CN110175705A - A kind of load forecasting method and the memory comprising this method, system - Google Patents
A kind of load forecasting method and the memory comprising this method, system Download PDFInfo
- Publication number
- CN110175705A CN110175705A CN201910373982.1A CN201910373982A CN110175705A CN 110175705 A CN110175705 A CN 110175705A CN 201910373982 A CN201910373982 A CN 201910373982A CN 110175705 A CN110175705 A CN 110175705A
- Authority
- CN
- China
- Prior art keywords
- historical
- data
- classification
- cluster
- date
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000013277 forecasting method Methods 0.000 title claims description 19
- 230000006870 function Effects 0.000 claims abstract description 66
- 238000007781 pre-processing Methods 0.000 claims description 11
- 240000002853 Nelumbo nucifera Species 0.000 claims description 6
- 235000006508 Nelumbo nucifera Nutrition 0.000 claims description 6
- 235000006510 Nelumbo pentapetala Nutrition 0.000 claims description 6
- 238000005314 correlation function Methods 0.000 description 5
- 238000001514 detection method Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000005611 electricity Effects 0.000 description 4
- 238000004140 cleaning Methods 0.000 description 3
- 238000004836 empirical method Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000012937 correction Methods 0.000 description 2
- 238000011156 evaluation Methods 0.000 description 2
- 238000012360 testing method Methods 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 201000004569 Blindness Diseases 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007423 decrease Effects 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 230000007774 longterm Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 235000015277 pork Nutrition 0.000 description 1
- 238000000513 principal component analysis Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 238000002922 simulated annealing Methods 0.000 description 1
- 230000009885 systemic effect Effects 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/23—Clustering techniques
- G06F18/232—Non-hierarchical techniques
- G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
- G06F18/23211—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with adaptive number of clusters
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/06—Energy or water supply
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Tourism & Hospitality (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Probability & Statistics with Applications (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Public Health (AREA)
- Water Supply & Treatment (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of Methods of electric load forecasting, for this method according to whether festivals or holidays carry out Preliminary division to data, the secondary division of result cluster progress according to temperature, humidity, wind speed to Preliminary division obtains multiple subclassifications；Then, fitting of a polynomial is carried out to each subclassification and obtains multiple fitting functions；Final anticipation function is finally integrated into according to each classification.The present invention is effectively improved polynomial fitting, can improve the accuracy of algorithm under operational circumstances guaranteeing the simple of prediction algorithm.
Description
Technical field
The present invention relates in smart grid load forecast field more particularly to a kind of load forecasting method and comprising
The memory of this method, system.
Background technique
Basis of the load forecast as Operation of Electric Systems, can refer to the configuration for leading power equipment and the supply of electric power.
The accuracy of prediction is directly related to the effect of entire electric system, if prediction result is lower than actual value, electricity needs can not
It is satisfied, the power failure of scale will occurs, systemic breakdown is even resulted in when serious；If prediction result is higher than actual value, configuration
Power equipment will be idle, to result in waste of resources.
Tradition uses empirical method to the prediction of long Electric Power Load, and " dividing pork from top to bottom " is either " from lower
Upper assorted cold dishes ", theoretical property is weaker, and the result of distribution determines by rule of thumb entirely.And in intelligent electric power field, mainly apply nerve
Network algorithm uses historical data training neural network to find the case where being close with historical data, thus to identical
The date electric load of meteorologic factor is predicted, but prediction model fails to pay attention to the area of festivals or holidays electric power and working day electric power
Not, it is concerned about the weight for usually assigning subjectivity of festivals or holidays and working day electricity consumption difference, the confidence level of prediction is caused to reduce.
Meanwhile existing intelligent algorithm has been also faced with the problem of model complexity, causes operand larger, multi-intelligence algorithm is needed to match
It closes, convergence speed of the algorithm could be improved, therefore not can avoid storage resource using problem larger, that the calculating time is grown, cause
Methods of electric load forecasting slowly cannot be promoted and applied widely.
Summary of the invention
In order to overcome the blindness of empirical method in the prior art, subjectivity, system complexity, the effect of intelligent prediction algorithms
Forthright problem, the present invention provides a kind of load forecasting methods, to solve the above problems.
The purpose of the present invention is implemented with the following technical solutions:
A kind of load forecasting method, it is characterised in that the following steps are included:
S1: acquiring the historical data of historical date, the historical data include historical load data and with the history
The corresponding history meteorological data of load data, the history meteorological data include temperature, humidity, wind speed；
S2: the historical data for belonging to the workaday historical date is classified as the first classification, inoperative will be belonged to
The historical data of the historical date of day is classified as the second classification；
S3: the p cluster set of the historical data of first classification, the q of the historical data of second classification are obtained
A cluster set；
S4: the fitting cluster set obtains function relevant to first classificationIt obtains and described the
The relevant function of two classification
S5: being integrated into the number occurred in period and the first classification or the second control by kinds data bulk ratio according to cluster,
Calculate the contribution probability of the cluster setAndObtain the first output functionObtain the second output function
S6: the forecast date is mapped to the historical date T by the input prediction date；
S7: the historical date T is inputted into first output function or second output function, it is negative to obtain prediction
Lotus.
Further, in step sl, first the historical data is cleaned when acquisition.
Further, in step s3, the cluster is obtained using Euclidean distance calculating method to gather.
Further, the Euclidean distance calculating method comprises the steps of:
In S31: Yu Suoshu first classification or second classification, p or q initial point Z are takeni(xi,yi,zi), wherein i
≤ n or m, i are positive integer, xiFor temperature, yiFor humidity, ziFor wind speed；By the point Zi(xi,yi,zi) be set as in initial clustering
The heart；
S32: any of the first classification or the second classification point P are successively calculatedf(xf,yf,zf) arrive the cluster centre most
Short distance obtains p or q cluster according to shortest distance principle；
S33: calculating the cluster centre of each cluster, and new cluster centre is exactly the average value of data object distance in class,
New cluster centre is denoted as Zi(xi,yi,zi)；
S34: if reaching termination condition, stopping calculating, and otherwise returns to S32 step.
Further, the termination condition of step S34 is to have reached the maximum number of iterations or step S33 meter of setting
The new cluster centre calculated, is no longer changed relative to the cluster centre in step S32.
Further, the relevant function is polynomial function.
Further, the coefficient of the polynomial function is solved using intelligent algorithm.
Further, the coefficient of the polynomial function is solved using the method for seeking partial derivative.
Further, further include that historical data is inputted to first output function and second output function, calculate
Square mean error amount, relative error maximum value, relative error minimum value；Fitting result is evaluated.
The invention also includes a kind of load prediction system, scheme is as follows:
A kind of load prediction system, including data preprocessing module, fitting of a polynomial module, forecast date input module,
Load prediction output module；
Data preprocessing module, for acquiring the historical data of historical date, the historical data includes historical load number
Accordingly and history meteorological data corresponding with the historical load data, the history meteorological data include temperature, humidity, wind
Speed；
Cluster module will belong to workaday described for handling the historical data of the data preprocessing module output
The historical data of historical date is classified as the first classification, will belong to the historical data of the historical date of nonworkdays
It is classified as the second classification；The p cluster set of the historical data of first classification is obtained, the historical data of second classification
Q cluster set；
Fitting of a polynomial module obtains function relevant to first classification for being fitted the cluster setObtain function relevant to second classificationAccording to cluster be integrated into the number that occurs in period with
First classification or the second control by kinds data bulk ratio calculate the contribution probability of the cluster setAndObtain first
Output functionObtain the second output function
Forecast date input module is used for the input prediction date, the forecast date is mapped to the historical date T；
Load prediction output module, for storing the first output functionAnd second output functionThe historical date T is inputted into first output function or second output function, it is negative to obtain prediction
Lotus.
The invention also includes a kind of memory, load forecasting method is stored in memory.
Compared with prior art, the beneficial effects of the present invention are:
(1) present invention according to influence electric load multiple factors, i.e., whether festivals or holidays, temperature, humidity, wind speed, logarithm
According to being clustered, cluster is relatively easy intuitive, can rule of thumb select initial cluster center and cluster number, the effect of cluster can
Direct Test.
(2) the final anticipation function of the present invention, it is contemplated that the probability that each anticipation function occurs considers each cluster
It is as a result more accurate, reliable to the contribution rate of predicted time electric load amount.
Detailed description of the invention
Fig. 1: for load forecasting method of the invention；
Fig. 2: for data preprocessing module shown in FIG. 1；
Fig. 3: for cluster module shown in FIG. 1；
Fig. 4: for the fitting of a polynomial module of Fig. 1；
Fig. 5: for the correction verification module of Fig. 1.
Specific embodiment
The present invention provides a kind of load forecasting method.Several big factors that will affect electric load extract, and are made with this
For the foundation of cluster, then fitting of a polynomial is carried out for each cluster result and obtains multiple functions, finally according to the contribution of cluster
Rate is integrated into final anticipation function.The present invention is effectively improved polynomial fitting, is guaranteeing prediction algorithm
Simple can improve the accuracy of algorithm under operational circumstances.In the following, in conjunction with attached drawing and specific embodiment, to the present invention
It is described further:
The present invention analyzes data by Principal Component Analysis, and having obtained temperature, humidity, wind speed is to influence power load
An important factor for lotus, and in same weather, the electric load of working day and nonworkdays has significant difference.Cause
This acquires basis of the historical data on phase of history date as anticipation function training, and historical data includes historical load number
Accordingly and history meteorological data corresponding with historical load data, history meteorological data include temperature, humidity, wind speed.History number
According to acquisition time will be close to forecast date, and the historical data of long period need to be obtained as far as possible.And anticipation function needs to determine
Shi Gengxin can according to circumstances choose whether to be modified model before prediction, such as predict the electricity consumption on 1 day May the year two thousand thirty, selection
2010-2019 4, the data in May as historical data, this is clearly inappropriate, once 2019-2029 year-climate has
The prediction effect of significant change, model will be very inaccurate.
Referring to attached drawing 2, after historical data acquisition, data incompleteness is likely to occur in historical data, such as lack temperature, humidity,
It is one or more in wind speed, it is also possible to appear in the electric power for occurring that an exception is high or exception is low in continuous historical date
Load needs to reject the above abnormal data, i.e., cleans, had to historical data when obtaining historical data
There is the historical data of reference value.The cleaning method of historical data includes but is not limited to following methods: value lacks detection method, mistake
It is worth detection method, repeats record detection method, inconsistency detection method.Due to the main electricity consumption main body of working day and nonworkdays, set
It is standby inconsistent, it is therefore desirable to which that the load prediction of working day and nonworkdays is distinguished.Workaday described go through will be belonged to
The historical data on history date is classified as the first classification, and the historical data for belonging to the historical date of nonworkdays is returned
For the second classification.
Referring to attached drawing 3, the p cluster set of the historical data of first classification is obtained using clustering algorithm, described the
The q cluster set of the historical data of two classification.The clustering algorithm includes but is not limited to K-Means clustering procedure, mean value drift
It moves cluster, density clustering method, examined with the greatest hope cluster of gauss hybrid models, Agglomerative Hierarchical Clustering, figure group
Survey clustering procedure.
The present invention takes full advantage of Conventional wisdom, selects the totally different point of weather that can effectively improve as the initial point of cluster
Computational efficiency, and calculated result is made to meet objective law.The empirical method can be the Chinese twenty-four solar terms can also be with
It is that the weather counted according to historical time, region characteristic changes key node.And cluster is calculated using Euclidean distance calculating method.
I.e. in first classification or second classification, p or q initial point Z are takeni(xi,yi,zi), wherein i≤n
Or m, i are positive integer, xiFor temperature, yiFor humidity, ziFor wind speed；By the point Zi(xi,yi,zi) it is set as initial cluster center.
Start clustering algorithm, any of the first classification or the second classification point Pf(xf,yf,zf) arrive the cluster centre
The shortest distance:
According to the calculating of the shortest distance, it is most short to can be obtained distance of the arbitrary point apart from which cluster centre, from
And the cluster centre of the arbitrary point and its shortest distance is classified as one kind.By all historical data points according to above method meter
It calculates, p cluster can be obtained in the first classification；Q cluster is obtained in the second classification.
The cluster centre of each cluster is recalculated, new cluster centre is exactly the average value of data object distance in class,
New cluster centre is denoted as Zi(xi,yi,zi):
Wherein n indicates the number of the historical data in the cluster.
The new cluster centre of acquisition is substituted into clustering algorithm, the clustering algorithm is restarted and is calculated, when
The maximum number of iterations or cluster centre for reaching setting are no longer changed, and stop calculating, and export cluster result.
Referring to attached drawing 4, start fitting algorithm and obtain curve matching to above-mentioned cluster, fitting algorithm include but is not limited to
Lower method: Lagrange's interpolation, Newton interpolating method, Newton iteration method, section dichotomy, Secant Method, Jacobi iterative method and
Newton Ke Tesi numerical integrating.
The present invention predicts the electric load of each cluster using fitting of a polynomial, before fitting, due to right
Data have carried out clustering processing, and prediction process can not consider further that the influence of temperature, humidity, wind speed, are guaranteeing prediction accuracy
While, simplify the complexity of prediction model:
(the first classification correlation function)
(the second classification correlation function)
Wherein max is a precision index, and value can be customized, but is not that the higher the better, it is, in general, that choosing
The result for selecting the higher sample data of number is more preferable, but the result of test data can decline instead.The value of max in the present invention
Range is 10 to 100, preferably 25.
The coefficient solution of the first classification correlation function and the second classification correlation function can use genetic algorithm, particle
Group method or simulated annealing etc..
In the present invention since data sample sufficiently uses high-efficient partial differentiation, i.e. construction range formula is minimized:
It enables
WhenIt is equally that distance is minimum, due to minimum value, that is, partial derivative when minimum
It is at zero, it may be assumed that
It can get max local derviation equation, the coefficient of correlation function can be obtained by needing to obtain at least max historical data,
To obtain the relevant function of the first classificationObtain function relevant to second classification
However each influence clustered in the period of certain to load prediction is inconsistent, therefore to be calculated every
The contribution probability of one cluster.The contribution probability of each cluster can be using within one period, which goes out in this time
Existing number and total time is than obtaining.Specifically contribution probability be any cluster in historical data number with first classification or
The ratio of the historical data number in second classification:
O=n when for the first classification, o=m when for the second classification.To obtain the contribution probability of the cluster setAndObtain the first output functionObtain the second output function
Since in same year, influence of a certain cluster to load is also not the same, therefore is also needed in calculating ratio
Impact factor a is added in valuei, it may be assumed that
Obtain the first output functionObtain the second output function
During prediction, the forecast date is mapped to the historical date T by user's input prediction date；It is described
It maps including but not limited to following methods: belonging to a certain date in year or belong to a certain date of the lunar calendar.
The historical date T is inputted into first output function or second output function, obtains prediction load.
Further include that historical data is inputted to first output function and second output function referring to attached drawing 5, calculates
Square mean error amount, relative error maximum value, minimum value；Fitting result is evaluated.By evaluation method, first can be obtained
The accuracy of output function and the second output function.To be adjusted in due course.
Further, first output function and the second output function should periodically update, that is, execute the several years
When afterwards or weather produces variation, resurveys historical data and adjust the first output function and the second output function, so as to its energy
It is long-term to meet prediction requirement.
The present invention contains a kind of memory, and memory can store load forecasting method wherein.
The solution of the present invention further includes a kind of load prediction system, referring to attached drawing 1, including data preprocessing module, cluster
Module, fitting of a polynomial module, forecast date input module, load prediction output module.
Referring to attached drawing 2, data preprocessing module, for acquiring the historical data of historical date, the historical data includes
Historical load data and history meteorological data corresponding with the historical load data, the history meteorological data include temperature
Degree, humidity, wind speed.
Referring to attached drawing 3, cluster module will belong to work for handling the historical data of the data preprocessing module output
The historical data for making the historical date of day is classified as the first classification, will belong to the institute of the historical date of nonworkdays
It states historical data and is classified as the second classification；The p cluster set of the historical data of first classification is obtained, second classification
Q cluster of historical data is gathered；
Referring to attached drawing 4, fitting of a polynomial module obtains related to first classification for being fitted the cluster set
FunctionObtain function relevant to second classificationIt is integrated into according to cluster and is occurred in period
Number and the first classification or the second control by kinds data bulk ratio calculate the contribution probability of the cluster setAndIt obtains
First output functionObtain the second output function
Forecast date input module is used for the input prediction date, the forecast date is mapped to the historical date T.
Load prediction output module, for storing the first output functionAnd second output functionThe historical date T is inputted into first output function or second output function, it is negative to obtain prediction
Lotus.
In data preprocessing module, the historical data of acquisition is needed through over cleaning, and it is negative that cleaning method can be found in the present invention
Lotus prediction technique.
In cluster module, the cluster is obtained using Euclidean distance calculating method and is gathered, the Euclidean distance calculating method packet
Containing following steps: in the classification of Yu Suoshu first or second classification, taking p or q initial point Zi(xi,yi,zi), wherein i≤
N or m, i are positive integer, xiFor temperature, yiFor humidity, ziFor wind speed；By the point Zi(xi,yi,zI) it is set as initial cluster center；
Successively calculate any of the first classification or the second classification point Pf(xf,yf,zf) arrive the cluster centre the shortest distance, according to
Shortest distance principle obtains p or q cluster；The cluster centre of each cluster is calculated, new cluster centre is exactly data in class
The average value of object distance, new cluster centre are denoted as zi(xi,yi,zi)；If reaching termination condition, stop calculating, otherwise will
Cluster centre zi(xi,yi,zi) return, recalculate cluster.The termination condition is to have reached the maximum number of iterations of setting
Or the new cluster centre that step S33 is calculated, it is no longer changed relative to the cluster centre in step S32.It is European
Apart from calculating method reference can be made to load forecasting method of the present invention.
In fitting of a polynomial module, the relevant function is polynomial function, is solved using intelligent algorithm described multinomial
The coefficient of formula function or the coefficient that the polynomial function is solved using the method for seeking partial derivative.The polynomial function solution side
The visible load forecasting method of the invention of method.
Referring to attached drawing 5, load prediction system further includes correction verification module, for historical data to be inputted the first output letter
Several and second output function calculates square mean error amount, relative error maximum value, minimum value；Fitting result is evaluated.
By evaluation method, the accuracy of the first output function and the second output function can be obtained.To be adjusted in due course.
It will be apparent to those skilled in the art that can make various other according to the above description of the technical scheme and ideas
Corresponding change and deformation, and all these changes and deformation all should belong to the protection scope of the claims in the present invention
Within.
Claims (10)
1. a kind of load forecasting method, it is characterised in that the following steps are included:
S1: acquiring the historical data of historical date, the historical data include historical load data and with the historical load
The corresponding history meteorological data of data, the history meteorological data include temperature, humidity, wind speed；
S2: the historical data for belonging to the workaday historical date is classified as the first classification, nonworkdays will be belonged to
The historical data of the historical date is classified as the second classification；
S3: obtaining the p cluster set of the historical data of first classification, and q of the historical data of second classification are poly-
Class set；
S4: the fitting cluster set obtains function relevant to first classificationIt obtains and described second point
The relevant function of class
S5: the number occurred in period and the first classification or the second control by kinds data bulk ratio are integrated into according to cluster, calculated
The contribution probability of the cluster setAndObtain the first output functionObtain the second output function
S6: the forecast date is mapped to the historical date T by the input prediction date；
S7: inputting first output function or second output function for the historical date T, obtains prediction load.
2. load forecasting method according to claim 1, it is characterised in that: in step sl, when acquisition is first gone through to described
History data are cleaned.
3. load forecasting method according to claim 1 or 2, it is characterised in that: in step s3, using Euclidean distance meter
Algorithm obtains the cluster set.
4. load forecasting method according to claim 3, it is characterised in that: the Euclidean distance calculating method includes following step
It is rapid:
In S31: Yu Suoshu first classification or second classification, p or q initial point Z are takeni(xi,yi,zi), wherein i≤n or
M, i are positive integer, xiFor temperature, yiFor humidity, ziFor wind speed；By the point Zi(xi,yi,zi) it is set as initial cluster center；
S32: any of the first classification or the second classification point P are successively calculatedf(xf,yf,zf) arrive the cluster centre most short distance
From according to p or q cluster of shortest distance principle acquisition；
S33: calculating the cluster centre of each cluster, and new cluster centre is exactly the average value of data object distance in class, new
Cluster centre is denoted as Zi(xi,yi,zi)；
S34: if reaching termination condition, stopping calculating, and otherwise returns to S32 step.
5. load forecasting method according to claim 4, it is characterised in that: the termination condition of step S34 is
The new cluster centre that the maximum number of iterations or step S33 for reaching setting calculate, relative to described in step S32
Cluster centre is no longer changed.
6. load forecasting method according to claim 1 or 2, it is characterised in that: the relevant function is multinomial letter
Number, the coefficient of the polynomial function is solved using intelligent algorithm.
7. load forecasting method according to claim 6, it is characterised in that: solved using the method for seeking partial derivative described more
The coefficient of item formula function.
8. load forecasting method according to claim 1 or 2, it is characterised in that: further include historical data is inputted described in
First output function and second output function calculate square mean error amount, relative error maximum value, relative error minimum value；
Fitting result is evaluated.
9. a kind of load prediction system, including data preprocessing module, fitting of a polynomial module, forecast date input module are born
Lotus predicts output module, it is characterised in that:
Data preprocessing module, for acquiring the historical data of historical date, the historical data include historical load data with
And history meteorological data corresponding with the historical load data, the history meteorological data include temperature, humidity, wind speed；
Cluster module will belong to the workaday history for handling the historical data of the data preprocessing module output
The historical data on date is classified as the first classification, and the historical data for belonging to the historical date of nonworkdays is classified as
Second classification；The p cluster set of the historical data of first classification is obtained, the q of the historical data of second classification is a
Cluster set；
Fitting of a polynomial module obtains function relevant to first classification for being fitted the cluster set
Obtain function relevant to second classificationThe number occurred in period and the first classification are integrated into according to cluster
Or the second control by kinds data bulk ratio, calculate the contribution probability of the cluster setAndObtain the first output functionObtain the second output function
Forecast date input module is used for the input prediction date, the forecast date is mapped to the historical date T；
Load prediction output module, for storing the first output functionAnd second output function
The historical date T is inputted into first output function or second output function, obtains prediction load.
10. a kind of memory, it is characterised in that: be stored with load prediction described in claim 1 to 8 any one in memory
Method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910373982.1A CN110175705B (en) | 2019-05-07 | 2019-05-07 | Load prediction method and memory and system comprising same |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910373982.1A CN110175705B (en) | 2019-05-07 | 2019-05-07 | Load prediction method and memory and system comprising same |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110175705A true CN110175705A (en) | 2019-08-27 |
| CN110175705B CN110175705B (en) | 2021-11-09 |
Family
ID=67691361
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910373982.1A Active CN110175705B (en) | 2019-05-07 | 2019-05-07 | Load prediction method and memory and system comprising same |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110175705B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114219144A (en) * | 2021-12-13 | 2022-03-22 | 广西电网有限责任公司北海供电局 | Main transformer load prediction method and system |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102722642A (en) * | 2012-05-25 | 2012-10-10 | 重庆市电力公司长寿供电局 | Large-fluctuation power grid load short-term forecasting method |
| CN103123665A (en) * | 2012-07-31 | 2013-05-29 | 上海交通大学 | Short-term power load forecasting method based on fuzzy clustering similar day |
| WO2015077754A1 (en) * | 2013-11-25 | 2015-05-28 | Siemens Corporation | A statistical approach to modeling and forecast of cchp energy and cooling demand and optimization cchp control setpoints |
| CN109242189A (en) * | 2018-09-12 | 2019-01-18 | 国网安徽省电力有限公司合肥供电公司 | A kind of Short-Term Load Forecasting of Electric Power System based on meteorologic factor |
| CN109583503A (en) * | 2018-12-03 | 2019-04-05 | 国网江苏省电力有限公司扬州供电分公司 | A kind of interruptible load prediction technique |
-
2019
- 2019-05-07 CN CN201910373982.1A patent/CN110175705B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102722642A (en) * | 2012-05-25 | 2012-10-10 | 重庆市电力公司长寿供电局 | Large-fluctuation power grid load short-term forecasting method |
| CN103123665A (en) * | 2012-07-31 | 2013-05-29 | 上海交通大学 | Short-term power load forecasting method based on fuzzy clustering similar day |
| WO2015077754A1 (en) * | 2013-11-25 | 2015-05-28 | Siemens Corporation | A statistical approach to modeling and forecast of cchp energy and cooling demand and optimization cchp control setpoints |
| CN109242189A (en) * | 2018-09-12 | 2019-01-18 | 国网安徽省电力有限公司合肥供电公司 | A kind of Short-Term Load Forecasting of Electric Power System based on meteorologic factor |
| CN109583503A (en) * | 2018-12-03 | 2019-04-05 | 国网江苏省电力有限公司扬州供电分公司 | A kind of interruptible load prediction technique |
Non-Patent Citations (1)
| Title |
|---|
| 陈皓等: "基于族群进化计算的多项式回归电力负荷预测", 《软件》 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114219144A (en) * | 2021-12-13 | 2022-03-22 | 广西电网有限责任公司北海供电局 | Main transformer load prediction method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110175705B (en) | 2021-11-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108846517B (en) | Integration method for predicating quantile probabilistic short-term power load | |
| CN103093288B (en) | Based on the partition power grid bus load prognoses system of weather information | |
| CN114792156B (en) | Photovoltaic output power prediction method and system based on curve characteristic index clustering | |
| CN106095639A (en) | A kind of cluster subhealth state method for early warning and system | |
| CN106682763B (en) | Power load optimization prediction method for large amount of sample data | |
| CN104850612B (en) | Distribution network user load characteristic classification method based on enhanced aggregation hierarchical clustering | |
| CN110263984A (en) | Ultra-short term net load prediction technique based on phase space reconfiguration and deep neural network | |
| CN105701562B (en) | Training method, applicable method for predicting generated power and respective system | |
| CN113515512A (en) | Quality control and improvement method for industrial internet platform data | |
| CN113744089B (en) | Transformer area household variable relation identification method and device | |
| CN110969306A (en) | Power distribution low-voltage distribution area load prediction method and device based on deep learning | |
| CN110188956A (en) | Load forecasting method based on ant colony neural network and the system comprising this method, memory | |
| CN113991711B (en) | Capacity configuration method for energy storage system of photovoltaic power station | |
| CN110298765A (en) | A kind of power distribution network electric power method for detecting abnormality based on objective correlative factor | |
| CN109919382A (en) | Consider the non-precision probability forecasting method of photovoltaic power climbing event and system that date periodicity influences | |
| CN109858668B (en) | Coordination prediction method for power load region in thunder and lightning climate | |
| CN117851908B (en) | Improved on-line low-voltage transformer area electric energy meter misalignment monitoring method and device | |
| CN109858785A (en) | A kind of method and system for evaluating intelligent electric energy meter operating status | |
| CN115375205A (en) | Method, device and equipment for determining water user portrait | |
| CN115423264A (en) | Line loss interval self-adaptive evaluation method and system based on deep learning | |
| CN110175705A (en) | A kind of load forecasting method and the memory comprising this method, system | |
| CN117910668A (en) | Power system evolution path planning method considering multiple uncertainty factors | |
| CN107330548A (en) | Residential electricity consumption Forecasting Methodology based on time series models | |
| CN116306290A (en) | Method for predicting concrete dam displacement confidence interval by using measured temperature | |
| CN109886455A (en) | Very Short-Term Load Forecasting Method under thunder and lightning weather |
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 | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: Room 509, building D, 39 Ruihe Road, Huangpu District, Guangzhou City, Guangdong Province Applicant after: Guangzhou Shuimu Qinghua Technology Co.,Ltd. Address before: Room 936, Room 333 Jiufojian Road, Zhongxin Guangzhou Knowledge City, Huangpu District, Guangzhou City, Guangdong Province Applicant before: Guangzhou Shuimu Qinghua Technology Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP02 | Change in the address of a patent holder | ||
| CP02 | Change in the address of a patent holder |
Address after: Room 604, Building 2, No. 5 Yunkai Road, Huangpu District, Guangzhou City, Guangdong Province, 510535 Patentee after: Guangzhou Shuimu Qinghua Technology Co.,Ltd. Address before: 510700 room 509, building D, 39 Ruihe Road, Huangpu District, Guangzhou City, Guangdong Province Patentee before: Guangzhou Shuimu Qinghua Technology Co.,Ltd. |