CN106779129A  A kind of ShortTerm Load Forecasting Method for considering meteorologic factor  Google Patents
A kind of ShortTerm Load Forecasting Method for considering meteorologic factor Download PDFInfo
 Publication number
 CN106779129A CN106779129A CN201510799160.1A CN201510799160A CN106779129A CN 106779129 A CN106779129 A CN 106779129A CN 201510799160 A CN201510799160 A CN 201510799160A CN 106779129 A CN106779129 A CN 106779129A
 Authority
 CN
 China
 Prior art keywords
 load
 firefly
 meteorologic factor
 factor
 data
 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
 230000001537 neural Effects 0.000 claims abstract description 14
 241000131894 Lampyris noctiluca Species 0.000 claims abstract description 9
 230000000875 corresponding Effects 0.000 claims abstract description 9
 238000000034 methods Methods 0.000 claims abstract description 8
 230000002159 abnormal effects Effects 0.000 claims abstract description 5
 241000254158 Lampyridae Species 0.000 claims description 61
 230000001264 neutralization Effects 0.000 claims description 40
 238000010219 correlation analysis Methods 0.000 claims description 8
 238000010521 absorption reactions Methods 0.000 claims description 4
 239000003570 air Substances 0.000 claims description 4
 238000010183 spectrum analysis Methods 0.000 claims description 4
 280000781629 Regional Electric companies 0.000 claims description 3
 230000035852 Tmax Effects 0.000 claims description 3
 239000010410 layers Substances 0.000 description 10
 230000000694 effects Effects 0.000 description 2
 238000005457 optimization Methods 0.000 description 2
 241001269238 Data Species 0.000 description 1
 241001081830 Degeneriaceae Species 0.000 description 1
 239000007789 gases Substances 0.000 description 1
 230000002068 genetic Effects 0.000 description 1
 238000009499 grossing Methods 0.000 description 1
 238000001556 precipitation Methods 0.000 description 1
 238000000611 regression analysis Methods 0.000 description 1
 238000001228 spectrum Methods 0.000 description 1
Classifications

 Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSSSECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSSREFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
 Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
 Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
 Y02A30/00—Adapting or protecting infrastructure or their operation

 Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSSSECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSSREFERENCE 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 loadside end user applications
Abstract
The invention discloses a kind of ShortTerm Load Forecasting Method of the consideration meteorologic factor for belonging to Techniques for Prediction of Electric Loads field.Historical load data and meteorological data are collected, abnormal data is detected and correct；The correlation of analysis load data and each meteorologic factor, it is determined that crucial meteorologic factor；Comprehensive meteorologic factor is set up with the correlation of crucial meteorologic factor according to load；The variation characteristic of area power grid daily load curve is summarized, the typical similar day of prediction day is found out；Elman neural network shortterm load forecasting models are set up with comprehensive meteorologic factor using selected load, using glowworm swarm algorithm training network parameter；By the comprehensive meteorologic factor at moment to be predicted and corresponding load data input Elman neural network shortterm load forecasting models, the predicted load at moment to be predicted is exported；Display predicted load.Can Accurate Prediction working day, the load data of weekend and legal festivals and holidays, precision of prediction is high and strong applicability, and reliable basis are provided for power grid operation personnel formulate generation schedule.
Description
Technical field
The invention belongs to Techniques for Prediction of Electric Loads field, more particularly to a kind of ShortTerm Load Forecasting Method for considering meteorologic factor.
Background technology
Shortterm load forecasting is the critical function of energy management system, is the basis of power system security, economy, reliability service.The precision of load prediction directly affects security, economy and the power supply quality of power system.Therefore the emphasis that precision of prediction is current Short Term load Forecasting Technique research how is improved.
Mainly there are traditional prediction method and the modern major class of Forecasting Methodology two currently used for the method for shortterm load forecasting.Traditional prediction method is including exponential smoothing, regression analysis, time series method, grey method etc..Wherein it is most widely used with time series method.Time series method is directed to Onedimension Time Series, is, according to the numerical value change for inferring future load, not account for influence of the meteorologic factor to load with historical load data, causes the missing of important information.Even if some methods consider the influence of meteorologic factor, also mostly it is the relations for analyzing the single meteorologic factors such as daily load and maximum temperature, minimum temperature, all of weather information can not be reflected, easily cause analysis result and error occur, so as to influence the precision of shortterm load forecasting, it is difficult to the need for meeting regional load prediction.Modern Forecasting Methodology mainly has expert system approach, genetic algorithm, neural network, SVMs etc..Neutral net is due to selflearning capability and to Complex Nonlinear System disposal ability, a kind of important method as shortterm load forecasting.Determine according to subjective experience because the structure and parameter of neutral net is more, the effect therefore, it is difficult to ensure prediction.Rationally determining the structure and parameter of neutral net can effectively improve the precision of load prediction.
For the deficiency that existing shortterm load forecasting method is present, it is an object of the present invention to provide a kind of ShortTerm Load Forecasting Method for considering meteorologic factor, the method has considered influence of multiple meteorologic factors to load, the crucial meteorologic factor of selection is used for load prediction, the workload of load prediction is effectively reduced, the accuracy and reliability of load prediction is improved；Using the parameter of firefly optimized algorithm optimization Elman neutral nets, the precision and realtime of load prediction are further improved, reliable basis are provided for power grid operation personnel formulate generation schedule, ensure power grid security, efficient and stable operation.
The content of the invention
It is an object of the invention to propose a kind of ShortTerm Load Forecasting Method for considering meteorologic factor, it is characterised in that comprise the following steps：
1) historical load data and corresponding history meteorological data of area power grid are collected, and type carries out category filter to load data and meteorological data by date, detects and correct abnormal data；
2) correlation analysis, analytical procedure 1 are used) degree of correlation of the load data that obtains and each meteorologic factor, it is determined that the crucial meteorologic factor of influence this area's load；
3) comprehensive meteorologic factor is set up with the correlation of crucial meteorologic factor according to regional electric load；
4) Fast Fourier Transform (FFT) being used, to step 1) load data that obtains carries out spectrum analysis, summarizes the variation characteristic of area power grid daily load curve, finds out the typical similar day of prediction day, and sort by date；
5) Elman neutral nets are set up with comprehensive meteorologic factor using selected load, determine the structure of Elman neutral nets, Elman neutral nets are trained using glowworm swarm algorithm and optimizes network parameter, the Elman neutral nets to training are tested to set up Elman neural network shortterm load forecasting models；
6) the weather forecast data at moment to be predicted are obtained, calculate comprehensive meteorologic factor, by comprehensive meteorologic factor and corresponding load data genaration test input vector, Elman neural network shortterm load forecasting models are input into, its output is the predicted load at moment to be predicted；
7) predicted load is shown.
The history meteorological data includes：Temperature, relative humidity, wind speed, rainfall, air pressure, radiation, comfort index, Body Comfort Index.
The formula of the correlation analysis is：
In formula (1), x_{i}To need the meteorologic factor of its correlation of identification, such as temperature, humidity, wind speed, y_{i}It is load data,It is the average value of the meteorologic factor,It is the average value of load
It is described use glowworm swarm algorithm train Elman neutral nets and the step of optimize network parameter for：
Step 401：Initialization firefly population scale n, maximum iteration T_{max}, each firefly encodes the connection weight of Elman neutral nets, structure sheaf initial input and self feed back gain factor successively；
Step 402：Calculate the fitness function f of each firefly_{i}, and as the brightness of firefly, according to f_{i}The smaller firefly brightness of value principle higher, firefly is ranked up, find most bright firefly.The computing formula of fitness function is as follows：
In formula (2),It is predicted value, y_{j}It is actual value, N is total sample number.
Step 403：The Attraction Degree β of each firefly is calculated, computing formula is as follows：
In formula (3), β_{0}It is the maximum Attraction Degree of firefly；β_{min}It is minimum Attraction Degree；γ is the Absorption of Medium factor.
Step 404：The position of all nonmost bright fireflies is updated, formula is as follows：
x_{i}(t+1)=x_{i}(t)+β(x_{j}(t)x_{i}(t))+rand_{1}(x_{g}(t)x_{i}(t)) (4)
In formula (4), x_{i}(t+1) it is firefly x_{i}In the position in t+1 generations；x_{i}F () is firefly x_{i}In the position in f generations；x_{j}F () is firefly x_{j}In the position in t generations；x_{g}T () is most bright firefly in the position in f generations；rand_{1}Uniform random factor is obeyed for [0,1] is interval.
Step 405：Position to most bright firefly carries out random perturbation, it is to avoid be absorbed in local optimum too early.Formula is as follows：
In formula (5), rand_{2}It is [0,1] interval random number；α is step factor, is [0,1] interval constant；
Step 406：According to the position of firefly after renewal, the fitness function of firefly is recalculated according to formula (2).
Step 407：Judge whether to meet required precision or reached predefined iterations, if meeting, algorithm terminates；Otherwise, 203 are gone to step, into searching for next time.
Step 408：The most bright firefly information of output, i.e., the parameter of optimal Elman neutral nets.
The beneficial effects of the invention are as follows the deficiency existed for existing shortterm load forecasting method, propose a kind of ShortTerm Load Forecasting Method for considering meteorologic factor, consider influence of multiple meteorologic factors to load, the crucial meteorologic factor of selection influence load constitutes comprehensive meteorologic factor is used for Elman neutral net load predictions, it is effectively reduced the workload of load prediction, ensure precision of prediction, improve forecasting efficiency, strengthen the applicability of load forecasting model；The parameter of Elman neutral nets quickly, is reasonably determined using firefly optimized algorithm, the precision of prediction and realtime of Elman neural network prediction models is improve；The present invention can be calculated to a nicety the load data of working day, weekend and legal festivals and holidays, and reliable basis are provided for power grid operation personnel formulate generation schedule, ensure power grid security, efficient and stable operation；The present invention is easily operated, be suitable for actual engineer applied.
Brief description of the drawings
Fig. 1 is a kind of ShortTerm Load Forecasting Method flow chart for considering meteorologic factor.
Specific embodiment
The present invention proposes a kind of ShortTerm Load Forecasting Method for considering meteorologic factor, and the present invention is elaborated with specific embodiment below in conjunction with the accompanying drawings.
Fig. 1 show a kind of ShortTerm Load Forecasting Method flow chart for considering meteorologic factor, comprises the following steps：
1) historical load data and corresponding history meteorological data of area power grid are collected, and type carries out category filter to load data and meteorological data by date, detects and correct abnormal data；
2) correlation analysis, analytical procedure 1 are used) degree of correlation of the load data that obtains and each meteorologic factor, it is determined that the crucial meteorologic factor of influence this area's load；
3) comprehensive meteorologic factor is set up with the correlation of crucial meteorologic factor according to regional electric load；
4) Fast Fourier Transform (FFT) being used, to step 1) load data that obtains carries out spectrum analysis, summarizes the variation characteristic of area power grid daily load curve, finds out the typical similar day of prediction day, and sort by date；
5) Elman neutral nets are set up with comprehensive meteorologic factor using selected load, determine the structure of Elman neutral nets, Elman neutral nets are trained using glowworm swarm algorithm and optimizes network parameter, the Elman neutral nets to training are tested to set up Elman neural network shortterm load forecasting models；
6) the weather forecast data at moment to be predicted are obtained, calculate comprehensive meteorologic factor, by comprehensive meteorologic factor and corresponding load data genaration test input vector, Elman neural network shortterm load forecasting models are input into, its output is the predicted load at moment to be predicted.
7) predicted load is shown.
The history meteorological data includes：Temperature, relative humidity, wind speed, rainfall, air pressure, radiation, comfort index, Body Comfort Index.
The formula of the correlation analysis is：
In formula (1), x_{i}To need the meteorologic factor of its correlation of identification, such as temperature, humidity, wind speed, y_{i}It is load data,It is the average value of the meteorologic factor,It is the average value of load.
It is described use glowworm swarm algorithm train Elman neutral nets and the step of optimize network parameter for：
Step 401：Initialization firefly population scale n, maximum iteration T_{max}, each firefly encodes the connection weight of Elman neutral nets, structure sheaf initial input and self feed back gain factor successively；
Step 402：Calculate the fitness function f of each firefly_{i}, and as the brightness of firefly, according to f_{i}The smaller firefly brightness of value principle higher, firefly is ranked up, find most bright firefly.The computing formula of fitness function is as follows：
In formula (2),It is predicted value, y_{j}It is actual value, N is total sample number.
Step 403：The Attraction Degree β of each firefly is calculated, computing formula is as follows：
In formula (3), β_{0}It is the maximum Attraction Degree of firefly；β_{min}It is minimum Attraction Degree；γ is the Absorption of Medium factor.
Step 404：The position of all nonmost bright fireflies is updated, formula is as follows：
x_{i}(t+1)=x_{i}(t)+β(x_{j}(t)x_{i}(t))+rand_{1}(x_{g}(t)x_{i}(t)) (4)
In formula (4), x_{i}(t+1) it is firefly x_{i}In the position in t+1 generations；x_{i}T () is firefly x_{i}In the position in f generations；x_{j}T () is firefly x_{j}In the position in f generations；x_{g}T () is most bright firefly in the position in t generations；rand_{1}Uniform random factor is obeyed for [0,1] is interval.
Step 405：Position to most bright firefly carries out random perturbation, it is to avoid be absorbed in local optimum too early.Formula is as follows：
In formula (5), rand_{2}It is [0,1] interval random number；α is step factor, is [0,1] interval constant；
Step 406：According to the position of firefly after renewal, the fitness function of firefly is recalculated according to formula (2).
Step 407：Judge whether to meet required precision or reached predefined iterations, if meeting, algorithm terminates；Otherwise, 203 are gone to step, into searching for next time.
Step 408：The most bright firefly information of output, i.e., the parameter of optimal Elman neutral nets.
The present invention collects 2010 to 2014 historical load datas in this area and corresponding history meteorological data with somewhere actual electric network as embodiment.Wherein, history meteorological data includes temperature, relative humidity, wind speed, rainfall, air pressure, radiation, comfort index, Body Comfort Index.Date type is divided into working day and nonworkdays, and wherein nonworkdays is divided into legal festivals and holidays and weekend again.Load data and meteorological data are classified according to different date types, summarizes the rule of its change.Abnormal data is recognized by the vertical treatment and horizontal processing of data, bad data therein, such as load burr is rejected；For missing data, carry out curve fitting to correct missing data using the normal data of same type adjacent day.
For numerous meteorologic factors, to recognize the correlation of each meteorologic factor and load, the degree of correlation of each meteorologic factor and load data is calculated using Pearson correlation coefficient formula, computing formula is as follows：
In formula, x_{i}To need the meteorologic factor of its correlation of identification, such as temperature, humidity, wind speed, y_{i}It is load data,It is the average value of the meteorologic factor,It is the average value of load., it is necessary to be normalized to each data before degree of correlation analysis is carried out.Correlation analysis are carried out to 2010 to the 2014 per day loads in this area and each meteorologic factor, load is as shown in table 1 with the coefficient correlation of each meteorologic factor.Wherein, Body Comfort Index is the comprehensive functions of the meteorological element to human body such as measurement temperature, humidity, wind speed, characterizes human body whether comfortable in atmospheric environment.The computing formula of Body Comfort Index is as follows：
In formula, D is human comfort, and T is temperature (DEG C), and U is relative humidity (%), and V is wind speed (m/s).The computing formula of comfort index is as follows：
THI=T_{H}(0.550.55U)×(T_{H}58)
In formula, T_{H}It is Fahrenheit temperature, U is relative humidity (%).
The coefficient correlation of the per day load of table 1 and each meteorologic factor
Meteorologic factor  2010  2011  2012  2013  2014 
Mean temperature  0.8227  0.6913  0.7891  0.7056  0.8185 
Maximum temperature  0.6444  0.5533  0.6917  0.6983  0.6980 
Minimum temperature  0.5496  0.5286  0.6612  0.6497  0.6373 
Medial humidity  0.2349  0.3527  0.3953  0.4722  0.2624 
Mean wind speed  0.0224  0.0185  0.0290  0.1041  0.0483 
Average gas pressure  0.1650  0.2052  0.3140  0.2079  0.1962 
Average precipitation  0.1734  0.1010  0.1812  0.0756  0.0598 
Average radiation  0.3876  0.2657  0.3102  0.2865  0.2706 
Average comfort index  0.7332  0.6528  0.6924  0.6489  0.7264 
Average ride number  0.8334  0.7215  0.7412  0.7298  0.8165 
Found by calculating coefficient correlation, the load of this area is maximum with mean temperature and the average ride number degree of correlation, illustrate that influence of the two meteorologic factors to this area's load is larger with respect to other factors.Therefore, mean temperature and average ride number are exactly the key factor for influenceing this area's load.
Because weather forecast is typically for districts and cities' scope, not directly against the weather forecast for saving net, therefore when the relation of provincial total load and meteorologic factor is analyzed, it is necessary to carry out certain treatment to meteorological condition.If in view of the weather information of citylevel cities of the whole province is carried out, comprehensive workload is very huge and be difficult to realize, the present invention is directed to two critical load factors of mean temperature and ride number, have chosen 4 representative prefecturelevel cities.Using 4 Practical Meteorological Requirements factors of prefecturelevel city, the average weighted comprehensive meteorologic factor of crucial meteorologic factor is formed with reference to the actual electricity consumption situation in each city, as the meteorologic factor of Power ShortTerm Load Forecasting Model.
Using Fast Fourier Transform (FFT), spectrum analysis is carried out to this area's historical load data, compare the spectrum position and respective magnitudes of load curve, summarize the variation characteristic of this area's power network daily load curve, find out the typical similar day of prediction day, and sorted by date.
Set up Elman neural network shortterm load forecasting models.Choose 3 the actual negative charge values and comprehensive meteorologic factor of typical similar day synchronization, 3 the actual negative charge values and comprehensive meteorologic factor of typical similar day previous hour, predict the actual negative charge values and comprehensive meteorologic factor of day previous hour, the comprehensive meteorologic factor of prediction time, the date type of prediction time as Elman neutral nets input quantity, the predicted load of prediction time as Elman neutral nets output quantity.The structure of Elman neutral nets has 4 layers, comprising input layer, hidden layer, structure sheaf and output layer, the number of wherein input layer and output node layer is determined that the parameter such as input layer to hidden layer weights, hidden layer to output layer weights, structure sheaf initial input, self feed back gain factor is obtained using the optimization of firefly optimized algorithm by input quantity and output quantity.
The parameter setting of glowworm swarm algorithm is as follows：The population scale n=100 of firefly, Absorption of Medium factor gamma=1.0, maximum Attraction Degree β_{0}=1.0, step factor α=0.02, maximum iteration T=200.Firefly optimized algorithm trains Elman neutral nets, and algorithm exports the positional information of optimal firefly, the as optimal parameter information of Elman neutral nets after terminating.Then, the Elman neutral nets for training are tested to set up Elman neural network shortterm load forecasting models.
The weather forecast data at moment to be predicted are obtained, comprehensive meteorologic factor is calculated, by comprehensive meteorologic factor and corresponding load data genaration test input vector, Elman neural network shortterm load forecasting models is input into, its output is the predicted load at moment to be predicted.To predict the outcome and compare with actual load data, calculate the error of prediction.
This area's on July 1st, 2014 is chosen to the August load data of 17 days and meteorological data as sample data set, Elman neutral nets are trained and tested.18 working day of August (Monday) in 2014,23 weekend of Augusts (Saturday) in 2014 and the 24 hours loads of National Day of on October 3rd, 2014 are predicted using the Elman neural network prediction models set up.Table 2 is three kinds of mean errors of the hourly load forecasting of method 24, wherein Elman refers to the Elman neutral nets for not accounting for meteorologic factor and being obtained using gradient descent method training, but FAElman refers to the Elman neutral nets for not accounting for meteorologic factor using glowworm swarm algorithm Optimal Parameters, and FAElman (consideration meteorologic factor) refers to the method for the present invention.
2 three kinds of methods of table are on weekdays, weekend and legal festivals and holidays (National Day) predicts the outcome
From table 2 it can be seen that relative to Elman neutral nets and FAElman neutral nets, the method for the present invention can obtain best predicting the outcome.
The present invention proposes a kind of ShortTerm Load Forecasting Method for considering meteorologic factor, has considered multiple meteorologic factors, it is ensured that the precision and reasonability of load prediction；Using comprehensive meteorologic factor as the input of Elman neutral nets, the workload of load prediction is effectively reduced, improve the degree of accuracy and the efficiency of prediction；The parameter of Elman neutral nets is determined using firefly optimized algorithm, the accuracy and realtime of load prediction is further increased；The method of the present invention is capable of the load data of prediction work day, weekend and legal festivals and holidays exactly.
The above; the only present invention preferably specific embodiment, but protection scope of the present invention is not limited thereto, any one skilled in the art the invention discloses technical scope in; the change or replacement that can be readily occurred in, should all be included within the scope of the present invention.Therefore, protection scope of the present invention should be defined by scope of the claims.
Claims (4)
1. it is a kind of consider meteorologic factor ShortTerm Load Forecasting Method, it is characterised in that the basic step of the method is as follows：
1) historical load data and corresponding history meteorological data of area power grid are collected, and type carries out category filter to load data and meteorological data by date, detects and correct abnormal data；
2) correlation analysis, analytical procedure 1 are used) degree of correlation of the load data that obtains and each meteorologic factor, it is determined that the crucial meteorologic factor of influence this area's load；
3) comprehensive meteorologic factor is set up with the correlation of crucial meteorologic factor according to regional electric load；
4) Fast Fourier Transform (FFT) being used, to step 1) load data that obtains carries out spectrum analysis, summarizes the variation characteristic of area power grid daily load curve, finds out the typical similar day of prediction day, and sort by date；
5) Elman neutral nets are set up with comprehensive meteorologic factor using selected load, determine the structure of Elman neutral nets, Elman neutral nets are trained using glowworm swarm algorithm and optimizes network parameter, the Elman neutral nets to training are tested to set up Elman neural network shortterm load forecasting models；
6) the weather forecast data at moment to be predicted are obtained, comprehensive meteorologic factor is calculated, by comprehensive meteorologic factor and corresponding load data genaration input vector, Elman neural network shortterm load forecasting models is input into, its output is the predicted load at moment to be predicted.
7) predicted load is shown.
2. according to claim 1 it is a kind of consider meteorologic factor ShortTerm Load Forecasting Method, it is characterised in that the history meteorological data includes：Temperature, relative humidity, wind speed, rainfall, air pressure, radiation, comfort index, Body Comfort Index.
3. a kind of ShortTerm Load Forecasting Method for considering meteorologic factor according to claim 1, it is characterised in that the formula of the correlation analysis is：
In formula (1), x_{i}To need the meteorologic factor of its correlation of identification, such as temperature, humidity, wind speed, y_{i}It is load data,It is the average value of the meteorologic factor,It is the average value of load.
4. a kind of ShortTerm Load Forecasting Method for considering meteorologic factor according to claim 1, it is characterised in that it is described use glowworm swarm algorithm train Elman neutral nets and the step of optimize network parameter for：
Step 401：Initialization firefly population scale n, maximum iteration T_{max}, each firefly encodes the connection weight of Elman neutral nets, structure sheaf initial input and self feed back gain factor successively；
Step 402：Calculate the fitness function f of each firefly_{i}, and as the brightness of firefly, according to f_{i}The smaller firefly brightness of value principle higher, firefly is ranked up, find most bright firefly.The computing formula of fitness function is as follows：
In formula (2),It is predicted value, y_{j}It is actual value, N is total sample number.
Step 403：The Attraction Degree β of each firefly is calculated, computing formula is as follows：
In formula (3), β_{0}It is the maximum Attraction Degree of firefly；β_{min}It is minimum Attraction Degree；γ is the Absorption of Medium factor.
Step 404：The position of all nonmost bright fireflies is updated, formula is as follows：
x_{i}(t+1)=x_{i}(t)+β(x_{j}(t)x_{i}(t))+rand_{1}(x_{g}(t)x_{i}(t)) (4)
In formula (4), x_{i}(t+1) it is firefly x_{i}In the position in t+1 generations；x_{i}T () is firefly x_{i}In the position in t generations；x_{j}T () is firefly x_{j}In the position in t generations；x_{g}T () is most bright firefly in the position in t generations；rand_{1}Uniform random factor is obeyed for [0,1] is interval.
Step 405：Position to most bright firefly carries out random perturbation, it is to avoid be absorbed in local optimum too early.Formula is as follows：
In formula (5), rand_{2}It is [0,1] interval random number；α is step factor, is [0,1] interval constant；
Step 406：According to the position of firefly after renewal, the fitness function of firefly is recalculated according to formula (2).
Step 407：Judge whether to meet required precision or reached predefined iterations, if meeting, algorithm terminates；Otherwise, 203 are gone to step, into searching for next time.
Step 408：The most bright firefly information of output, i.e., the parameter of optimal Elman neutral nets.
Priority Applications (1)
Application Number  Priority Date  Filing Date  Title 

CN201510799160.1A CN106779129A (en)  20151119  20151119  A kind of ShortTerm Load Forecasting Method for considering meteorologic factor 
Applications Claiming Priority (1)
Application Number  Priority Date  Filing Date  Title 

CN201510799160.1A CN106779129A (en)  20151119  20151119  A kind of ShortTerm Load Forecasting Method for considering meteorologic factor 
Publications (1)
Publication Number  Publication Date 

CN106779129A true CN106779129A (en)  20170531 
Family
ID=58915937
Family Applications (1)
Application Number  Title  Priority Date  Filing Date 

CN201510799160.1A Pending CN106779129A (en)  20151119  20151119  A kind of ShortTerm Load Forecasting Method for considering meteorologic factor 
Country Status (1)
Country  Link 

CN (1)  CN106779129A (en) 
Cited By (25)
Publication number  Priority date  Publication date  Assignee  Title 

CN107330518A (en) *  20170621  20171107  国家电网公司  Energy management control method and system based on temperature adjustment load prediction 
CN107609671A (en) *  20170810  20180119  燕山大学  A kind of ShortTerm Load Forecasting Method based on composite factor evaluation model 
CN107967541A (en) *  20171221  20180427  联想(北京)有限公司  A kind of information forecasting method, information prediction device and server cluster 
CN107992975A (en) *  20171214  20180504  北京交通大学  A kind of load forecasting method that becomes more meticulous for merging historical data and Real Time Effect factor 
CN108229754A (en) *  20180131  20180629  杭州电子科技大学  Shortterm load forecasting method based on similar day segmentation and LMBP networks 
CN108335014A (en) *  20171229  20180727  国网北京市电力公司  Load Analytic Method, device, storage medium and processor 
CN108446808A (en) *  20180408  20180824  广东电网有限责任公司  A kind of shortterm load forecasting method of glowworm swarm algorithm optimization neural network 
CN108460486A (en) *  20180305  20180828  国网上海市电力公司  A kind of voltage deviation prediction technique based on improvement clustering algorithm and neural network 
CN108764522A (en) *  20180413  20181106  华南理工大学  A kind of localised load prediction technique based on multivariable input 
CN108830405A (en) *  20180529  20181116  东北电力大学  Realtime electric power load prediction system and method based on multi objective Dynamic Matching 
CN108932557A (en) *  20180428  20181204  云南电网有限责任公司临沧供电局  A kind of Shortterm Load Forecasting Model based on temperature cumulative effect and grey relational grade 
CN109002937A (en) *  20180907  20181214  深圳供电局有限公司  Load Forecasting, device, computer equipment and storage medium 
CN109002915A (en) *  20180713  20181214  福州大学  Photovoltaic plant short term power prediction technique based on KmeansGRAElman model 
CN109034500A (en) *  20180904  20181218  湘潭大学  A kind of midterm electric load forecasting method of multiple timings collaboration 
CN109034495A (en) *  20180830  20181218  珠海吉瓦科技有限公司  Electric Load Prediction System based on edge calculations 
CN109118025A (en) *  20180925  20190101  新智数字科技有限公司  A kind of method and apparatus of electric system prediction 
CN109146063A (en) *  20180827  20190104  广东工业大学  A kind of more segmentation shortterm load forecasting methods based on vital point segmentation 
CN109190856A (en) *  20181030  20190111  深圳供电局有限公司  A kind of urban distribution network shortterm load forecasting method that distinguishing day to be predicted and system 
CN109387753A (en) *  20181008  20190226  国网福建省电力有限公司  A kind of composite insulator degradation level division methods of diversification 
CN109858700A (en) *  20190201  20190607  华北水利水电大学  BP neural network heating system energy consumption prediction technique based on similar screening sample 
CN110033134A (en) *  20190409  20190719  国网安徽省电力有限公司  A kind of shortterm load forecasting algorithm of segmentation day by day considering meteorologic factor 
CN110097236A (en) *  20190516  20190806  南京工程学院  A kind of shortterm load forecasting method based on FA optimization Elman neural network 
CN110147908A (en) *  20190522  20190820  东莞理工学院  A kind of wind power forecasting method based on threedimensional optimal similarity and improvement cuckoo algorithm 
CN110728401A (en) *  20191010  20200124  郑州轻工业学院  Shortterm power load prediction method of neural network based on squirrel and weed hybrid algorithm 
CN112696728A (en) *  20210122  20210423  北京嘉洁能科技股份有限公司  Control system for balancing electric load and reducing electric capacity increase 
Citations (4)
Publication number  Priority date  Publication date  Assignee  Title 

US7305282B2 (en) *  20030513  20071204  Siemens Power Transmission & Distribution, Inc.  Very short term load prediction in an energy management system 
CN103606022A (en) *  20131205  20140226  国家电网公司  Shortterm load prediction method 
CN104318332A (en) *  20141029  20150128  国家电网公司  Power load predicting method and device 
CN104778503A (en) *  20150316  20150715  国家电网公司  Method for forecasting public building air conditioner shorttime base wire load with consideration of realtime weather factors 

2015
 20151119 CN CN201510799160.1A patent/CN106779129A/en active Pending
Patent Citations (4)
Publication number  Priority date  Publication date  Assignee  Title 

US7305282B2 (en) *  20030513  20071204  Siemens Power Transmission & Distribution, Inc.  Very short term load prediction in an energy management system 
CN103606022A (en) *  20131205  20140226  国家电网公司  Shortterm load prediction method 
CN104318332A (en) *  20141029  20150128  国家电网公司  Power load predicting method and device 
CN104778503A (en) *  20150316  20150715  国家电网公司  Method for forecasting public building air conditioner shorttime base wire load with consideration of realtime weather factors 
NonPatent Citations (2)
Title 

胡长洪: "考虑实时气象因素的短期负荷预测", 《中国优秀硕士学位论文全文数据库工程科技II辑》 * 
马彦追: "萤火虫算法的改进及其应用研究", 《中国优秀硕士学位论文全文数据库信息科技辑》 * 
Cited By (26)
Publication number  Priority date  Publication date  Assignee  Title 

CN107330518A (en) *  20170621  20171107  国家电网公司  Energy management control method and system based on temperature adjustment load prediction 
CN107609671A (en) *  20170810  20180119  燕山大学  A kind of ShortTerm Load Forecasting Method based on composite factor evaluation model 
CN107992975A (en) *  20171214  20180504  北京交通大学  A kind of load forecasting method that becomes more meticulous for merging historical data and Real Time Effect factor 
CN107967541A (en) *  20171221  20180427  联想(北京)有限公司  A kind of information forecasting method, information prediction device and server cluster 
CN108335014A (en) *  20171229  20180727  国网北京市电力公司  Load Analytic Method, device, storage medium and processor 
CN108229754A (en) *  20180131  20180629  杭州电子科技大学  Shortterm load forecasting method based on similar day segmentation and LMBP networks 
CN108460486A (en) *  20180305  20180828  国网上海市电力公司  A kind of voltage deviation prediction technique based on improvement clustering algorithm and neural network 
CN108446808A (en) *  20180408  20180824  广东电网有限责任公司  A kind of shortterm load forecasting method of glowworm swarm algorithm optimization neural network 
CN108764522A (en) *  20180413  20181106  华南理工大学  A kind of localised load prediction technique based on multivariable input 
CN108932557A (en) *  20180428  20181204  云南电网有限责任公司临沧供电局  A kind of Shortterm Load Forecasting Model based on temperature cumulative effect and grey relational grade 
CN108830405A (en) *  20180529  20181116  东北电力大学  Realtime electric power load prediction system and method based on multi objective Dynamic Matching 
CN109002915A (en) *  20180713  20181214  福州大学  Photovoltaic plant short term power prediction technique based on KmeansGRAElman model 
CN109146063A (en) *  20180827  20190104  广东工业大学  A kind of more segmentation shortterm load forecasting methods based on vital point segmentation 
CN109034495A (en) *  20180830  20181218  珠海吉瓦科技有限公司  Electric Load Prediction System based on edge calculations 
CN109034500A (en) *  20180904  20181218  湘潭大学  A kind of midterm electric load forecasting method of multiple timings collaboration 
CN109002937A (en) *  20180907  20181214  深圳供电局有限公司  Load Forecasting, device, computer equipment and storage medium 
CN109118025A (en) *  20180925  20190101  新智数字科技有限公司  A kind of method and apparatus of electric system prediction 
WO2020063690A1 (en) *  20180925  20200402  新智数字科技有限公司  Electrical power system prediction method and apparatus 
CN109387753A (en) *  20181008  20190226  国网福建省电力有限公司  A kind of composite insulator degradation level division methods of diversification 
CN109190856A (en) *  20181030  20190111  深圳供电局有限公司  A kind of urban distribution network shortterm load forecasting method that distinguishing day to be predicted and system 
CN109858700A (en) *  20190201  20190607  华北水利水电大学  BP neural network heating system energy consumption prediction technique based on similar screening sample 
CN110033134A (en) *  20190409  20190719  国网安徽省电力有限公司  A kind of shortterm load forecasting algorithm of segmentation day by day considering meteorologic factor 
CN110097236A (en) *  20190516  20190806  南京工程学院  A kind of shortterm load forecasting method based on FA optimization Elman neural network 
CN110147908A (en) *  20190522  20190820  东莞理工学院  A kind of wind power forecasting method based on threedimensional optimal similarity and improvement cuckoo algorithm 
CN110728401A (en) *  20191010  20200124  郑州轻工业学院  Shortterm power load prediction method of neural network based on squirrel and weed hybrid algorithm 
CN112696728A (en) *  20210122  20210423  北京嘉洁能科技股份有限公司  Control system for balancing electric load and reducing electric capacity increase 
Similar Documents
Publication  Publication Date  Title 

CN106779129A (en)  A kind of ShortTerm Load Forecasting Method for considering meteorologic factor  
Xue et al.  Fault detection and operation optimization in district heating substations based on data mining techniques  
CN103853106B (en)  A kind of energy consumption Prediction Parameters optimization method of building energy supplied equipment  
CN105069525A (en)  Allweather 96point daily load curve prediction and optimization correction system  
Meng et al.  Degreeday based nondomestic building energy analytics and modelling should use building and type specific base temperatures  
US10175709B2 (en)  Consumer electric power control system and consumer electric power control method  
CN104484715A (en)  Neural network and particle swarm optimization algorithmbased building energy consumption predicting method  
CN104700321A (en)  Analytical method of state running tendency of transmission and distribution equipment  
KR101800286B1 (en)  Method and system for managing energy usage with using big date of energy usage  
Luo et al.  Feature extraction and genetic algorithm enhanced adaptive deep neural network for energy consumption prediction in buildings  
CN106951611A (en)  A kind of severe cold area energysaving design in construction optimization method based on user's behavior  
Qiu et al.  Data mining based framework to identify rule based operation strategies for buildings with power metering system  
Do et al.  Residential building energy consumption: A review of energy data availability, characteristics, and energy performance prediction methods  
CN107730031A (en)  A kind of ultrashort term peak load forecasting method and its system  
CN102509173B (en)  A kind of based on markovian power system load Accurate Prediction method  
CN103440410A (en)  Main variable individual defect probability forecasting method  
CN104252647B (en)  Electroload forecast method based on antidistance weighting interpolation method  
Nikolaou et al.  Review and state of the art on methodologies of buildings’ energyefficiency classification  
CN109858668B (en)  Coordination prediction method for power load region in thunder and lightning climate  
CN110880055A (en)  Building intelligent ammeter system  
Song et al.  An Indoor Temperature Prediction Framework Based on Hierarchical Attention Gated Recurrent Unit Model for Energy Efficient Buildings  
CN110533213A (en)  Transmission line of electricity defect Risk Modeling and its prediction technique based on support vector machines  
CN110033134A (en)  A kind of shortterm load forecasting algorithm of segmentation day by day considering meteorologic factor  
Sun et al.  Implementation of hybrid shortterm load forecasting system with analysis of temperature sensitivities  
CN109035067A (en)  Building energy consumption processing method and processing device based on RF and ARMA algorithm 
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  
WD01  Invention patent application deemed withdrawn after publication 
Application publication date: 20170531 

WD01  Invention patent application deemed withdrawn after publication 