CN107194509B - Wind power prediction method based on time interval fuzzy operator and approximate weight integration - Google Patents
Wind power prediction method based on time interval fuzzy operator and approximate weight integration Download PDFInfo
- Publication number
- CN107194509B CN107194509B CN201710357797.4A CN201710357797A CN107194509B CN 107194509 B CN107194509 B CN 107194509B CN 201710357797 A CN201710357797 A CN 201710357797A CN 107194509 B CN107194509 B CN 107194509B
- Authority
- CN
- China
- Prior art keywords
- data
- wind power
- time point
- group
- approximate weight
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/06—Electricity, gas or water supply
-
- 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
Abstract
The invention discloses a wind power prediction method based on time interval fuzzy operator and approximate weight integration, and relates to a wind power prediction method based on time interval fuzzy operator and approximate weight integration. The invention aims to solve the defects that the existing method has a certain time delay from the time when the blades of the generator are driven by wind power to rotate and finally shows that the output power changes, the data collected from the wind power plant are possibly interfered by noise factors to fluctuate in a short time and the good prediction is difficult to carry out. The invention comprises the following steps: firstly, the method comprises the following steps: calculating the power variation trend of each time point precursor time point; II, secondly: dividing all data into G groups, wherein each group of data corresponds to a central point; thirdly, the method comprises the following steps: obtaining G regression prediction models from model (1) to model (G); fourthly, the method comprises the following steps: constructing an approximate weight of each central point based on the distance between the description vector V and the central point of the grouped data; fifthly: and obtaining a wind power prediction result. The invention is used in the technical field of wind power.
Description
Technical Field
The invention relates to a wind power prediction method.
Background
Wind power generation is influenced by various factors such as wind direction, wind speed and air pressure, so that the wind power generation has the characteristics of fluctuation, intermittence and randomness, the wind power is continuously fluctuated and changed due to the characteristics, the voltage stability of a power grid in the whole area is further influenced, and the safe and stable operation of the whole power grid is greatly influenced. Therefore, the wind power is very necessary to be predicted, and the condition and the trend of possible change of the wind power are extracted and obtained; the capacity of a standby thermal power generation system for wind power consumption can be effectively reduced through wind power prediction, the running cost of the whole system is reduced, and meanwhile, the safety and the stability of the whole power grid are improved, so that the wind power prediction has very important practical application value.
A large number of wind power prediction methods are available at present, and the methods tend to improve the wind power prediction accuracy by using a single model or a plurality of model voting modes, so that certain achievements are achieved. However, the current method also needs to face the following problems: firstly, data collected from a wind power plant are transient, and a certain time delay exists from the moment that a wind power pushes a generator blade to rotate and finally an output power change is reflected, and the delay characteristic needs to be reflected in an algorithm; secondly, the data collected from the wind power plant can be interfered by noise factors to fluctuate in a short time, and the fluctuation is not necessarily related to the change of the wind power; thirdly, the relation between the wind direction, the wind speed, the air pressure and other factors and the result of the wind power is not linear and is influenced by various factors, and all existing data are combined into a training set to train one model or multiple models to be difficult to predict well. Meanwhile, in some methods, the prediction is divided into a plurality of groups (for example, data is divided into three groups of high, medium and low) by wind speed for grouping prediction, a splitting phenomenon among the groups can occur, and when the wind speed is near the boundary of the groups, more fluctuation of the prediction result can occur.
Therefore, a method for further improving the wind power prediction accuracy is needed to solve the above problems.
Disclosure of Invention
The invention aims to provide a wind power prediction method based on integration of a time interval fuzzy operator and approximate weight, aiming at solving the problems that a certain time delay exists between the time when a wind power pushes a generator blade to rotate in a wind power plant and the time when the output power changes, and the problem that data collected from the wind power plant by the existing wind power prediction method is possibly interfered by noise factors to fluctuate in a short time.
A wind power prediction method based on time interval fuzzy operator and approximate weight integration comprises the following steps:
aiming at the problems in the prior art, the invention provides a wind power prediction method based on time interval fuzzy operator and approximate weight integration. The influence of noise and mutation data on the prediction precision is reduced through a time interval fuzzy operator, and meanwhile, historical data are introduced to a prediction point to reflect the delay reaction of output power change on instantaneous data; meanwhile, by utilizing approximate weight integration, the weight-based combination of a plurality of groups of model results according to the approximation degree of data is realized, the nonlinear relation between natural conditions and power output can be fully expressed, and meanwhile, the prediction results among the groups are smoother, and more accurate prediction results are obtained.
The method comprises the following steps: collecting wind power plant operation data, establishing a time interval operator based on a time interval L, describing the wind power plant data by using the time interval operator, and calculating the power variation trend of a precursor time point of each time point and the wind power variation trend of a subsequent time point of the time point t;
step two: grouping wind power plant data, and dividing all the data into G groups, namely group (1) to group (G), wherein each group of data corresponds to a central point;
step three: learning each group of data obtained in the second step by using a support vector machine algorithm to obtain G regression prediction models (1) to (G);
step four: constructing a description vector V for the operation data of a to-be-predicted time point tc of the wind power plant, and constructing an approximate weight of each central point based on the distance between the description vector V and the grouped data central point;
step five: inputting a description vector V into each regression prediction model to obtain a corresponding prediction result; performing approximate weight integration based on the approximate weight of each central point; and obtaining a wind power prediction result.
The invention has the beneficial effects that:
the invention provides a wind power prediction method based on time interval fuzzy operator and approximate weight integration. The influence of noise and mutation data on the prediction precision is reduced through the time interval fuzzy operator, and meanwhile, the historical data can be added into the current time point through the time interval fuzzy operator in a certain weight, so that the historical data also has a certain influence on the current power prediction result, and the delayed response of the power to the data is reflected. Meanwhile, by utilizing approximate weight integration, the weight-based combination of a plurality of groups of model results according to the approximation degree of data is realized, the nonlinear relation between natural conditions and power output can be fully expressed, and meanwhile, the prediction results among the groups are smoother, and more accurate prediction results are obtained. The method can realize high-precision wind power prediction. The method can effectively reduce the capacity of the thermal power generation standby power system for wind power consumption, reduce the operation cost of the whole system, and simultaneously improve the safety and stability of the whole power grid; the method has good application and popularization values for expanding the wind power utilization rate, promoting energy conservation and environmental protection, reducing carbon emission, improving the power grid management capability and reducing the operation cost.
Drawings
FIG. 1 is a flow chart of the present invention;
FIG. 2 is a flow chart of step one;
FIG. 3 is a flowchart of step two;
FIG. 4 is a flowchart of step three;
FIG. 5 is a flowchart of step four;
FIG. 6 is a flowchart of step five;
FIG. 7 is a graph of the results of prediction differences;
FIG. 8 is a histogram of predicted differences versus number of samples.
Detailed Description
The first embodiment is as follows: as shown in fig. 1, a wind power prediction method based on integration of a time interval fuzzy operator and an approximate weight includes the following steps:
the method comprises the following steps: collecting wind power plant operation data, establishing a time interval operator based on a time interval L, describing the wind power plant data by using the time interval operator, and calculating the power variation trend of a precursor time point of each time point and the wind power variation trend of a subsequent time point of the time point t;
step two: performing characteristic grouping on the data of the wind power plant, and dividing all the data into G groups, namely group (1) to group (G), wherein each group of data corresponds to a central point;
step three: learning each group of data obtained in the second step by using a Support Vector Machine (SVM) algorithm to obtain G regression prediction models (1) to (G);
step four: constructing a description vector V for the operation data of a to-be-predicted time point tc of the wind power plant, and constructing an approximate weight of each central point based on the distance between the description vector V and the grouped data central point;
step five: inputting a description vector V into each regression prediction model to obtain a corresponding prediction result; performing approximate weight integration based on the approximate weight of each central point; and obtaining a wind power prediction result.
The second embodiment is as follows: the first difference between the present embodiment and the specific embodiment is: as shown in fig. 2, the first step of collecting the wind farm operation data, establishing a time interval operator based on a time interval L, describing the wind farm data by using the time interval operator, and calculating the power variation trend of each time point at a previous time point comprises the following specific processes:
the method comprises the following steps: collecting the operation data of a wind power plant every Gape second; the Gape corresponds to the number of seconds of data collection intervals, and the default value is 900 seconds; the operational data of the wind farm includes: wind power F1, wind speed F2, humidity F3, temperature F4, air pressure F5 and wind direction F6;
storing the collected data in a database; each time point corresponds to a record in the database, and for a time point t, each record comprises the following fields: id (t) indicates time information accurate to seconds, F1(t) indicates output wind power of the wind farm at time point t, F2(t) indicates wind speed of the wind farm at time point t, F3(t) indicates ambient humidity at time point t, F4(t) indicates ambient temperature at time point t, F5(t) indicates air pressure of the wind farm at time point t, and F6(t) indicates wind direction of the wind farm at time point t;
the first step is: counting a mean value F1 mu and a standard deviation F1 sigma of F1 wind power, a mean value F2 mu and a standard deviation F2 sigma of F2 wind speed, a mean value F3 mu and a standard deviation F3 sigma of F3 humidity, a mean value F4 mu and a standard deviation F4 sigma of F4 temperature and a mean value F5 mu and a standard deviation F5 sigma of F5 air pressure;
step one is three: establishing a fuzzy Operator of a time interval L; the default value of the time zone L is 1800 seconds;
step one is: calculating the data of each time point of the wind power plant by using a time interval fuzzy Operator to obtain interval description fields Q1-Q5;
step one and five: calculating the wind power change trend of the predecessor time point of the time point t, and obtaining trend description fields Q6, Q7 and Q8; the predecessor time point of the time point t refers to a time point before the time point t;
step one is six: calculating a wind power change trend D1 of a subsequent time point of the time point t; the subsequent time point to the time point t refers to a time point after the time point t.
Other steps and parameters are the same as those in the first embodiment.
The third concrete implementation mode: the present embodiment differs from the first or second embodiment in that: in the third step, for a time interval L, a specific process of establishing a time interval fuzzy Operator is as follows:
the fuzzy membership of the time point with the distance t of k seconds is described as follows:
the time region blurring operator is as follows:
other steps and parameters are the same as those in the first or second embodiment.
The fourth concrete implementation mode: the difference between this embodiment mode and one of the first to third embodiment modes is: in the fourth step, for the data at each time point of the wind farm, a time interval fuzzy Operator is used for calculation, and the specific process of obtaining the interval description fields Q1 to Q5 is as follows:
the Q1 to Q5 for a time t are calculated as follows:
other steps and parameters are the same as those in one of the first to third embodiments.
The fifth concrete implementation mode: the difference between this embodiment and one of the first to fourth embodiments is: in the step one five, the wind power change trend of the predecessor time point of the time point t is calculated, and the specific process of obtaining the trend description fields Q6, Q7 and Q8 is as follows:
setting 3 precursor time variables as PR1, PR2 and PR 3; the default value of PR1 is 1800 seconds, the default value of PR2 is 2700 seconds, and the default value of PR3 is 3600 seconds;
the formula for Q6 is: q6(t) ═ Q1(t) -Q1(t-PR1)
The formula for Q7 is: q7(t) ═ Q1(t) -Q1(t-PR2)
The formula for Q8 is: q8(t) ═ Q1(t) -Q1(t-PR3)
Other steps and parameters are the same as in one of the first to fourth embodiments.
The sixth specific implementation mode: the difference between this embodiment and one of the first to fifth embodiments is: the concrete process of calculating the wind power change trend D1 of the subsequent time point of the time point t in the first step six is as follows:
d1 is the wind power change trend after the subsequent AFT seconds, the AFT default value is 1800 seconds, and the calculation formula of D1 is as follows:
D1(t)＝(Q1(t)-Q1(t+AFT))。
other steps and parameters are the same as those in one of the first to fifth embodiments.
The seventh embodiment: the difference between this embodiment and one of the first to sixth embodiments is: as shown in fig. 3, in the second step, the wind farm data are grouped according to characteristics, and all data are divided into G groups, i.e., group (1) to group (G), where a specific process that each group of data corresponds to one central point is as follows:
g represents the number of the wind power plant data groups, and the default value of the number is 16 groups;
step two, firstly: all contents of fields Q1-Q8 of the wind farm data processed in the step one are extracted;
step two: performing clustering calculation on the data taken out in the second step by using a K-Means algorithm (K mean algorithm), and dividing all the data into G groups, namely group (1) to group (G);
step two and step three: obtaining a central point vector of the G groups of data;
for the group (j) th data, the calculation formula of the center point vector is as follows:
Center(j)＝(avg(Q1),avg(Q2),avg(Q3),avg(Q4),avg(Q5),avg(Q6),avg(Q7),avg(Q8))
wherein avg (Q1) is the average of all Q1 fields of group (j) th data, avg (Q2) is the average of all Q2 fields of group (j) th data, avg (Q3) is the average of all Q3 segments of group (j) th data, avg (Q4) is the average of all Q4 fields of group (j) th data, avg (Q5) is the average of all Q5 fields of group (j) th data, avg (Q6) is the average of all Q6 fields of group (j) th data, avg (Q7) is the average of all Q7 fields of group (j) th data, and avg (Q8) is the average of all Q8 fields of group (j) th data.
Other steps and parameters are the same as those in one of the first to sixth embodiments.
The specific implementation mode is eight: the present embodiment differs from one of the first to seventh embodiments in that: as shown in fig. 4, in the third step, learning is performed on each group of data obtained in the second step by using a support vector machine algorithm, and a specific process of obtaining G regression prediction models (1) to (G) is as follows:
step three, firstly: converting all data in the group (j) th data into a power data change table, wherein the power data change table comprises the following two groups of data:
attribute data: corresponding to fields Q1 through Q8 in group (j);
data to be predicted: corresponding to the D1 field in group (j);
wherein, group (1) is less than or equal to group (j) is less than or equal to group (G);
step three: and (5) learning a power data change table by using a support vector machine algorithm to obtain a regression prediction model (j).
Other steps and parameters are the same as those in one of the first to seventh embodiments.
The specific implementation method nine: the present embodiment differs from one of the first to seventh embodiments in that: as shown in fig. 5, in the fourth step, for the operation data of the to-be-predicted time point tc of the wind farm, a description vector V is constructed, and a specific process of constructing the approximate weight of each central point based on the distance between the description vector V and the grouped data central point is as follows:
step four, firstly: constructing a description vector V of the current wind power plant for the data of the current time point;
V＝(Q1(tc),Q2(tc),Q3(tc),Q4(tc),Q5(tc),Q7(tc),Q8(tc))
step four and step two: calculating the distance dis (j) of V from each data packet center point center (j);
dis(j)＝|V-center(j)|
step four and step three: calculating the distance sum DisSum;
step four: obtaining an approximate weight light (j) corresponding to each central point;
other steps and parameters are the same as those in one to eight of the embodiments.
The detailed implementation mode is ten: the present embodiment differs from one of the first to ninth embodiments in that: as shown in fig. 6, in the fifth step, each regression prediction model inputs a description vector V to obtain a corresponding prediction result; performing approximate weight integration based on the approximate weight of each central point; the specific process for obtaining the wind power prediction result comprises the following steps:
step five, first: inputting a description vector V to each regression prediction model to obtain a corresponding prediction result;
the formula of predicting by using the prediction function prediction of model (j) is as follows:
pvalue(j)＝predict(v)
wherein, predict (V) is to use the model generated by the support vector machine algorithm to carry out regression prediction on the vector V to obtain a prediction result;
step five two: calculating an approximate weight integration result based on the approximate weight of each central point;
the calculation formula of the integration result ensemble is as follows:
step five and step three: calculating a result of wind power prediction;
the calculation formula of the wind power prediction result is as follows:
result＝(ensemble+Q1(tc))×F1σ×2+F1μ。
other steps and parameters are the same as those in one of the first to ninth embodiments.
The following examples were used to demonstrate the beneficial effects of the present invention:
the first embodiment is as follows:
in order to verify the effectiveness of the algorithm, the power generation data of 30 wind power generators in a certain northeast wind power plant in 2012, year 2 and month is introduced into the algorithm as training data, and the data of 2013, year 2 and month is used as test data. 960 time sampling points were continuously introduced at 15 minute intervals and predicted using the method of the present invention. The predicted results were compared with the actual results and evaluated using the following formula:
prediction of actual generated power-prediction result of patent of the invention
The corresponding prediction difference results are shown in fig. 7, and the histogram of the prediction difference and the number of samples is shown in fig. 8.
It can be seen that the vast majority of the predictions range from-5 MW to +5MW, with a small percentage ranging from-8 to-5 and 5 to 6. The average predicted absolute error of 960 test points is only 2.53, which shows that the predicted result of the invention is closer to the actual operation result and can be used as reference data for wind power plant and power grid management.
The present invention is capable of other embodiments and its several details are capable of modifications in various obvious respects, all without departing from the spirit and scope of the present invention.
Claims (5)
1. A wind power prediction method based on time interval fuzzy operator and approximate weight integration is characterized in that: the wind power prediction method based on the integration of the time interval fuzzy operator and the approximate weight comprises the following steps:
the method comprises the following steps: collecting wind power plant operation data, establishing a time interval operator based on a time interval L, describing the wind power plant data by using the time interval operator, and calculating the power variation trend of a precursor time point of each time point and the wind power variation trend of a subsequent time point of the time point t; the specific process is as follows:
the method comprises the following steps: collecting the operation data of a wind power plant every Gape second; the Gape is the number of seconds of the data collection interval; the operational data of the wind farm includes: wind power F1, wind speed F2, humidity F3, temperature F4, air pressure F5 and wind direction F6;
storing the collected data in a database; each time point corresponds to a record in the database, and for the time point t, the field of each record comprises: id (t) indicates time information accurate to seconds, F1(t) indicates output wind power of the wind farm at time point t, F2(t) indicates wind speed of the wind farm at time point t, F3(t) indicates ambient humidity at time point t, F4(t) indicates ambient temperature at time point t, F5(t) indicates air pressure of the wind farm at time point t, and F6(t) indicates wind direction of the wind farm at time point t;
the first step is: counting a mean value F1 mu and a standard deviation F1 sigma of F1 wind power, a mean value F2 mu and a standard deviation F2 sigma of F2 wind speed, a mean value F3 mu and a standard deviation F3 sigma of F3 humidity, a mean value F4 mu and a standard deviation F4 sigma of F4 temperature and a mean value F5 mu and a standard deviation F5 sigma of F5 air pressure;
step one is three: establishing a fuzzy Operator of a time interval L; the specific process is as follows:
the fuzzy membership of the time point with the distance t of k seconds is described as follows:
the time region blurring operator is as follows:
step one is: calculating the data of each time point of the wind power plant by using a time interval fuzzy Operator to obtain interval description fields Q1-Q5; the specific process is as follows:
the Q1 to Q5 for a time t are calculated as follows:
Q1:
Q2:
Q3:
Q4:
Q5:
step one and five: calculating the wind power change trend of the predecessor time point of the time point t, and obtaining trend description fields Q6, Q7 and Q8; the predecessor time point of the time point t refers to a time point before the time point t; the specific process is as follows:
setting 3 precursor time variables as PR1, PR2 and PR 3;
the formula for Q6 is: q6(t) ═ Q1(t) -Q1(t-PR1)
The formula for Q7 is: q7(t) ═ Q1(t) -Q1(t-PR2)
The formula for Q8 is: q8(t) ═ Q1(t) -Q1(t-PR 3));
step one is six: calculating a wind power change trend D1 of a subsequent time point of the time point t; the subsequent time point of the time point t refers to a time point after the time point t; the specific process is as follows:
d1 is the wind power change trend after the subsequent AFT seconds, and the calculation formula of D1 is as follows:
D1(t)＝(Q1(t)-Q1(t+AFT))；
step two: grouping wind power plant data, and dividing all the data into G groups, namely group (1) to group (G), wherein each group of data corresponds to a central point;
step three: learning each group of data obtained in the second step by using a support vector machine algorithm to obtain G regression prediction models (1) to (G);
step four: constructing a description vector V for the operation data of a to-be-predicted time point tc of the wind power plant, and constructing an approximate weight of each central point based on the distance between the description vector V and the grouped data central point;
step five: inputting a description vector V into each regression prediction model to obtain a corresponding prediction result; performing approximate weight integration based on the approximate weight of each central point; and obtaining a wind power prediction result.
2. The wind power prediction method based on the integration of the time interval fuzzy operator and the approximate weight according to claim 1, characterized in that: grouping the wind power plant data in the second step, and dividing all the data into G groups, namely group (1) to group (G), wherein the specific process that each group of data corresponds to one central point is as follows:
g represents the number of the wind power plant data groups, and the default value of the number is 16 groups;
step two, firstly: all contents of fields Q1-Q8 of the wind farm data processed in the step one are extracted;
step two: performing clustering calculation on the data extracted in the second step by using a K-means algorithm, and dividing all the data into G groups, namely group (1) to group (G);
step two and step three: obtaining a central point vector of the G groups of data;
for the group (j) th data, the calculation formula of the center point vector is as follows:
Center(j)＝(avg(Q1),avg(Q2),avg(Q3),avg(Q4),avg(Q5),avg(Q6),avg(Q7),avg(Q8))
wherein avg (Q1) is the average of all Q1 fields of group (j) th data, avg (Q2) is the average of all Q2 fields of group (j) th data, avg (Q3) is the average of all Q3 segments of group (j) th data, avg (Q4) is the average of all Q4 fields of group (j) th data, avg (Q5) is the average of all Q5 fields of group (j) th data, avg (Q6) is the average of all Q6 fields of group (j) th data, avg (Q7) is the average of all Q7 fields of group (j) th data, and avg (Q8) is the average of all Q8 fields of group (j) th data.
3. The wind power prediction method based on the integration of the time interval fuzzy operator and the approximate weight as claimed in claim 2, characterized in that: in the third step, learning is performed on each group of data obtained in the second step by using a support vector machine algorithm, and the specific process of obtaining G regression prediction models (1) to (G) is as follows:
step three, firstly: converting all data in the group (j) th data into a power data change table, wherein the power data change table comprises the following two groups of data:
attribute data: corresponding to fields Q1 through Q8 in group (j);
data to be predicted: corresponding to the D1 field in group (j);
wherein, group (1) is less than or equal to group (j) is less than or equal to group (G);
step three: and (5) learning a power data change table by using a support vector machine algorithm to obtain a regression prediction model (j).
4. The wind power prediction method based on the integration of the time interval fuzzy operator and the approximate weight as claimed in claim 3, wherein: in the fourth step, a description vector V is constructed for the operation data of the wind power plant to-be-predicted time point tc, and the specific process of constructing the approximate weight of each central point based on the distance between the description vector V and the grouped data central points is as follows:
step four, firstly: constructing a description vector V of the current wind power plant for the data of the current time point;
V＝(Q1(tc),Q2(tc),Q3(tc),Q4(tc),Q5(tc),Q7(tc),Q8(tc))
step four and step two: calculating the distance dis (j) of V from each data packet center point center (j);
dis(j)＝|V-center(j)|
step four and step three: calculating the distance sum DisSum;
step four: obtaining an approximate weight light (j) corresponding to each central point;
5. the wind power prediction method based on the integration of the time interval fuzzy operator and the approximate weight as claimed in claim 4, wherein: inputting a description vector V into each regression prediction model in the step five to obtain a corresponding prediction result; performing approximate weight integration based on the approximate weight of each central point; the specific process for obtaining the wind power prediction result comprises the following steps:
step five, first: inputting a description vector V to each regression prediction model to obtain a corresponding prediction result;
the formula of predicting by using the prediction function prediction of model (j) is as follows:
pvalue(j)＝predict(v)
wherein, predict (V) is to use the model generated by the support vector machine algorithm to carry out regression prediction on the vector V to obtain a prediction result;
step five two: calculating an approximate weight integration result based on the approximate weight of each central point;
the calculation formula of the integration result ensemble is as follows:
step five and step three: calculating a result of wind power prediction;
the calculation formula of the wind power prediction result is as follows:
result＝(ensemble+Q1(tc))×F1σ×2+F1μ。
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710357797.4A CN107194509B (en) | 2017-05-19 | 2017-05-19 | Wind power prediction method based on time interval fuzzy operator and approximate weight integration |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710357797.4A CN107194509B (en) | 2017-05-19 | 2017-05-19 | Wind power prediction method based on time interval fuzzy operator and approximate weight integration |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107194509A CN107194509A (en) | 2017-09-22 |
| CN107194509B true CN107194509B (en) | 2020-06-09 |
Family
ID=59875587
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710357797.4A Active CN107194509B (en) | 2017-05-19 | 2017-05-19 | Wind power prediction method based on time interval fuzzy operator and approximate weight integration |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107194509B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107704968B (en) * | 2017-10-18 | 2021-07-06 | 吉林省电力科学研究院有限公司 | Distributed distance integrated parallelized wind power plant output power prediction method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102298707A (en) * | 2011-08-24 | 2011-12-28 | 辽宁力迅风电控制系统有限公司 | Wind power prediction method based on continuous time slice clustering and support vector machine (SVM) modeling |
| CN103996084A (en) * | 2014-06-06 | 2014-08-20 | 山东大学 | Wind power probabilistic forecasting method based on longitudinal moment Markov chain model |
| CN104299044A (en) * | 2014-07-01 | 2015-01-21 | 沈阳工程学院 | Clustering-analysis-based wind power short-term prediction system and prediction method |
| CN105279582A (en) * | 2015-11-20 | 2016-01-27 | 中国水利水电第十四工程局有限公司 | An ultra-short-term wind electricity power prediction method based on dynamic correlation characteristics |
| CN105956720A (en) * | 2016-06-07 | 2016-09-21 | 中南大学 | Wind power ultra short period prediction method based on T-S fuzzy model |
-
2017
- 2017-05-19 CN CN201710357797.4A patent/CN107194509B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102298707A (en) * | 2011-08-24 | 2011-12-28 | 辽宁力迅风电控制系统有限公司 | Wind power prediction method based on continuous time slice clustering and support vector machine (SVM) modeling |
| CN103996084A (en) * | 2014-06-06 | 2014-08-20 | 山东大学 | Wind power probabilistic forecasting method based on longitudinal moment Markov chain model |
| CN104299044A (en) * | 2014-07-01 | 2015-01-21 | 沈阳工程学院 | Clustering-analysis-based wind power short-term prediction system and prediction method |
| CN105279582A (en) * | 2015-11-20 | 2016-01-27 | 中国水利水电第十四工程局有限公司 | An ultra-short-term wind electricity power prediction method based on dynamic correlation characteristics |
| CN105956720A (en) * | 2016-06-07 | 2016-09-21 | 中南大学 | Wind power ultra short period prediction method based on T-S fuzzy model |
Non-Patent Citations (2)
| Title |
|---|
| 基于模糊信息粒化和最小二乘支持向量机的风电功率联合预测建模;王恺等;《电力系统保护与控制》;20150116;第43卷(第2期);全文 * |
| 基于连续时间段聚类的支持向量机风电功率预测方法;丁志勇等;《电力系统自动化》;20120725;第36卷(第14期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107194509A (en) | 2017-09-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109751206B (en) | Fan blade icing fault prediction method and device and storage medium | |
| Chaudhary et al. | Short term wind power forecasting using machine learning techniques | |
| Dubey et al. | A Review of Intelligent Systems for the Prediction of Wind Energy Using Machine Learning | |
| CN114021483A (en) | Ultra-short-term wind power prediction method based on time domain characteristics and XGboost | |
| CN112651576A (en) | Long-term wind power prediction method and device | |
| CN109236589B (en) | It is a kind of for assessing the method and device of fan blade deicing capital project | |
| CN107194509B (en) | Wind power prediction method based on time interval fuzzy operator and approximate weight integration | |
| CN114386718A (en) | Wind power plant output power short-time prediction algorithm combined with particle swarm neural network | |
| Mulders et al. | Efficient tuning of individual pitch control: a Bayesian optimization machine learning approach | |
| Jiang et al. | Understanding wind turbine interactions using spatiotemporal pattern network | |
| CN116702957A (en) | New energy power prediction method, equipment and storage medium for extreme weather | |
| Yang | Short-term load monitoring of a power system based on neural network | |
| CN105932669B (en) | Wind power swing component decomposer and wind power output wave characteristic appraisal procedure | |
| Rajeevan et al. | ARIMA modeling of wind speed for wind farm reliability analysis | |
| Yesilbudak et al. | kNN Classifier Applications in Wind and Solar Energy Systems | |
| Khan et al. | Wind energy potential estimation for different regions of Bangladesh | |
| Scappaticci et al. | Analysing Wind Turbine States and SCADA Data for Fault Diagnosis | |
| Gao et al. | Very-short-term prediction of wind speed based on chaos phase space reconstruction and NWP | |
| CN111539577A (en) | Short-term wind power prediction method based on wind speed change rate and Gaussian process regression | |
| CN111191857A (en) | Wind power generation prediction system model based on gradient lifting algorithm | |
| Yang et al. | Wind-storage combined system based on just-in-time-learning prediction model with dynamic error compensation | |
| Yuan et al. | State transition ANNs for hourly wind speed forecasting | |
| Ma et al. | Photovoltaic time series aggregation method based on K-means and MCMC algorithm | |
| Huang et al. | An Ensemble Learning Approach for Wind Power Forecasting | |
| Cardoso Filho et al. | Wind Power Curve Modeling Through Data-driven Approaches: Evaluating Piecewise Linear Fitting and Machine Learning Applications in a Real-Unit Case |
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 |