CN109886455A - Very Short-Term Load Forecasting Method under thunder and lightning weather - Google Patents
Very Short-Term Load Forecasting Method under thunder and lightning weather Download PDFInfo
- Publication number
- CN109886455A CN109886455A CN201910031331.4A CN201910031331A CN109886455A CN 109886455 A CN109886455 A CN 109886455A CN 201910031331 A CN201910031331 A CN 201910031331A CN 109886455 A CN109886455 A CN 109886455A
- Authority
- CN
- China
- Prior art keywords
- load
- thunder
- curve
- rainfall
- lightning
- 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
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention relates to very Short-Term Load Forecasting Methods under a kind of thunder and lightning weather comprising preprocess method and real-time predicting method.Preprocess method is used for load data and temperature data and rainfall product data using the load data and temperature data and rainfall product data, existing power grid of power grid under existing non-thunder and lightning weather under thunder and lightning weather, obtains load difference and temperature difference, the relational expression of rainfall difference.Real-time predicting method then on the basis of carrying out ultra-short term to power grid using load data stream and obtain basic forecast curve, is superimposed the corresponding load variations amount of thunder and lighting process that will occur, to obtain ultra-short term curve under thunder and lightning weather.The present invention is based on the online cutting techniques of data flow, influence of the variation of change in weather bring temperature and rainfall to load variations is considered simultaneously, the load prediction results of Changes in weather violent period are modified, the accuracy of the load prediction under the extreme conditions such as thunder and lightning is improved.
Description
Technical field
The invention belongs to load forecast fields, and in particular to very Short-Term Load Forecasting Method under a kind of thunder and lightning weather.
Background technique
As electric system is to intelligence, networking, marketing development, it is desirable that, can be fast when system loading changes
The unit output in each power plant in speed and adaptive adjustment power grid, therefore, a kind of predicted time is shorter, predetermined speed faster, in advance
Survey the prediction of accuracy higher ultra-short term real-time load by there is an urgent need to.Meanwhile under the conditions of thunder and lightning, load is by thunder and lightning weather shadow
Sound is larger, and load fluctuation is more violent, so that the accuracy of load prediction substantially reduces.
Currently, these intelligent algorithms of neural network, wavelet analysis are mainly used in terms of ultra-short term,
Such as it is existing based on the very Short-Term Load Forecasting Method for improving data flow, more accurate ultra-short term may be implemented.
However, above-mentioned intelligent algorithm is difficult to adapt under the conditions of thunder and lightning.The reason is as follows that: 1. due to the existing load data of thunderstorm weather
It is less, it is difficult to meet requirement of the above-mentioned intelligent algorithm to data volume；2. due to the arriving of thunder and lightning weather, can make load fluctuation compared with
Greatly, above-mentioned intelligent algorithm does not consider influence of the meteorologic factor to load usually, reduces the forecasting accuracy in change in weather；
3. since ultra-short term requires implementation and accuracy simultaneously, and above-mentioned intelligent algorithm real-time and accuracy often phase
It mutually influences mutually to restrict, there are intrinsic Contradictory constrainings, it is difficult to realize while improve；4. since load data itself has centainly
Changing rule, above-mentioned intelligent algorithm is excessively intended to " mathematicization ", excessively focuses on the improvement to algorithm, has ignored load data
Changing rule itself loses the characteristic of electric system.
Summary of the invention
The object of the present invention is to provide the variations of a kind of consideration change in weather bring temperature and rainfall to load variations
Influence improve under the extreme conditions such as thunder and lightning to be modified to the load prediction results of Changes in weather violent period
Very Short-Term Load Forecasting Method under the thunder and lightning weather of load prediction accuracy.
In order to achieve the above objectives, the technical solution adopted by the present invention is that:
Very Short-Term Load Forecasting Method under a kind of thunder and lightning weather, it is ultrashort for being carried out to power grid under thunder and lightning weather condition
Phase load prediction, very Short-Term Load Forecasting Method includes preprocess method and real-time predicting method under the thunder and lightning weather；
The preprocess method the following steps are included:
Step 1.1: utilizing the load data and temperature data and rainfall of the power grid under existing non-thunder and lightning weather
Data obtain the basic load curve of the power grid and its corresponding basal temperature curve and basic rainfall under non-thunder and lightning weather
Curve；
Step 1.2: by the basic load curve of the power grid under non-thunder and lightning weather and its corresponding basal temperature curve and
Basic rainfall curve is in conjunction with load data of the existing power grid under thunder and lightning weather and temperature data and rainfall number
According to obtaining, multiple groups load difference, temperature difference and rainfall are poor；
Step 1.3: being fitted to obtain load difference and temperature difference, rainfall using multiple groups load difference, temperature difference and rainfall difference
The relational expression of difference；
The real-time predicting method the following steps are included:
Step 2.1: using the forecast information in weather forecast to the thunder and lighting process that will occur, including thunder and lightning start it is latter
Temperature data and rainfall product data in the section time and after thunder and lightning in a period of time, in conjunction with the basis temperature under non-thunder and lightning weather
It writes music line and basic rainfall curve, obtains temperature difference in the thunder and lighting process to be occurred and rainfall is poor, and will send out
Temperature difference and rainfall difference band in raw thunder and lighting process enter load difference and temperature difference, the relational expression of rainfall difference and obtain to want
The corresponding load variations amount of the thunder and lighting process of generation；
Step 2.2: ultra-short term is carried out to power grid using load data stream and obtains basic forecast curve, and
On the basis of the basic forecast curve, it is superimposed the corresponding load variations amount of thunder and lighting process that will occur, to obtain thunder and lightning
Ultra-short term curve under weather.
Preferably, in the step 1.1, the typical day under multiple non-thunder and lightning weather is chosen, by each typical day
Load curve, temperature curve, rainfall curve are averaged and the basic load as the power grid under non-thunder and lightning weather respectively
Curve and its corresponding basal temperature curve and basic rainfall curve.
Preferably, it the typical day chooses temperature in the Spring Festival and autumn and is lower than 20 DEG C and multiple dates without thunderstorm weather.
Preferably, in the step 1.3, the load difference is obtained using multilinear fitting and temperature difference, rainfall are poor
Relational expression.
Preferably, in the step 2.1,25 minutes temperature datas after starting latter 25 minutes and thunder and lightning using thunder and lightning
And rainfall product data.
Preferably, in the step 2.2, ultra-short term is carried out to power grid using load data stream and obtains basis
The method of prediction curve are as follows: the load of the power grid is carried out real-time using data flow is met described in the segmentation of segmented linear method
Trend shift analysis, to determine corresponding piecewise prediction model according to different load trends and divide described in adjacent two sections
The cut-point of prediction model, and then obtain the basic forecast curve.
Preferably, in the step 2.2, ultra-short term is carried out to power grid using load data stream and obtains basis
The method of prediction curve includes following sub-step:
Sub-step 2.2.1: using first n minutes in the period to be predicted of actual load data, using linear regression method
It treats predicted time and carries out load prediction, obtain tentative prediction as a result, and using the tentative prediction result as when the last period is pre-
Survey model；
Sub-step 2.2.2: the bottom threshold p of default deviation accumulation valueh1With upper threshold ph2, then circulation executes sub-step
Rapid 2.2.3 to sub-step 2.2.4；
Sub-step 2.2.3: acquisition current time is corresponding when the last period prediction model, by the actual load number of the power grid
It is compared, and is obtained when the corresponding each secondary comparison of the last period prediction model according to the prediction data worked as in the last period prediction model
Cumulative errors C (t), t is sampling instant；
Step 2.2.4: judge the cumulative errors C (t) and the bottom threshold ph1, the upper threshold ph2Size
Relationship:
When | C (t) | < ph1When, it continues with when the last period prediction model is predicted；
Work as ph1≤ | C (t) | < ph2When, it continues with and is predicted when the last period prediction model, and store sampling instant t
And its corresponding actual negative charge values y (t)；
Work as ph2≤ | C (t) | when, using next sampling instant t+ Δ t as the cut-point, and utilize ph1≤ | C (t) | <
ph2When each actual negative charge values y (t) for storing, load prediction is carried out using linear regression method, obtains next section of prediction model, and
Cumulative errors C (t) is reset to 0, the actual negative charge values of storage are reset.
Preferably, the actual load of the power grid is sampled according to 5 minutes sampling intervals and realizes and utilizes load
Data flow carries out ultra-short term to power grid and obtains basic forecast curve.
Preferably, it in the sub-step 2.2.1, is carried out using first 30 minutes in the period to be predicted actual load data
Load prediction.
Preferably, load prediction is carried out using least square method and obtains prediction model.
Due to the above technical solutions, the present invention has the following advantages over the prior art: the present invention is based on data
Online cutting techniques are flowed, while considering influence of the variation of change in weather bring temperature and rainfall to load variations, to day
The load prediction results that gas changes the violent period are modified, and improve the accurate of under the extreme conditions such as thunder and lightning load prediction
Property.
Specific embodiment
The present invention will be further described below with reference to examples.
A kind of embodiment one: thunder and lightning weather that ultra-short term is carried out to power grid under thunder and lightning weather condition
Lower very Short-Term Load Forecasting Method, including preprocess method and real-time predicting method two parts.
One, preprocess method
Preprocess method the following steps are included:
Step 1.1: utilizing the load data and temperature data and rainfall product data of power grid under existing non-thunder and lightning weather
Obtain the basic load curve of power grid and its corresponding basal temperature curve and basic rainfall curve under non-thunder and lightning weather.
Specifically, choosing the typical day under multiple non-thunder and lightning weather first, typical day the Spring Festival and autumn medium temperature can be chosen
Degree is lower than 20 DEG C and multiple dates without thunderstorm weather.Such as China's most area, the distribution for observing thunderstorm weather can
Know, thunderstorm weather is typically distributed across the 6-9 month, therefore can generally choose temperature in April and 10 months and be lower than 20 DEG C and without thundery sky
Certain day of gas, as spring or the typical day in autumn.To choose typical day in a Spring Festival and an autumn in the present embodiment
It is illustrated for typical day.
It chooses a lot of typical case in the future, load curve, temperature curve, the rainfall curve of each typical day is made even respectively
Mean value and basic load curve and its corresponding basal temperature curve as power grid under non-thunder and lightning weather and basic rainfall are bent
Line.If the load curve of typical case's day spring is YA, temperature curve TA, rainfall curve is PA(value is 0), if typical case's day autumn
Load curve be YB, temperature curve TB, rainfall curve is PB(value is 0), then by the load curve in spring, typical day in autumn
Temperature curve, rainfall curve are averaged respectively:
It is electric under as non-thunder and lightning weather
The basic load curve of net,For corresponding basal temperature curve,For corresponding basic rainfall curve.
Step 1.2: by the basic load curve of power grid under non-thunder and lightning weather and its corresponding basal temperature curve and basis
It is more that rainfall curve combines load data and temperature data and rainfall product data of the existing power grid under thunder and lightning weather to obtain
Group load difference, temperature difference and rainfall are poor.
It is corresponding negative with the basis of same time period using the load under existing thunder and lightning weather, temperature, rainfall product data
Lotus, temperature, rainfall make the difference, and it is poor to obtain multiple groups load difference, temperature difference, rainfall, i.e., [Δ Y, Δ T, Δ P].
Step 1.3: being fitted to obtain load difference and temperature difference, rainfall using multiple groups load difference, temperature difference and rainfall difference
The relational expression of difference.
In the step, load difference and temperature difference, the relational expression of rainfall difference are obtained using multilinear fitting
Δ Y=a Δ T+b Δ P (1-4)
Wherein, a, b are respectively fitting coefficient.
Two, real-time predicting method
Real-time predicting method the following steps are included:
Step 2.1: using the forecast information in weather forecast to the thunder and lighting process that will occur, including thunder and lightning start it is latter
In the section time after (such as 25 minutes, 5 sampled points) and thunder and lightning in a period of time (such as 25 minutes, 5 sampled points) temperature
Data and rainfall product data obtain occurring in conjunction with the basal temperature curve and basic rainfall curve under non-thunder and lightning weather
Thunder and lighting process in temperature difference Δ T and rainfall difference Δ P, and by the temperature difference Δ T and drop in the thunder and lighting process that will occur
Rainfall difference Δ P is brought into load difference and temperature difference, the relational expression (1-4) of rainfall difference, obtains the thunder and lighting process pair to be occurred
The load variations amount Δ Y answered.The corresponding load variations amount Δ Y of the thunder and lighting process that will occur is in subsequent progress load prediction
As correction value.The reason of being modified for 25 minutes after 25 minutes are wherein chosen after thunder and lightning starts with thunder and lightning is: at this section
Between be the most violent point of Changes in weather, load variations are maximum.
Step 2.2: ultra-short term is carried out to power grid using load data stream and obtains basic forecast curve, and
On the basis of the basic forecast curve, it is superimposed the corresponding load variations amount Δ Y of thunder and lighting process that will occur, to obtain thunder and lightning
Ultra-short term curve under weather.
In the step 2.2, ultra-short term is carried out to power grid using load data stream and obtains basic forecast curve
Method are as follows: data flow is met using the segmentation of segmented linear method and real-time tendency shift analysis is carried out to the load of power grid, fastly
Speed detects appropriate load trend cut-point (hereinafter referred to as cut-point), to be determined according to different load trends corresponding
The cut-point of piecewise prediction model and the adjacent two sections of prediction models of segmentation, and then obtain basic forecast curve.Cut-point be
The point for separating future load data and available data when load trend changes greatly forms one between adjacent two cut-point
A trend data section, same trend section obey the least-squares linear regression prediction model of identical parameters, and different trend sections are established
Corresponding prediction model carries out real-time segmentation prediction.
This method is existing method, is realized by following sub-step:
Sub-step 2.2.1: it using the actual load data of n minutes first (such as 30 minutes) in the period to be predicted, uses
Least square method linear approximation treats predicted time and carries out load prediction, obtains tentative prediction as a result, and by tentative prediction result
As when the last period prediction model.
In the implementation process of this method, by the way that (such as according to 5 minutes sampling intervals, i.e. whole day has 288 altogether at equal intervals
Load sampled point, these sampled points, that is, subsequent load prediction point) it samples to be sampled to the actual load of power grid.Then in son
In step 2.2.1, using first 30 minutes in the period to be predicted actual load data (totally 6 sampled points), to carry out minimum
Square law linear approximation.
Sub-step 2.2.2: the bottom threshold p of default deviation accumulation valueh1With upper threshold ph2, then circulation executes sub-step
Rapid 2.2.3 to sub-step 2.2.4.
Sub-step 2.2.3: acquisition current time t is corresponding when the last period prediction model, by the actual load data of power grid
It is compared, and is obtained when the corresponding each secondary comparison of the last period prediction model with the prediction data worked as in the last period prediction model
Cumulative errors C (t), t are sampling instant.
Specifically, prediction result is compared with the actual load result of newest monitoring, cumulative errors C (t) is obtained, such as
Formula (2-1):
Wherein, t is current sample time；Δ t is the sampling interval, and the present invention takes 5min, i.e. it is pre- to have 288 loads altogether for whole day
Measuring point；The prediction error value e (t) of t moment=y (t)-y*(t)；Y (t) is the load actual value of t moment, y*It (t) is t moment
Predicted load；C (t) is the prediction deviation accumulation value of t moment；T is corresponding to a cut-point nearest from current time
Time；N is since a upper cut-point to the load number of samples current time, i, n ∈ N.
Step 2.2.4: judge cumulative errors C (t) and bottom threshold ph1, upper threshold ph2Size relation: according to error
Aggregate-value C (t) is measured in real time and judges to cut-point (load trend cut-point), when prediction deviation accumulation value C (t) is more than
The upper threshold p of settingh2When, show that the current practical trend of load prediction trend deviation is very remote, which cannot reach
To good prediction result, t+ time Δt is judged as cut-point, and establishes newly pre- using the effective sample data of this trend section
Model is surveyed, the load data of next trend section is predicted in real time.Specific distinguishing rule is as follows:
1. when | C (t) | < ph1When, point t is not cut-point, and current predictive model is suitable for next sampling instant t+ Δ
T, therefore continue with when the last period prediction model is predicted.
2. working as ph1≤ | C (t) | < ph2When, point t is not cut-point, and current predictive model is suitable for next sampling instant t
+ Δ t is then continued with and is predicted when the last period prediction model, and stores sampling instant t and its corresponding actual negative charge values y
(t), the forecast sample data as next load trend section.
3. working as ph2≤ | C (t) | when, using next sampling instant t+ Δ t as cut-point, and utilize ph1≤ | C (t) | < ph2
When each actual negative charge values y (t) for storing, i.e., the forecast sample data of next load trend section are linearly forced using least square method
It is close to carry out load prediction, next section of prediction model is obtained, and cumulative errors C (t) is reset to 0, by the actual negative charge values of storage
It resets.
Sub-step 2.2.3 is returned after having executed above-mentioned judgement.When meeting criterion 3., predicted in real time into new one piece of data stream
Stage obtains when then executing sub-step 2.2.3 again when the last period prediction model as executes sub-step 2.2.4 before this
Prediction model.
Pass through the above sub-step 2.2.1 to sub-step 2.2.4, it is already possible to carry out ultra-short term, however only examine
Load data itself is considered, has not considered the influences of the meteorologic factors to load such as temperature, rainfall, therefore has also needed to utilize meteorology
The temperature of office's prediction and the load data of these data corrections of rainfall prediction, i.e., on the basis of basic forecast curve, superposition
The corresponding load variations amount Δ Y of the thunder and lighting process that will occur obtained in step 2.1, to obtain under complete thunder and lightning weather
Ultra-short term curve.
Least square method linear approximation involved in above method, illustrate as follows: load data stream is divided in advance online
The update for surveying model parameter depends on the Real-time segmentation of data flow.Before the appearance of new cut-point, for newly arrived data, only
It is substituted into and carries out error calculation and storage in established model, without being updated calculating to prediction model parameters.When new
When cut-point occurs, prediction model parameters are then adjusted according to new forecast sample data in real time.If upper trend section load
Predict that value sequence isThe practical value sequence of load isWhen corresponding
Between sequence beJ indicates the trend section serial number where data, njThe load for indicating that jth section includes is adopted
Sample data amount check, j, nj∈N+.It can be obtained according to Linear Regression Forecasting Model, the predicted value of new trend section load data is
Wherein, aj+1With bj+1The prediction model parameters for indicating+1 section of load data of jth solve formula using least square method
Unknown parameter a in (3-1)j+1With bj+1It can be obtained predicted load.Enable error sum of squares
In time tj,i(i=1,2 ..., nj) variation when have a minimum value, that is, meet:
So as to solve the prediction model parameters in formula (9-1):
The present invention is based on the online cutting techniques of data flow, ultra-short term is divided into multiple periods and is carried out respectively in advance
It surveys, while utilizing existing less thunderstorm weather load data, consider the variation pair of change in weather bring temperature and rainfall
The influence of load variations is modified the load prediction results of Changes in weather violent period, improves in the extreme item such as thunder and lightning
The accuracy of load prediction under part.Its advantage is characterized in particular in:
1. calculation amount is small, can satisfy ultrashort since the present invention need to only use less load, temperature, rainfall product data
Requirement of the phase load prediction to real-time.
2. since the present invention is to be predicted according to nearest historical data and have cumulative errors threshold value, it is contemplated that load sheet
The changing rule of body has very high accuracy.
3. it is corrected since the present invention arrives to thunder and lightning with the load prediction results of thunder and lightning finish time, it can be significantly
The accuracy of load prediction, can be well adapted for the extreme climates such as thunder and lightning when increasing change in weather.
In the solution of the present invention, in step 2.2 can also using other linear regression methods replace least square method into
Row prediction, can achieve same or similar result.Phase can be reached using the choosing method of other typical days in step 1.1
Same or similar result.
The above embodiments merely illustrate the technical concept and features of the present invention, and its object is to allow person skilled in the art
Scholar cans understand the content of the present invention and implement it accordingly, and it is not intended to limit the scope of the present invention.It is all according to the present invention
Equivalent change or modification made by Spirit Essence, should be covered by the protection scope of the present invention.
Claims (10)
1. very Short-Term Load Forecasting Method under a kind of thunder and lightning weather, for carrying out ultra-short term to power grid under thunder and lightning weather condition
Load prediction, it is characterised in that: very Short-Term Load Forecasting Method includes preprocess method and prediction in real time under the thunder and lightning weather
Method；
The preprocess method the following steps are included:
Step 1.1: utilizing the load data and temperature data and rainfall product data of the power grid under existing non-thunder and lightning weather
Obtain the basic load curve of the power grid and its corresponding basal temperature curve and basic rainfall curve under non-thunder and lightning weather；
Step 1.2: by the basic load curve of the power grid under non-thunder and lightning weather and its corresponding basal temperature curve and basis
Rainfall curve is obtained in conjunction with load data of the existing power grid under thunder and lightning weather and temperature data and rainfall product data
It is poor to multiple groups load difference, temperature difference and rainfall；
Step 1.3: being fitted to obtain load difference and temperature difference, rainfall difference using multiple groups load difference, temperature difference and rainfall difference
Relational expression；
The real-time predicting method the following steps are included:
Step 2.1: using the forecast information in weather forecast to the thunder and lighting process that will occur, including when thunder and lightning starts latter section
Temperature data and rainfall product data after interior and thunder and lightning in a period of time, it is bent in conjunction with the basal temperature under non-thunder and lightning weather
Line and basic rainfall curve, obtain temperature difference in the thunder and lighting process to be occurred and rainfall is poor, and will will occur
Temperature difference and rainfall difference band in thunder and lighting process enter load difference and temperature difference, the relational expression of rainfall difference and obtain occurring
The corresponding load variations amount of thunder and lighting process；
Step 2.2: ultra-short term being carried out to power grid using load data stream and obtains basic forecast curve, and described
On the basis of basic forecast curve, it is superimposed the corresponding load variations amount of thunder and lighting process that will occur, to obtain thunder and lightning weather
Lower ultra-short term curve.
2. very Short-Term Load Forecasting Method under thunder and lightning weather according to claim 1, it is characterised in that: the step 1.1
In, the typical day under multiple non-thunder and lightning weather is chosen, load curve, temperature curve, the rainfall of each typical day is bent
Line is averaged and basic load curve and its corresponding basal temperature curve as the power grid under non-thunder and lightning weather respectively
With basic rainfall curve.
3. very Short-Term Load Forecasting Method under thunder and lightning weather according to claim 1, it is characterised in that: typical case's day choosing
Temperature in the Spring Festival and autumn is taken to be lower than 20 DEG C and multiple dates without thunderstorm weather.
4. very Short-Term Load Forecasting Method under thunder and lightning weather according to claim 1, it is characterised in that: the step 1.3
In, the load difference and temperature difference, the relational expression of rainfall difference are obtained using multilinear fitting.
5. very Short-Term Load Forecasting Method under thunder and lightning weather according to claim 1, it is characterised in that: the step 2.1
In, 25 minutes temperature datas and rainfall product datas after starting latter 25 minutes and thunder and lightning using thunder and lightning.
6. very Short-Term Load Forecasting Method under thunder and lightning weather according to claim 1, it is characterised in that: the step 2.2
In, ultra-short term is carried out and the method that obtains basic forecast curve to power grid using load data stream are as follows: using segmentation
Meet data flow described in linear method segmentation and real-time tendency shift analysis is carried out to the load of the power grid, thus according to difference
Load trend determine the cut-point of corresponding piecewise prediction model and the adjacent two sections of prediction models of segmentation, and then obtain
The basic forecast curve.
7. very Short-Term Load Forecasting Method under thunder and lightning weather according to claim 6, it is characterised in that: the step 2.2
In, the method for carrying out ultra-short term to power grid using load data stream and obtaining basic forecast curve includes following sub-step
It is rapid:
Sub-step 2.2.1: it using first n minutes in the period to be predicted of actual load data, is treated using linear regression method
Predicted time carries out load prediction, obtains tentative prediction as a result, and predicting mould using the tentative prediction result as when the last period
Type；
Sub-step 2.2.2: the bottom threshold p of default deviation accumulation valueh1With upper threshold ph2, then circulation executes sub-step
2.2.3 to sub-step 2.2.4；
Sub-step 2.2.3: it is corresponding when the last period prediction model to obtain current time, by the actual load data of the power grid with
When the prediction data in the last period prediction model is compared, and obtain tired when the corresponding each secondary comparison of the last period prediction model
It counts error C (t), t is sampling instant；
Step 2.2.4: judge the cumulative errors C (t) and the bottom threshold ph1, the upper threshold ph2Size relation:
When | C (t) | < ph1When, it continues with when the last period prediction model is predicted；
Work as ph1≤ | C (t) | < ph2When, it continues with and is predicted when the last period prediction model, and store sampling instant t and its right
The actual negative charge values y (t) answered；
Work as ph2≤ | C (t) | when, using next sampling instant t+ Δ t as the cut-point, and utilize ph1≤ | C (t) | < ph2When
Each actual negative charge values y (t) of storage carries out load prediction using linear regression method, obtains next section of prediction model, and will tire out
Meter error C (t) resets to 0, and the actual negative charge values of storage are reset.
8. very Short-Term Load Forecasting Method under thunder and lightning weather according to claim 7, it is characterised in that: according to 5 minutes
Sampling interval samples the actual load of the power grid and realizes and carry out super short period load to power grid using load data stream
It predicts and obtains basic forecast curve.
9. very Short-Term Load Forecasting Method under thunder and lightning weather according to claim 7, it is characterised in that: the sub-step
2.2.1 in, load prediction is carried out using first 30 minutes in the period to be predicted actual load data.
10. very Short-Term Load Forecasting Method under thunder and lightning weather according to claim 7, it is characterised in that: using minimum two
Multiplication carries out load prediction and obtains prediction model.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910031331.4A CN109886455B (en) | 2019-01-14 | 2019-01-14 | Method for predicting ultra-short-term load in thunder and lightning climate |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910031331.4A CN109886455B (en) | 2019-01-14 | 2019-01-14 | Method for predicting ultra-short-term load in thunder and lightning climate |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109886455A true CN109886455A (en) | 2019-06-14 |
| CN109886455B CN109886455B (en) | 2020-12-08 |
Family
ID=66925924
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910031331.4A Active CN109886455B (en) | 2019-01-14 | 2019-01-14 | Method for predicting ultra-short-term load in thunder and lightning climate |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109886455B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115798900A (en) * | 2023-01-13 | 2023-03-14 | 江苏凡高电气有限公司 | Intelligent capacity-adjusting transformer convenient to disassemble and assemble |
| CN117670073A (en) * | 2023-10-30 | 2024-03-08 | 国家电网有限公司华东分部 | Prediction model construction method and device based on load air temperature correlation analysis |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102567814A (en) * | 2012-01-31 | 2012-07-11 | 河南省电力公司南阳供电公司 | Method for predicting and calculating cooling load of regional power grid |
| CN104036327A (en) * | 2014-06-20 | 2014-09-10 | 国家电网公司 | Fast bus load forecasting method for smart distribution network |
| CN104156783A (en) * | 2014-07-29 | 2014-11-19 | 广西电网有限责任公司 | Maximum daily load prediction system and method of electric system considering meteorological accumulative effect |
| CN105069525A (en) * | 2015-07-30 | 2015-11-18 | 广西大学 | All-weather 96-point daily load curve prediction and optimization correction system |
-
2019
- 2019-01-14 CN CN201910031331.4A patent/CN109886455B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102567814A (en) * | 2012-01-31 | 2012-07-11 | 河南省电力公司南阳供电公司 | Method for predicting and calculating cooling load of regional power grid |
| CN104036327A (en) * | 2014-06-20 | 2014-09-10 | 国家电网公司 | Fast bus load forecasting method for smart distribution network |
| CN104156783A (en) * | 2014-07-29 | 2014-11-19 | 广西电网有限责任公司 | Maximum daily load prediction system and method of electric system considering meteorological accumulative effect |
| CN105069525A (en) * | 2015-07-30 | 2015-11-18 | 广西大学 | All-weather 96-point daily load curve prediction and optimization correction system |
Non-Patent Citations (2)
| Title |
|---|
| JING LUO: "Short Term Power load Forecasting Considering Meteorological Factors", 《2ND INTERNATIONAL CONFERENCE ON MACHINERY, ELECTRONICS AND CONTROL SIMULATION》 * |
| 苏适 等: "计及气象因素的用电负荷短期分时分类预测模型与方法", 《电力建设》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115798900A (en) * | 2023-01-13 | 2023-03-14 | 江苏凡高电气有限公司 | Intelligent capacity-adjusting transformer convenient to disassemble and assemble |
| CN115798900B (en) * | 2023-01-13 | 2023-05-30 | 江苏凡高电气有限公司 | Intelligent capacity-regulating transformer convenient to disassemble and assemble |
| CN117670073A (en) * | 2023-10-30 | 2024-03-08 | 国家电网有限公司华东分部 | Prediction model construction method and device based on load air temperature correlation analysis |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109886455B (en) | 2020-12-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108846517B (en) | Integration method for predicating quantile probabilistic short-term power load | |
| CN108229754B (en) | Short-term load prediction method based on similar day segmentation and LM-BP network | |
| CN110084367B (en) | Soil moisture content prediction method based on LSTM deep learning model | |
| CN110474339B (en) | Power grid reactive power control method based on deep power generation load prediction | |
| CN111369070B (en) | Multimode fusion photovoltaic power prediction method based on envelope clustering | |
| CN111783953A (en) | 24-point power load value 7-day prediction method based on optimized LSTM network | |
| CN110110912B (en) | Photovoltaic power multi-model interval prediction method | |
| CN111079989B (en) | DWT-PCA-LSTM-based water supply amount prediction device for water supply company | |
| CN102682349A (en) | Electricity consumption intelligent prediction system and method | |
| CN106682763B (en) | Power load optimization prediction method for large amount of sample data | |
| CN109389238B (en) | Ridge regression-based short-term load prediction method and device | |
| CN112149879A (en) | New energy medium-and-long-term electric quantity prediction method considering macroscopic volatility classification | |
| CN110033134A (en) | A kind of short-term load forecasting algorithm of segmentation day by day considering meteorologic factor | |
| CN112085285B (en) | Bus load prediction method, device, computer equipment and storage medium | |
| CN105701562B (en) | Training method, applicable method for predicting generated power and respective system | |
| CN110378504B (en) | Photovoltaic power generation climbing event probability prediction method based on high-dimensional Copula technology | |
| CN103002164A (en) | Telephone traffic forecasting method of electric power call center | |
| CN110969306B (en) | Power distribution low-voltage transformer area load prediction method and device based on deep learning | |
| CN113361195B (en) | Power distribution network line ultra-short term load prediction method based on LSTM and similar days | |
| CN109886455A (en) | Very Short-Term Load Forecasting Method under thunder and lightning weather | |
| CN110909958A (en) | Short-term load prediction method considering photovoltaic grid-connected power | |
| CN112036598A (en) | Charging pile use information prediction method based on multi-information coupling | |
| CN109214458A (en) | A kind of city load quantization method based on historical data | |
| CN104915727B (en) | Various dimensions allomer BP neural network optical power ultra-short term prediction method | |
| CN111882114A (en) | Short-term traffic flow prediction model construction method and prediction method |
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 |