chromosome by being randomly assigned 0 and 1 creation initial population
Equal to the number of the quasi- selection influence factor of photovoltaic power generation quantity；Arrange to form character subset, RBF neural with 1 feature
It is trained based on this feature subset；Maximum number of iterations is set, and juxtaposition primary iteration number is 1；
B, the fitness of each chromosome is calculated, and calculates optimal adaptation degree；
C, the number of iterations adds 1, if the number of iterations is greater than maximum number of iterations, then goes to step H；
D, duplication operation is carried out by duplication probability；
E, crossover operation is carried out to chromosome in population by crossover probability；
F, mutation operation is carried out to chromosome by mutation probability；
G, circulation step B~F；
H, optimal adaptation degree and its corresponding optimal chromosome are exported, wherein optimal chromosome represents optimal photovoltaic power generation quantity shadow
The factor of sound, optimal adaptation degree represent the error bounds of the extensive error model in optimal part.
6. the method for forecasting photovoltaic power generation quantity according to claim 5 based on RBF neural, which is characterized in that calculate
The formula of each chromosome fitness are as follows:
Wherein, fitnessiFor the fitness of i-th of chromosome,It is to be calculated using the extensive error model in part
The error bounds of i-th of RBF neural；SQRepresent the union of the Q neighborhood of whole sample points；
RBF neural is single outputting radial basis function neural network, and expression-form is as follows:
Wherein, f (x) is the true output of RBF neural, and M represents the number of hidden nodes, by formulaIt acquires；N
For the characteristic in training sample input vector；A is the constant between 1 to 10；wjRepresent j-th of hidden neuron be connected to it is defeated
The weight of layer neuron out；ujFor the center vector value of j-th of hidden neuron；vjFor the gaussian basis letter of j-th of hidden neuron
Several width；X is the input of RBF neural；
It is acquired using following formula:
Wherein:
Wherein, F (x) is the desired output of RBF neural；L is number of training；WithRespectively represent ith feature
It is expected that and variance；C is a normal number；Q is a given positive real number；ED[] is mathematic expectaion, ujiFor ujI-th of component, Remp
For the mean square deviation of RBF neural output；γj、ξj、sjIt is intermediate variable, no physical meaning.
7. the method for forecasting photovoltaic power generation quantity according to claim 6 based on RBF neural, which is characterized in that described
The center of RBF neural is acquired by K-means clustering algorithm, and the specific method is as follows:
1-a, arbitrarily select kk data object as initial cluster center from nn data object；
1-b, each data object of calculating draw respective data object at a distance from cluster centre, and according to minimum range
Point；
1-c, each mean value for changing cluster is recalculated, using the mean value acquired as cluster centre；
1-d, step 1-b, 1-c is repeated, until the variation of all mean values is less than given threshold value；
In the case where the center of RBF neural has determined, the width of RBF neural is sought using following methods, is had
Body method is as follows:
2-a, the distance for finding out all cluster centres pairm1=1,2 ..., M, n1=m1+1,m1+2,...,
M, and m1≠n1；Wherein,Represent m1A cluster centre；Represent n-th1A cluster centre；M indicates the weight for needing to solve
Number；
2-b, to the distance-taxis of gained cluster centre pair, each RBF neural width vjIdentical value v is taken, and before being proportional to
The average value of p minimum range.
8. the method for forecasting photovoltaic power generation quantity according to claim 7 based on RBF neural, which is characterized in that use
The weight that brainstorming algorithm acquires RBF neural is improved, the specific method is as follows:
3-a, n is randomly generated2Individual, each individual have M dimension, represent a RBF neural weight per one-dimensional；
3-b, with k-means by this n2Individual is divided into m2Class；
3-c, this n is assessed2Individual；
3-d, the individual in every one kind is ranked up, selects individual optimal in every one kind as such cluster centre；
3-e, be randomly generated one 0 to 1 between numerical value r1；
(3-e1) is if r1< probability parameter p1,
I. a cluster centre is randomly choosed；
Ii. an individual is randomly generated instead of the cluster centre；
3-f, the update for carrying out individual:
(3-f1) be randomly generated one 0 to 1 between numerical value r2；
(3-f2) is if r2Less than probability parameter p2；
I. a cluster is randomly choosed by probability；
Ii. the numerical value r between being randomly generated one 0 to 13；
Iii. if r3Less than probability parameter p3, cluster centre is selected to generate new individual plus random perturbation；
Iv. otherwise, the numerical value r between being randomly generated one 0 to 14；
V. if r4Less than probability parameter p4, from an individual is randomly choosed in such, in addition random perturbation generates new individual；
Vi. otherwise, 2 individuals are randomly choosed from such, are merged into each other and are generated a new individual plus random perturbation；
(3-f3) otherwise, randomly chooses two classes and generates new individual
I. the numerical value r between being randomly generated one 0 to 15；
Ii. if r5Less than probability parameter p5, the cluster centre of two classes is merged into each other and generates one new plus random perturbation
Body；
Iii. otherwise, the numerical value r between being randomly generated one 0 to 16；
Iv. if r6Less than probability parameter p6, by the cluster centre of first class with from second class it is randomly selected one by one
Body, which is merged into each other, generates a new individual plus random perturbation；
V. otherwise, select an individual to merge into each other from two classes respectively and generate a new individual plus random perturbation；
(3-f4) newly generated individual is compared with current individual, new individual of the good individual of adaptive value as next iteration；
If 3-g, having generated n new individuals, 3-h is gone to step, 3-f is otherwise gone to step；
3-h, stop if reaching maximum number of iterations, otherwise go to step 3-b；
3-i, output optimum individual and optimal adaptation degree；
It is related to 2 individual fusions, fusion process carries out as the following formula:
Inew=α1I1+(1-α1)I2 (12)
In formula: InewThe offspring individual generated for 2 individual fusions；I1And I2Expression receives 2 individuals of mixing operation；α1It is one
Random number between a 0 to 1, for adjusting 2 individual weights；
Random perturbation is added during new individual generates, mode is as follows:
In formula:To choose value of the individual in d dimension；Newly to generate value of the individual in d dimension；N (μ, σ) is Gauss
Random function, mean value μ, variance σ；ξ1For step-length, for adjusting the value range of random perturbation；
ξ1Value is calculated as follows to obtain:
ξ1=logsig ((0.5*max_iteration-current_iteration)/k1)*rand() (14)
In formula: logsig () is a logarithm S-transformation function, and max_iteration is maximum number of iterations, current_
Iteration is current iteration number, k1For the slope gradient for controlling logsig (), rand () is generated between one 0 to 1
Random number.
9. the method for forecasting photovoltaic power generation quantity according to claim 1 based on RBF neural, which is characterized in that photovoltaic
Place should be normalized in the day data to be predicted of generated energy influence factor before being input to RBF neural as input vector
Reason predicts that output data obtains the photovoltaic power generation quantity predicted value of day to be predicted, the public affairs of anti-normalization processing through anti-normalization processing
Formula are as follows:
Wherein:For the photovoltaic power generation quantity prediction data before the anti-normalization processing predicted through RBF neural, y*It is anti-
Photovoltaic power generation quantity predicted value after normalized, ymin、ymaxMinimum respectively before normalized in history output data
Value and maximum value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810620653.8A CN108960491A (en) | 2018-06-15 | 2018-06-15 | Method for forecasting photovoltaic power generation quantity based on RBF neural |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810620653.8A CN108960491A (en) | 2018-06-15 | 2018-06-15 | Method for forecasting photovoltaic power generation quantity based on RBF neural |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN108960491A true CN108960491A (en) | 2018-12-07 |
Family
ID=64488906
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810620653.8A Pending CN108960491A (en) | 2018-06-15 | 2018-06-15 | Method for forecasting photovoltaic power generation quantity based on RBF neural |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108960491A (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109978280A (en) * | 2019-04-19 | 2019-07-05 | 上海交通大学 | A kind of generalization photovoltaic cell operating temperature prediction technique and device |
| CN110009098A (en) * | 2019-04-19 | 2019-07-12 | 上海交通大学 | A kind of photovoltaic cell operating temperature and generated output combined estimation method |
| CN110070145A (en) * | 2019-04-30 | 2019-07-30 | 天津开发区精诺瀚海数据科技有限公司 | LSTM wheel hub single-item energy consumption prediction based on increment cluster |
| CN110070227A (en) * | 2019-04-25 | 2019-07-30 | 中国人民解放军国防科技大学 | Migration neural network power prediction method suitable for grid-connected photovoltaic power generation |
| CN111602156A (en) * | 2019-01-18 | 2020-08-28 | 株式会社秀住房 | Method for precisely predicting power generation amount calculated from expected value, system for precisely predicting power generation amount calculated from expected value, and program for precisely predicting power generation amount calculated from expected value |
| CN112732008A (en) * | 2020-12-29 | 2021-04-30 | 江苏中信博新能源科技股份有限公司 | Method and device for controlling power generation amount and photovoltaic tracking support |
| CN112906860A (en) * | 2021-02-04 | 2021-06-04 | 深圳大学 | Network structure searching method and device |
| CN113469427A (en) * | 2021-06-24 | 2021-10-01 | 国网山东省电力公司东营供电公司 | Convolution LSTM-based photovoltaic power station day-ahead power generation amount prediction method and system |
| CN115912331A (en) * | 2022-10-30 | 2023-04-04 | 淮阴工学院 | Power adjusting method of intelligent wind-solar hybrid power generation system |
| CN116128150A (en) * | 2023-03-02 | 2023-05-16 | 南京邮电大学 | Photovoltaic intelligent output prediction method based on two-stage optimization |
| CN117522080A (en) * | 2024-01-02 | 2024-02-06 | 国网湖北省电力有限公司经济技术研究院 | Linkage optimization method, system and equipment for progress safety and cost of power transmission engineering |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110125684A1 (en) * | 2009-11-24 | 2011-05-26 | Hussain Al-Duwaish | Method for identifying multi-input multi-output hammerstein models |
| CN103927460A (en) * | 2014-05-05 | 2014-07-16 | 重庆大学 | Wind power plant short-term wind speed prediction method based on RBF |
| CN103926526A (en) * | 2014-05-05 | 2014-07-16 | 重庆大学 | Analog circuit fault diagnosis method based on improved RBF neural network |
| CN103974388A (en) * | 2013-01-29 | 2014-08-06 | 中国人民解放军总参谋部第六十一研究所 | Method and device for fusing data of wireless sensor network |
| CN104037757A (en) * | 2014-05-20 | 2014-09-10 | 西安理工大学 | Brainstorming-based thermal power plant economic environment scheduling method |
| CN104616085A (en) * | 2015-02-16 | 2015-05-13 | 河海大学常州校区 | Photovoltaic generating capacity predicting method based on BP neural network |
| CN105913150A (en) * | 2016-04-12 | 2016-08-31 | 河海大学常州校区 | BP neural network photovoltaic power station generating capacity prediction method based on genetic algorithm |
| CN107358014A (en) * | 2016-11-02 | 2017-11-17 | 华南师范大学 | The clinical pre-treating method and system of a kind of physiological data |
| CN107730044A (en) * | 2017-10-20 | 2018-02-23 | 燕山大学 | A kind of hybrid forecasting method of renewable energy power generation and load |
-
2018
- 2018-06-15 CN CN201810620653.8A patent/CN108960491A/en active Pending
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20110125684A1 (en) * | 2009-11-24 | 2011-05-26 | Hussain Al-Duwaish | Method for identifying multi-input multi-output hammerstein models |
| CN103974388A (en) * | 2013-01-29 | 2014-08-06 | 中国人民解放军总参谋部第六十一研究所 | Method and device for fusing data of wireless sensor network |
| CN103927460A (en) * | 2014-05-05 | 2014-07-16 | 重庆大学 | Wind power plant short-term wind speed prediction method based on RBF |
| CN103926526A (en) * | 2014-05-05 | 2014-07-16 | 重庆大学 | Analog circuit fault diagnosis method based on improved RBF neural network |
| CN104037757A (en) * | 2014-05-20 | 2014-09-10 | 西安理工大学 | Brainstorming-based thermal power plant economic environment scheduling method |
| CN104616085A (en) * | 2015-02-16 | 2015-05-13 | 河海大学常州校区 | Photovoltaic generating capacity predicting method based on BP neural network |
| CN105913150A (en) * | 2016-04-12 | 2016-08-31 | 河海大学常州校区 | BP neural network photovoltaic power station generating capacity prediction method based on genetic algorithm |
| CN107358014A (en) * | 2016-11-02 | 2017-11-17 | 华南师范大学 | The clinical pre-treating method and system of a kind of physiological data |
| CN107730044A (en) * | 2017-10-20 | 2018-02-23 | 燕山大学 | A kind of hybrid forecasting method of renewable energy power generation and load |
Non-Patent Citations (4)
| Title |
|---|
| 公茂法: "采煤机滚动轴承故障诊断新方法", 《工矿自动化》 * |
| 周静: "基于局部泛化误差模型的RBFNN的启发式训练方法", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 李旻炫: "基于ART2神经网络和K—means算法的光伏发电功率曲线聚类", 《电气制造》 * |
| 陈宏伟: "头脑风暴算法在多阈值Otsu分割法中的应用", 《湖北工业大学学报》 * |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111602156B (en) * | 2019-01-18 | 2021-07-30 | 株式会社秀住房 | Method for precisely predicting power generation amount calculated from expected value, system for precisely predicting power generation amount calculated from expected value, and program for precisely predicting power generation amount calculated from expected value |
| CN111602156A (en) * | 2019-01-18 | 2020-08-28 | 株式会社秀住房 | Method for precisely predicting power generation amount calculated from expected value, system for precisely predicting power generation amount calculated from expected value, and program for precisely predicting power generation amount calculated from expected value |
| US10998725B2 (en) | 2019-01-18 | 2021-05-04 | Hide Housing Corporation | Electric power generation prediction method based on expected value calculation, electric power generation prediction system based on expected value calculation, and electric power generation prediction program product based on expected value calculation |
| CN110009098A (en) * | 2019-04-19 | 2019-07-12 | 上海交通大学 | A kind of photovoltaic cell operating temperature and generated output combined estimation method |
| CN109978280A (en) * | 2019-04-19 | 2019-07-05 | 上海交通大学 | A kind of generalization photovoltaic cell operating temperature prediction technique and device |
| CN110070227A (en) * | 2019-04-25 | 2019-07-30 | 中国人民解放军国防科技大学 | Migration neural network power prediction method suitable for grid-connected photovoltaic power generation |
| CN110070227B (en) * | 2019-04-25 | 2021-05-04 | 中国人民解放军国防科技大学 | Migration neural network power prediction method suitable for grid-connected photovoltaic power generation |
| CN110070145A (en) * | 2019-04-30 | 2019-07-30 | 天津开发区精诺瀚海数据科技有限公司 | LSTM wheel hub single-item energy consumption prediction based on increment cluster |
| CN112732008A (en) * | 2020-12-29 | 2021-04-30 | 江苏中信博新能源科技股份有限公司 | Method and device for controlling power generation amount and photovoltaic tracking support |
| CN112906860A (en) * | 2021-02-04 | 2021-06-04 | 深圳大学 | Network structure searching method and device |
| CN113469427A (en) * | 2021-06-24 | 2021-10-01 | 国网山东省电力公司东营供电公司 | Convolution LSTM-based photovoltaic power station day-ahead power generation amount prediction method and system |
| CN115912331A (en) * | 2022-10-30 | 2023-04-04 | 淮阴工学院 | Power adjusting method of intelligent wind-solar hybrid power generation system |
| CN115912331B (en) * | 2022-10-30 | 2023-09-29 | 淮阴工学院 | Power adjusting method of intelligent wind-solar complementary power generation system |
| CN116128150A (en) * | 2023-03-02 | 2023-05-16 | 南京邮电大学 | Photovoltaic intelligent output prediction method based on two-stage optimization |
| CN117522080A (en) * | 2024-01-02 | 2024-02-06 | 国网湖北省电力有限公司经济技术研究院 | Linkage optimization method, system and equipment for progress safety and cost of power transmission engineering |
| CN117522080B (en) * | 2024-01-02 | 2024-04-09 | 国网湖北省电力有限公司经济技术研究院 | Linkage optimization method, system and equipment for progress safety and cost of power transmission engineering |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108960491A (en) | Method for forecasting photovoltaic power generation quantity based on RBF neural | |
| WO2021120787A1 (en) | Simulation operation method for large-scale reservoir group in main stream and tributaries of river basin | |
| CN105809279B (en) | A kind of water resources optimal operation method based on multi-target quantum shuffled frog leaping algorithm | |
| CN109886473B (en) | Watershed wind-solar water system multi-objective optimization scheduling method considering downstream ecology | |
| JP5888640B2 (en) | Photovoltaic power generation prediction apparatus, solar power generation prediction method, and solar power generation prediction program | |
| CN104484833A (en) | Photovoltaic power generation output power tracking algorithm based on genetics algorithm improved RBF-BP neural network | |
| CN108694473A (en) | Building energy consumption prediction technique based on RBF neural | |
| CN108205717A (en) | A kind of photovoltaic generation power Multiple Time Scales Forecasting Methodology | |
| CN109492287A (en) | A kind of solid electrolyte ionic conductivity prediction technique based on BP neural network | |
| CN104636985A (en) | Method for predicting radio disturbance of electric transmission line by using improved BP (back propagation) neural network | |
| CN113887787B (en) | Flood forecast model parameter multi-objective optimization method based on long-short-term memory network and NSGA-II algorithm | |
| CN103577679A (en) | Real-time computing method for theoretical line loss of low-voltage distribution room | |
| CN111119282A (en) | Pressure monitoring point optimal arrangement method for water supply pipe network | |
| CN105978732A (en) | Method and system for optimizing parameters of minimum complexity echo state network based on particle swarm | |
| Chitsazan et al. | Wind speed forecasting using an echo state network with nonlinear output functions | |
| CN110263915A (en) | A kind of wind power forecasting method based on deepness belief network | |
| CN112883632B (en) | Lithium battery equivalent circuit model parameter identification method based on improved ant colony algorithm | |
| CN109492768A (en) | Air Quality Forecast model training method and device | |
| Tian et al. | A network traffic hybrid prediction model optimized by improved harmony search algorithm | |
| CN110501020A (en) | A kind of multiple target three-dimensional path planning method | |
| Nhita et al. | Planting calendar forecasting system using evolving neural network | |
| CN110148936B (en) | Coordinated planning method for flexible multi-state switch and distributed power supply in active power distribution network | |
| CN112183721A (en) | Construction method of combined hydrological prediction model based on self-adaptive differential evolution | |
| CN109615142A (en) | A kind of wind farm wind velocity combination forecasting method based on wavelet analysis | |
| CN114492928A (en) | Method, device and equipment for constructing energy comfort degree prediction model for building |
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: 20181207 |
|
| RJ01 | Rejection of invention patent application after publication |