CN108596242B - Prediction method of power grid meteorological load based on wavelet neural network and support vector machine - Google Patents
Prediction method of power grid meteorological load based on wavelet neural network and support vector machine Download PDFInfo
- Publication number
- CN108596242B CN108596242B CN201810359331.2A CN201810359331A CN108596242B CN 108596242 B CN108596242 B CN 108596242B CN 201810359331 A CN201810359331 A CN 201810359331A CN 108596242 B CN108596242 B CN 108596242B
- Authority
- CN
- China
- Prior art keywords
- load
- data
- meteorological
- historical
- day
- 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.)
- Active
Links
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 60
- 238000012706 support-vector machine Methods 0.000 title claims abstract description 40
- 238000000034 method Methods 0.000 title claims abstract description 25
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 19
- 238000012549 training Methods 0.000 claims abstract description 18
- 238000012847 principal component analysis method Methods 0.000 claims abstract description 11
- 238000012545 processing Methods 0.000 claims description 20
- 230000001186 cumulative effect Effects 0.000 claims description 6
- 238000013277 forecasting method Methods 0.000 claims 2
- 230000000717 retained effect Effects 0.000 claims 2
- 238000012795 verification Methods 0.000 claims 1
- 238000003062 neural network model Methods 0.000 abstract description 5
- 238000004364 calculation method Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 6
- 238000004519 manufacturing process Methods 0.000 description 3
- 238000000513 principal component analysis Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000013519 translation Methods 0.000 description 2
- 230000006399 behavior Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000007621 cluster analysis Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 238000003064 k means clustering Methods 0.000 description 1
- 230000014759 maintenance of location Effects 0.000 description 1
- 238000012067 mathematical method Methods 0.000 description 1
- 238000013178 mathematical model Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
Images
Classifications
-
- 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
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—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)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Physics & Mathematics (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Tourism & Hospitality (AREA)
- Artificial Intelligence (AREA)
- General Business, Economics & Management (AREA)
- Marketing (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Primary Health Care (AREA)
- Public Health (AREA)
- Quality & Reliability (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Entrepreneurship & Innovation (AREA)
- Water Supply & Treatment (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Development Economics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a power grid meteorological load prediction method based on a wavelet neural network and a support vector machine. Extracting original data from the power data, simplifying the original data by combining a clustering algorithm and a principal component analysis method, standardizing the original data, inputting the standardized data to a support vector machine for training, and obtaining a load prediction model and first predicted load data; inputting the standardized data to wavelet neural network training to obtain a trained neural network and second predicted load data; and obtaining a load prediction model according to the first and second predicted load data. The invention fully considers the influence of meteorological data on load fluctuation, fully considers the scale of the data, simultaneously reduces the load and the meteorological data quantity through a clustering algorithm and a principal component analysis method, provides the prediction model, ensures the prediction precision of a support vector machine and a wavelet neural network model, improves the prediction precision, and solves the problem of low prediction precision caused by the consideration of meteorological factors.
Description
Technical Field
The invention relates to a power grid load data prediction method, in particular to a power grid meteorological load data prediction method based on a wavelet neural network and a support vector machine.
Background
The power system consists of a power grid and power consumers and has the function of economically providing reliable and standard-meeting electric energy for various consumers of the power system as far as possible so as to meet the requirements of various consumers at any time, namely the load requirement. However, in the present situation, the electric energy cannot be stored in large quantities, which requires that the system power generation should be dynamically balanced with the change of the system load at any time, otherwise, the quality of the power supply and utilization is affected slightly, and the safety and stability of the system are jeopardized seriously. The acquisition of the future load change of the system is realized through load prediction, so that the load prediction of the power system is developed, becomes an important research field in engineering science, and is an important content in the automation of the power system.
The load prediction of the power system is based on accurate statistical data and survey data, and a set of mathematical method for systematically processing past and future loads is researched or utilized on the basis of the history and the current situation of the power consumption under the condition of fully considering some important system operation characteristics, capacity increase decisions, natural conditions and social influences. Under the meaning of meeting certain precision requirement, the load numerical value of a certain future moment is determined.
The purpose of power load prediction is to provide the development condition and level of the load, provide a basis for the power production department and the management department to make a production plan and a development plan, and determine the power supply quantity and the production plan of each power supply area.
The result of the power load prediction is determined by the historical rule of the load, is influenced by a plurality of non-load factors, and is directly related to the applied prediction theory and the adopted prediction method. Over the years, many scholars have conducted intensive research on this subject and have proposed many methods.
The prior art has the following disadvantages:
1. most of the prior art does not fully consider the influence of meteorological data on load fluctuation and does not combine meteorological factors with load parameters.
2. Even if meteorological factors are considered, the load prediction of the prior art causes low prediction efficiency due to huge meteorological data and load data.
3. Even if meteorological factors are considered, the load prediction of the prior art has low prediction accuracy due to huge meteorological data and load data.
4. In the existing load prediction algorithm, the simplified processing of data only aims at load data or meteorological data, and the load data and the meteorological data are not simultaneously simplified and processed.
Disclosure of Invention
In order to solve the problems in the background art, the invention provides a power grid meteorological load data prediction method based on a wavelet neural network and a support vector machine, and meanwhile, the calculation amount is greatly reduced and high-precision prediction is achieved.
As shown in fig. 1, the technical scheme of the invention comprises the following steps:
the first step is as follows: extracting and obtaining first historical meteorological data, first historical load data, second historical meteorological data and second historical load data from power grid data, wherein the first historical meteorological data and the first historical load data form first original data, the second historical meteorological data and the second historical load data form second original data, the first historical meteorological data and the second historical meteorological data form original meteorological data, and the first historical load data and the second historical load data form original load data; the first historical meteorological data refers to the meteorological data of each day of the previous n years, the first historical load data refers to the load data of each day of the previous n years, the second historical meteorological data refers to the meteorological data of each day of the (n + 1) th year, the second historical load data refers to the load data of each day of the (n + 1) th year, the meteorological data of each day consists of a plurality of meteorological parameters, and the load data of each day consists of a plurality of load parameters;
the load data is composed of load parameters, and the meteorological data is composed of meteorological parameters.
The second step is that: simplifying the first original data and the second original data by adopting a mode of combining a clustering algorithm and a principal component analysis method to obtain original data corresponding to the reserved representative days, and simplifying the data volume;
the third step: standardizing the first original data and the second original data obtained by the second step, and respectively standardizing the load data and the meteorological data in a manner of taking the data of each day as a unit by adopting the following formula:
Xi＝Xireality/Xi mean
Wherein, XiWeather/load data representing the i-th representative day after normalization, XIrealityRaw weather/load data, X, representing the day represented by the ii meanRepresents the average of all the raw weather/load data for the day;
the fourth step: establishing a load prediction model by using a support vector machine, inputting standardized first historical meteorological data and first historical load data into the support vector machine for training by adopting daily prediction to obtain a trained load prediction model, and then inputting second historical meteorological data into the trained support vector machine for prediction to obtain first predicted load data;
the fifth step: adopting monthly prediction, taking the standardized first historical meteorological data as an input layer, taking the standardized first historical load data as an output layer, and adopting a wavelet neural network for training; after the training is finished, inputting second historical meteorological data into the trained neural network, and outputting to obtain second predicted load data;
the neural network is an arithmetic mathematical model which imitates the behavior characteristics of the animal neural network and performs distributed parallel information processing. The network achieves the aim of processing information by adjusting the mutual connection relationship among a large number of nodes in the network depending on the complexity of the system.
And a sixth step: and combining the results of the fourth step and the fifth step to obtain a load prediction model for predicting the load data of each day of the time period to be predicted.
And in the second step, the first original data and the second original data are respectively processed in the following modes by adopting a mode of combining a clustering algorithm and a principal component analysis method:
taking the average load parameter as a representative load parameter required by the simplification processing of the second-step clustering algorithm, and taking the load data of the representative load parameter in the (n + 1) th year as data required by the simplification processing of the second-step clustering algorithm;
selecting a representative day of the representative load parameters obtained by simplifying the clustering algorithm in the second step as a representative day of all the load parameters of the previous n +1 years;
selecting meteorological data of a representative day of the year of the first n +1 years as data required by the simplified processing of the second-step principal component analysis method;
s1: clustering load data of each day by using load data of each day as a unit by adopting a clustering algorithm, selecting load data of a middle day in each clustered class for retention, and removing the load data of the day which is not positioned in the middle from original data, thereby greatly reducing the load data amount and calling each day obtained by processing as a representative day;
the clustering algorithm adopts a K clustering algorithm.
S2: and then, processing all meteorological parameters in the meteorological data by adopting a principal component analysis method to obtain the cumulative contribution rate of each meteorological parameter, selecting the meteorological parameters with the cumulative contribution rate reaching 80% for reservation, and removing the meteorological parameters with the contribution rate not reaching 80% from the meteorological data of each day, thereby reducing the meteorological data quantity.
The invention reduces the meteorological data volume and the load data volume through the two steps, thereby relieving the difficult problem of complex big data calculation caused by considering meteorological factors.
The fourth step: the support vector machine establishes a load prediction model and processes the load prediction model in the following way:
inputting meteorological data of a certain representative day after the simplification and standardization in the second step and the standardization in the third step, outputting load data of the same representative day, traversing the representative day from the first year to the nth year, performing load prediction model training, substituting second historical meteorological data after the training is finished, and outputting first predicted load data of each representative day in the (n + 1) th year; verifying the prediction accuracy of the first predicted load data and the second historical load data, and calculating the first prediction accuracy E1。
In the fifth step, the specific structure of the wavelet neural network is as follows: the wavelet neural network is divided into three layers of an input layer, a hidden layer and an output layer: wherein the input layer contains axAn input unit, a is the number of meteorological parameters obtained by the simplified processing of the second step, dxFor the total number of days of the representative day of the Xth month obtained by the second simplified processing, each input unit represents a meteorological parameter of the representative day; output layer b x dxAn output unit, b is the number of the load parameters obtained by the simplified processing in the second step, dxFor the total number of days of the representative day of the Xth month obtained by the second simplification processing, each output unitRepresents a load parameter representative of the day; hidden layers having b x dxEach hidden unit is composed of a wavelet function, and the wavelet function adopts a Morlet mother wavelet basis function; the input data are weather and load data obtained after the second step of simplification and the third step of standardization, input and output data of all representative days in the Xth month are taken as a unit, the input and output data of all representative days in the Xth month are substituted into the neural network for training, first historical weather and first historical load data of the representative days in the Xth month of the previous n years obtained by screening in the first step are sequentially substituted into the neural network, the training is required for n times in total, and the neural network is stopped after n times; and taking 1 to 12 in turn, namely training the wavelet neural network by adopting the method every 12 months, thereby obtaining 12 groups of wavelet neural network models corresponding to 12 months. Inputting second historical meteorological data by taking a month as a basic unit, predicting by using the trained wavelet neural network to obtain second predicted load data, verifying the prediction precision of the second predicted load data and the second historical load data, and calculating second prediction precision E2。
In one embodiment, the second prediction accuracy E may be further adjusted2Judging if the second prediction accuracy E2If the prediction accuracy of the wavelet neural network is more than or equal to 90 percent, the wavelet neural network is considered to meet the requirement, and the next step is carried out. Otherwise, adjusting the wavelet function expansion factor, the translation factor, the network connection weight, the network connection threshold value and the network learning rate, and retraining the neural network until the prediction precision requirement is met.
First prediction accuracies E of the fourth and fifth steps1And a second prediction accuracy E2The method comprises the following steps:
calculating the first/second intermediate precision D of each representative day of the n +1 th year by using the historical load data and the first/second predicted load data of each representative day of the n +1 th year obtained by the processing of the fourth step and the fifth step and adopting the following formula1And D2：
Wherein D is the intermediate precision and is the first intermediate precision D of the load prediction model of the support vector machine1Or a second intermediate precision D of the wavelet neural network load prediction model2N represents the number of load parameters, i.e. daily maximum load, daily minimum load, daily peak-to-valley difference and daily average load,
a second historical load data representing a load parameter i,
first/second predicted load data representing a load parameter i;
then, the prediction accuracy is calculated by adopting the following formula:
wherein E is the prediction precision and is the first prediction precision E of the load prediction model of the support vector machine1Second prediction precision E of wavelet neural network load prediction model2A represents the total number of days of the representative day satisfying the requirement that the median accuracy D is 7% or less in the (n + 1) th year, and B represents the total number of days of the representative day in the (n + 1) th year.
The sixth step is specifically as follows: calculating a first intermediate precision D between the first predicted load data obtained by the support vector machine and the second historical load data obtained by the neural network at each load parameter of each representative day of the year1Calculating a second intermediate precision D between the second predicted load data and the second historical load data obtained by the neural network in the fifth step on each load parameter of each day2(ii) a Comparing the first intermediate precision with the corresponding second intermediate precision, taking the representative day as a basic unit, selecting the support vector machine/neural network corresponding to the smaller of the first intermediate precision and the second intermediate precision of the day as a load prediction model of the representative day, adopting the same class of days as the representative day and representing the daysAnd (4) a load prediction model of the same day, thereby obtaining a load prediction model for each day.
The plurality of meteorological parameters in the meteorological data include a maximum temperature, a minimum temperature, an average temperature, a relative humidity, and a rainfall.
The multiple load parameters in the load data comprise daily maximum load, daily minimum load, daily peak-valley difference and daily average load.
In the specific implementation of the invention, the prediction can be carried out according to the day and the month.
The daily prediction means that the load condition of the same day in the future is predicted by using the data of the same day every year in the historical data. For example, using load data of each day such as 11 days 1 month in 10 years, 11 days 1 month in 11 years, 11 days 1 month in 12 years, 11 days 1 month in 13 years, and 11 days 1 month in 14 years, by analogy with the above method, for predicting 11 days 1 month in 15 years, 11 days 12 months in 1 month, and the like,
the monthly prediction means that the daily load of the month to be predicted is predicted in chronological order based on the historical load data of the year before the month to be predicted, and for example, the total data of 10 years and 1 month, the total data of 11 years and 1 month, the total data of 12 years and 1 month, the total data of 13 years and 1 month and the total data of 14 years and 1 month are used to predict the total data of 15 years and 1 month.
In the invention, the four load parameters are selected to accurately show the approximate change condition of the load in one day, and the five meteorological factors are selected to approximately reflect the change characteristic of the meteorological phenomena in one day.
In specific implementation, the weather data used is more than the load data for one month without load data, and the weather data is used as a time period to be predicted.
The invention has the beneficial effects that:
compared with the prior art, the method fully considers the influence of meteorological data on load fluctuation, fully considers the scale of the data, simultaneously reduces the load and the meteorological data quantity through a clustering algorithm and a principal component analysis method, and ensures the prediction precision of a support vector machine and a wavelet neural network model through a defined first/second prediction precision calculation formula, thereby improving the prediction precision and solving the problem of low prediction precision caused by the consideration of meteorological factors.
The invention greatly reduces meteorological data and load parameters through cluster analysis and principal component analysis, thereby relieving huge data volume caused by considering meteorological factors, ensuring the prediction precision of a support vector machine and a wavelet neural network model through a defined first/second prediction precision calculation formula, and selecting a model with higher precision from two prediction models as a final prediction model of a day by taking the day as a unit, thereby realizing high precision while greatly improving the prediction efficiency and solving the problem of low prediction precision caused by considering meteorological factors.
Drawings
FIG. 1 is a logic diagram of the method of the present invention.
Fig. 2 is a diagram of an embodiment clustering result.
FIG. 3 is a graph comparing the predicted power in 2014 and the actual power obtained according to the support vector machine prediction model in the embodiment.
Fig. 4 is a graph of the comparison of the predicted power in 2015 and the actual power of the wavelet neural network prediction model according to the embodiment.
Detailed Description
The invention is further illustrated by the following figures and examples.
The examples of the invention are as follows:
the first step is as follows: and extracting and obtaining first historical meteorological data, first historical load data, second historical meteorological data and second historical load data from the power grid data.
The power load data (one sampling point every 15min, 96 points per day, and MW in dimension) from 1/2010 to 31/2015 in 12/2015 in a certain area and the meteorological data (highest daily temperature, lowest daily temperature, average daily temperature, relative daily humidity, and daily rainfall) from 1/2010 to 31/2016 in 1/2016 are known. And obtaining load data of the annual daily maximum load, the daily minimum load, the daily peak-valley difference and the daily load rate parameters of the region through statistics.
The first historical meteorological data and the first historical load data form first original data, and the second historical meteorological data and the second historical load data form second original data; the first historical meteorological data is meteorological data of each day of the previous 5 years, the first historical load data is load data of each day of the previous 5 years, the second historical meteorological data is meteorological data of each day of the 6 th year, and the second historical load data is load data of each day of the 6 th year. The weather data of each day is composed of a plurality of weather parameters, and the load data of each day is composed of a plurality of load parameters. The time period to be predicted is from No. 1/2016 to No. 1/31/2016
The second step is that: the load data of each day is clustered by adopting a K-Means clustering algorithm with the load data of each day as a unit, for each clustered class, the load data of the middle day in the class is selected and reserved, the load data of the day which is not positioned in the middle is removed from the original data, the days represented by the days connected by lines can be classified into one class by taking the clustering of month 1 2012 as an example, and as can be seen from figure 2, the numbers of 1 month 5, 13, 16, 20, 25 and 27 in year 2012 are represented days of the month.
The third step: and (3) processing all meteorological data of the meteorological parameters under each representative day by adopting a principal component analysis method to obtain the cumulative contribution rate of each meteorological parameter, selecting the meteorological parameters with the cumulative contribution rate of 80% for reservation, and removing the meteorological parameters with the contribution rate of less than 80% from the meteorological data of each day.
Table 1 shows the weight of each parameter, and it is found from the principal component analysis that the contribution ratio of the highest temperature and the lowest temperature is the highest, and the sum exceeds 0.8, so "the highest temperature and the lowest temperature" are selected as the main meteorological factors to be considered.
Table 1: weight occupied by each meteorological factor
| Weight of | Maximum temperature | Minimum temperature | Mean temperature | Relative humidity | Amount of rainfall |
| Rate of contribution | 0.651 | 0.2182 | 0.1236 | 0.0066 | 0.0006 |
The fourth step: establishing a load prediction model by adopting a support vector machine, inputting standardized first historical meteorological data and first historical load data into the support vector machine for training by adopting daily prediction to obtain a trained load prediction model, then inputting second historical meteorological data into the trained support vector machine for prediction to obtain first predicted load data, verifying prediction accuracy of the first predicted load data and the second historical load data, and calculating first prediction accuracy E1。
Further, if the first prediction accuracy E1And if the prediction accuracy is more than or equal to 90%, the prediction accuracy of the load prediction model of the support vector machine is considered to meet the requirement, and the next step is carried out. Otherwise, adjusting the relevant parameters until the prediction precision requirement is met.
The predicted result is shown in fig. 3, taking the first 60 representative days of the highest load data in 2014 as an example.
As can be seen from the above figure, the predicted data change trend in 2014 is basically consistent with the change trend in actual 2014, which indicates that the predicted data are consistent with the actual data and have high reliability. And the first prediction precision is 96%, which shows that the prediction precision is very high and the actual goodness of fit is high.
The fifth step: adopting monthly prediction, taking the standardized first historical meteorological data as an input layer, taking the standardized first historical load data as an output layer, and adopting a wavelet neural network for training; after the training is finished, inputting second historical meteorological data into the trained neural network, and outputting to obtain second predicted load data; verifying the prediction accuracy of the second predicted load data and the second historical load data, and calculating a second prediction accuracy E2。
Further, if the second prediction accuracy E2If the prediction accuracy of the wavelet neural network is more than or equal to 90 percent, the wavelet neural network is considered to meet the requirement, and the next step is carried out. Otherwise, adjusting the wavelet function expansion factor, the translation factor, the network connection weight, the network connection threshold value and the network learning rate, and retraining the neural network until the prediction precision requirement is met.
The predicted data of the wavelet neural network is shown in fig. 4, taking the data of the highest load of 2015 year as an example.
As can be seen from the above figure, the predicted data change trend in 2015 is basically consistent with the change trend in actual 2015, which shows that the predicted data are consistent with actual data and have high reliability. And the first prediction precision is 98.4%, which shows that the prediction precision is very high and the actual goodness of fit is high.
The specific implementation of the method is that the data load in neural network prediction can be greatly reduced and the prediction efficiency is accelerated by two algorithms of a clustering algorithm and a principal component analysis and the simplified processing of meteorological data and load data is considered at the same time, and the final prediction precision can be known. And the prediction accuracy of the support vector machine and the wavelet neural network model is ensured through the defined first/second prediction accuracy calculation formula.
And a sixth step: calculating the fourth step passing branchFirst intermediate precision D between first predicted load data obtained by a support vector machine and second historical load data obtained by a neural network on each load parameter for each representative day of the year1Calculating a second intermediate precision D between the second predicted load data and the second historical load data obtained by the neural network in the fifth step on each load parameter of each day2(ii) a And comparing the first intermediate precision with the corresponding second intermediate precision, taking the representative day as a basic unit, selecting the support vector machine/neural network corresponding to the smaller of the first intermediate precision and the second intermediate precision of the day as a load prediction model of the representative day, and adopting the same load prediction model as the representative day for the days belonging to the same class as the representative day, thereby obtaining the load prediction model for each day.
In the sixth step of the embodiment, the unit of day is used, and a model with higher precision is selected from the two prediction models as the final prediction model of the day, so that the prediction efficiency is greatly improved, the high precision is realized, and the problem of low prediction precision caused by considering meteorological factors is solved.
The prediction results of the average loads of the time periods to be predicted from No. 11 to No. 17 are shown in Table 2:
table 2: prediction results of average loads of No. 11-17 of time period to be predicted
The prediction model results provided by the invention are compared with the neural network prediction model considering all meteorological and load data in terms of time and prediction accuracy, and as shown in the following table 3, the prediction data are average load data from No. 1/1 in 2015 to No. 15 in 2015.
TABLE 3
| Algorithm | Average daily prediction accuracy E | Calculating time (second unit) |
| The prediction model provided by the invention | 97.4％ | 16.88 |
| Neural network prediction model | 83.2％ | 43.16 |
Therefore, the power grid meteorological load prediction method based on the wavelet neural network and the support vector machine shows good metering efficiency and prediction accuracy in both prediction accuracy and calculation time.
Claims (7)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810359331.2A CN108596242B (en) | 2018-04-20 | 2018-04-20 | Prediction method of power grid meteorological load based on wavelet neural network and support vector machine |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810359331.2A CN108596242B (en) | 2018-04-20 | 2018-04-20 | Prediction method of power grid meteorological load based on wavelet neural network and support vector machine |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108596242A CN108596242A (en) | 2018-09-28 |
| CN108596242B true CN108596242B (en) | 2021-03-23 |
Family
ID=63614224
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810359331.2A Active CN108596242B (en) | 2018-04-20 | 2018-04-20 | Prediction method of power grid meteorological load based on wavelet neural network and support vector machine |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108596242B (en) |
Families Citing this family (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109345027B (en) * | 2018-10-25 | 2021-11-23 | 国网江苏省电力有限公司盐城供电分公司 | Micro-grid short-term load prediction method based on independent component analysis and support vector machine |
| CN112418476A (en) * | 2019-08-23 | 2021-02-26 | 武汉剑心科技有限公司 | Ultra-short-term power load prediction method |
| CN110617927B (en) * | 2019-09-20 | 2022-04-05 | 长安大学 | Structural settlement deformation prediction method based on EMD-SVR-WNN |
| CN110826789B (en) * | 2019-10-30 | 2023-06-06 | 深圳市康必达控制技术有限公司 | Power load prediction method and device based on power system and terminal equipment |
| CN110991638B (en) * | 2019-11-29 | 2024-01-05 | 国网山东省电力公司聊城供电公司 | Generalized load modeling method based on clustering and neural network |
| CN112215459A (en) * | 2020-09-02 | 2021-01-12 | 南方电网能源发展研究院有限责任公司 | Power distribution method and device based on power grid investment scale prediction |
| CN112418533B (en) * | 2020-11-25 | 2024-09-17 | 江苏电力交易中心有限公司 | A clean energy electricity decomposition prediction method |
| CN115051864B (en) * | 2022-06-21 | 2024-02-27 | 郑州轻工业大学 | PCA-MF-WNN-based network security situation element extraction method and system |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005122517A (en) * | 2003-10-17 | 2005-05-12 | Fuji Electric Holdings Co Ltd | Energy demand forecasting method, energy demand forecasting apparatus, energy demand forecasting program, and recording medium |
| CN102982387A (en) * | 2012-10-18 | 2013-03-20 | 安徽工程大学 | Method for predicting short-term power load |
| CN103218675A (en) * | 2013-05-06 | 2013-07-24 | 国家电网公司 | Short-term load prediction method based on clustering and sliding window |
| CN104008430A (en) * | 2014-05-29 | 2014-08-27 | 华北电力大学 | Method for establishing virtual reality excavation dynamic smart load prediction models |
| CN104318332A (en) * | 2014-10-29 | 2015-01-28 | 国家电网公司 | Power load predicting method and device |
| CN106208388A (en) * | 2016-08-31 | 2016-12-07 | 科大智能电气技术有限公司 | A kind of intelligence charging system and short term basis load prediction implementation method thereof in order |
| CN106410781A (en) * | 2015-07-29 | 2017-02-15 | 中国电力科学研究院 | Power consumer demand response potential determination method |
-
2018
- 2018-04-20 CN CN201810359331.2A patent/CN108596242B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005122517A (en) * | 2003-10-17 | 2005-05-12 | Fuji Electric Holdings Co Ltd | Energy demand forecasting method, energy demand forecasting apparatus, energy demand forecasting program, and recording medium |
| CN102982387A (en) * | 2012-10-18 | 2013-03-20 | 安徽工程大学 | Method for predicting short-term power load |
| CN103218675A (en) * | 2013-05-06 | 2013-07-24 | 国家电网公司 | Short-term load prediction method based on clustering and sliding window |
| CN104008430A (en) * | 2014-05-29 | 2014-08-27 | 华北电力大学 | Method for establishing virtual reality excavation dynamic smart load prediction models |
| CN104318332A (en) * | 2014-10-29 | 2015-01-28 | 国家电网公司 | Power load predicting method and device |
| CN106410781A (en) * | 2015-07-29 | 2017-02-15 | 中国电力科学研究院 | Power consumer demand response potential determination method |
| CN106208388A (en) * | 2016-08-31 | 2016-12-07 | 科大智能电气技术有限公司 | A kind of intelligence charging system and short term basis load prediction implementation method thereof in order |
Non-Patent Citations (3)
| Title |
|---|
| An improved load forecast model using factor analysis: An Australian case study;Xi Chen 等;《Proceeding of the 2017 IEEE International Conference on Information and Automation(ICIA) Macau SAR,China 》;20170731;第903-908页 * |
| 基于知识挖掘技术的智能协同电力负荷预测研究;王建军;《中国博士学位论文全文数据库-工程科技Ⅱ辑》;20110915(第9期);第C042-36页第1.2.1,第3.4.6节，表5-1 * |
| 电力系统负荷预测研究综述与发展方向的探讨;康重庆 等;《电力系统自动化》;20040910;第28卷(第17期);第1-11页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108596242A (en) | 2018-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108596242B (en) | Prediction method of power grid meteorological load based on wavelet neural network and support vector machine | |
| CN108229742B (en) | A Load Forecasting Method Based on Meteorological Data and Data Trends | |
| CN103559655B (en) | The Forecasting Methodology of the novel feeder line load of microgrid based on data mining | |
| CN113505923B (en) | A method and system for short-term load forecasting of regional power grid | |
| CN111552923A (en) | A load forecasting method and load forecasting system based on general distribution | |
| CN106408223A (en) | Short-term load prediction based on meteorological similar day and error correction | |
| CN111695731A (en) | Load prediction method, system and equipment based on multi-source data and hybrid neural network | |
| CN112418485A (en) | Household load prediction method and system based on load characteristics and power consumption behavior mode | |
| CN113255900A (en) | Impulse load prediction method considering improved spectral clustering and Bi-LSTM neural network | |
| CN112421631A (en) | A method and system for evaluating new energy consumption capacity | |
| CN113379116A (en) | Cluster and convolutional neural network-based line loss prediction method for transformer area | |
| CN118970924A (en) | A short-term load forecasting method and system based on LSTM-DNN hybrid neural network | |
| CN114091776A (en) | K-means-based multi-branch AGCNN short-term power load prediction method | |
| CN117273077A (en) | CNN-LSTM-based central air conditioner demand response testing method and system | |
| CN111612244A (en) | A day-ahead photovoltaic power nonparametric probability prediction method based on QRA-LSTM | |
| CN118396036A (en) | Wind power prediction method, system, equipment and medium | |
| CN118312723A (en) | A slope displacement prediction method and system based on monitoring data | |
| CN108615091B (en) | Electric power meteorological load data prediction method based on cluster screening and neural network | |
| CN116307250A (en) | A short-term load forecasting method and system based on typical daily feature selection | |
| CN115952895A (en) | A wind power device power prediction method, system and storage medium | |
| CN103854073A (en) | Method for comprehensively predicting generation capacity of multi-radial flow type small hydropower station group area | |
| CN110175705B (en) | Load prediction method and memory and system comprising same | |
| CN108665090B (en) | Prediction method of saturated load of urban power grid based on principal component analysis and Verhulst model | |
| Deng et al. | Medium-term rolling load forecasting based on seasonal decomposition and long short-term memory neural network | |
| CN115796000A (en) | Short-term air temperature forecast set correction method based on stacked machine learning 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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |