Offshore wind power prediction method

Abstract

The invention discloses an offshore wind power prediction method, solves the problem of large offshore wind power prediction error, and belongs to the field of offshore wind power prediction. The method comprises the following steps: preprocessing offshore wind power data, and dividing a wind speed-power interval; performing first abnormal point identification and first abnormal point cleaning on the power data in the divided wind speed-power interval by using a Thompson tau method; performing second abnormal point identification and second abnormal point cleaning on the power data after the first abnormal point cleaning by using a quartile method; and dividing the offshore wind power data into normal data and abnormal data according to abnormal points of the first abnormal point cleaning and the second abnormal point cleaning, and performing data reconstruction on the abnormal data to complete prediction. According to the method, abnormal data can be eliminated aiming at the problem of relatively large offshore wind power prediction error. And meanwhile, the recognition rate of abnormal data is improved.

Description

Technical Field

[0001] The invention relates to an offshore wind power prediction method and belongs to the field of offshore wind power prediction. Background Art

[0002] With the development of the global economy and the continuous growth of the population, human demand for energy is increasing. In the current power generation structure, thermal power still occupies a dominant position, so it is necessary to increase the development of clean energy. Therefore, wind power generation has become the focus of energy development in countries around the world. So far, wind power generation has become one of the clean energy sources with the fastest installed capacity growth. Three-quarters of the earth is covered by the ocean. Offshore wind energy is extremely abundant, and the benefits of wind energy resources are higher than those of onshore wind farms. Vigorously developing offshore wind power technology is an important measure to improve the energy structure and environmental quality. It can be foreseen that offshore wind power will further expand the proportion of wind power generation in installed power generation capacity.

[0003] However, offshore wind power is very different from traditional thermal power generation. It has the characteristics of randomness, intermittency, and volatility. Large-scale offshore wind power joint operation will bring burdens to the safe operation of the power grid. Therefore, it is necessary to predict the power generation of offshore wind power. Since the offshore wind power is more abundant, changes very quickly, and long-term wind power data is usually lacking, the prediction error is large or even completely fails to reflect the actual situation, and reasonable error correction is required. Summary of the invention

[0004] In order to solve the problem of large offshore wind power prediction errors, the present invention provides an offshore wind power prediction method.

[0005] A method for predicting offshore wind power of the present invention comprises:

[0006] S1. Preprocess offshore wind power data and divide the wind speed-power interval;

[0007] S2, using the Thompson tau method to identify and clean the first abnormal points of the power data in the wind speed-power interval;

[0008] S3, using the quartile method to perform a second abnormal point identification and a second abnormal point cleaning on the power data after the first abnormal point cleaning;

[0009] S4. According to the abnormal points obtained by the first abnormal point cleaning and the second abnormal point cleaning, the offshore wind power data is divided into normal data and abnormal data, and the abnormal data is reconstructed to complete the prediction.

[0010] Preferably, S1 comprises:

[0011] Filter out wind speed-power data with wind speed greater than cut-in wind speed, less than cut-out wind speed, and power less than or equal to zero from offshore wind power data;

[0012] For wind speed-power data below the rated wind speed, the wind speed is divided into n intervals according to the set size based on the amount of wind speed-power data, where n is a positive integer;

[0013] The wind speed-power data above the rated wind speed and below the cut-out wind speed is divided into an interval, which is recorded as the n+1th wind speed-power interval.

[0014] Preferably, S2 comprises:

[0015] Calculate the mean of the power data within the interval for:

[0016]

[0017] Where: P i，j are the power values ​​in the i-th interval, and m is the number of power data in the i-th interval;

[0018] The standard deviation S of the power data in the i-th interval i for:

[0019]

[0020] The absolute value of the deviation of each power data in the interval δ i,j for:

[0021]

[0022] If in the i-th interval, when δ i，j ≥τS i , then the absolute value of the deviation δ i，j The corresponding power data is an abnormal point; when δ i，j <τS i When the absolute value of the deviation δ i，j The corresponding power data is the normal point; τ is the Thompson tau value:

[0023]

[0024] Where t is the t distribution value of the power sample data, and α is the significance level.

[0025] Preferably, α=0.01.

[0026] Preferably, S3 includes:

[0027] After the first outlier cleaning, the number of remaining power data in each interval is m′, so the power data of the i-th interval is Pi ′={P i, ′1,P i, ′2,…,P i, ' m′}, where P i, ′1≤P i, ′2≤…≤P i, ' m′ ;

[0028] The second quantile M i Indicates P i The median of ′, the first and third digits represent P i The values ​​represented by the positions of the 25% data points before and after the segmentation in ′;

[0029] When m′ is an even number, M i P i ′ is divided into two subsequences of the same length, namely P i_1 = {P i, ′1,P i, ′2,…,P i, ' (m′-1) / 2} and P i_2 = = {P i, ' (m′+1) / 2 , P i, ' (m′+2) / 2 ,…,P i, ' m′}, Q 1.i and Q 3.i Represent subsequence P i_1 and P i_2 the median;

[0030] k = 0, 1, ...;

[0031] When m′=4k+3, Q 1.i and Q 3.i The calculation formula is:

[0032]

[0033] When m′=4k+1, Q 1.i and Q 3.i The calculation formula is:

[0034]

[0035] Get P i The interquartile range of ′ i for:

[0036] I i =Q 3,i -Q 1,i

[0037] Calculate the upper limit W u,i and the lower limit W d,i :

[0038]

[0039] At the upper limit W u,i and the lower limit W d,i The power data in the middle is the normal point, and the power data outside is the abnormal point.

[0040] As a preference, the second quantile M i for:

[0041]

[0042] As a preferred embodiment, in S4, the abnormal data is reconstructed using a four-point interpolation algorithm, and the power point to be interpolated P2′ i+1 :

[0043]

[0044] Among them, p i-1 , p i , p i+1 , p i+2 They are the power values ​​corresponding to the four wind speeds closest to the power point to be interpolated.

[0045] The beneficial effect of the present invention is that the offshore wind power data is first preprocessed and divided into wind speed-power intervals. Then the Thompson tau method is used to perform the first identification of the power data in the interval until no abnormal points are found, and then the data after the first cleaning is secondary identified using the quartile method to complete the abnormal power value identification of the wind speed-power interval. Finally, according to the identification results, the abnormal data is cleaned and replaced. The present invention can eliminate abnormal data for the problem of large offshore wind power prediction errors. At the same time, the recognition rate of abnormal data is improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 The figure is a flow chart of the prediction method according to the embodiment of the present invention.

[0047] Figure 2 This is the original data diagram of the fan.

[0048] Figure 3 Thompson tau method identification results.

[0049] Figure 4 Results of other recognition methods. DETAILED DESCRIPTION

[0050] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0052] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0053] Offshore wind power is developing rapidly. The high randomness of offshore wind power and the insufficient amount of historical data have brought great challenges to the prediction of offshore wind power. When the prediction accuracy is required to be higher, more accurate offshore wind power output data can be obtained through error correction methods. The offshore wind power prediction method of this embodiment includes:

[0054] Step 1: pre-process the offshore wind power data and divide the wind speed-power interval;

[0055] For points where the wind speed is greater than the cut-in wind speed and less than the cut-out wind speed, and the power is less than or equal to zero, it appears as a horizontally stacked data cluster at the bottom in the wind speed-power scatter plot. After completing the preprocessing of the wind speed-power data, for the wind speed-power data below the rated wind speed, the wind speed is divided into n intervals according to a certain size based on the amount of wind speed-power data; for the wind speed-power data above the rated wind speed and below the cut-out wind speed, the output power of the wind turbine no longer increases with the increase of wind speed, so the wind speed-power data above the rated wind speed and below the cut-out wind speed is divided into one interval, recorded as the n+1th wind speed-power interval.

[0056] Divided into n+1 intervals, the power data sample of the i-th wind speed-power interval is P i = {P i，1 , P i，2 ,…,P i，m}, where i = 1, 2, ..., s, P i，1 ≤P i，2 ≤…≤P i，m , m is the number of power points in each wind speed-power interval. The Thompson tau method is used to segmentally identify and eliminate abnormal power data points in each interval.

[0057] Step 2: Use the Thompson tau method to identify and clean the first abnormal points of the power data in the wind speed-power interval:

[0058] First, calculate the mean of the power data within the interval:

[0059]

[0060] Where: P i，j is the power data in the i-th interval.

[0061] The standard deviation of the power data in the i-th interval is:

[0062]

[0063] The absolute value of the deviation of each power sample data in the interval is recorded as:

[0064]

[0065] If in the i-th interval, the absolute value of the deviation of the observed power data δ i，j The maximum value is the point that is likely to be an outlier to be removed. Because when the power deviation value is the largest, it means that the power data is the largest or smallest in this interval, and the possibility that the point is a power outlier is greater. The calculation of τ value in Thompson tau method is as follows:

[0066]

[0067] Where: t is the t distribution value of the power sample data; α is the significance level, whose value affects the adequacy of the power data. This project chooses the rejection level α = 0.01. i,j ≥τS i When , the detected power data is an abnormal point; on the contrary, if δ i,j <τS i When , the detected power data is the normal point.

[0068] The Thompson tau method can detect one power anomaly point in each wind speed interval each time. When the power data is detected as an abnormal value, it is removed and the mean and standard deviation in the interval are recalculated. The value of τ will also change with the number of power data in the interval until no new power anomaly is found.

[0069] The recognition results of different abnormal data recognition methods are as follows: Figure 3 , Figure 4 As shown. It can be seen that the Thompson tau method has higher accuracy and has better recognition ability for abnormal data in each wind speed-power interval. Step 3: Use the quartile method to perform a second abnormal point recognition and a second abnormal point cleaning on the power data after the first abnormal point cleaning:

[0070] After the first identification is completed by the Thompson tau method, the number of power points remaining in each wind speed interval is m′. Then the power data sample of the i-th wind speed-power interval after the first identification is P i ′={P i, ′1,P i, ′2,…,P i, ' m′}, where P i, ′1≤P i, ′2≤…≤P i, ' m′ The quartile is a kind of quantile in statistics, that is, all the data are divided into four equal parts from small to large, and the data at the three dividing points are the quartiles. The second quantile M i Represents sample data P i The median of ′ is calculated as:

[0071]

[0072] The first and third digits represent P i The values ​​represented by the positions of the 25% data points before and after the segmentation in ′. The total number of data points m in the interval is different, and the calculation formula is slightly different.

[0073] When m is an even number, M i P i ′ is divided into two subsequences of the same length, namely P i_1 = {P i, ′1,P i, ′2,...,P i, ' (m′-1) / 2} and P i_2 = = {P i, ' (m′+1) / 2 , P i, ' (m′+2) / 2 , ..., P i, ' m′}, Q 1，i and Q 3，i Represent subsequence P i_1 and P i_2 of the median.

[0074] When m=4k+3 (k=0, 1, ...), the calculation formula is:

[0075]

[0076] When m=4k+1 (k=0, 1, ...), the calculation formula is:

[0077]

[0078] According to the calculation results of formula (6) and formula (7), we can get P i The interquartile range of ′ i for:

[0079] I i =Q 3,i -Q 1,i (8)

[0080] In the quartile method, the upper limit W is used u，i , lower limit W d，i To exclude outliers in the data, the calculation formulas are:

[0081]

[0082] According to the calculation result of formula (9), the power data sample P is defined as i In W u，i and W d，i The data points between are normal values, and the data points outside are outliers.

[0083] By dividing the wind speed intervals, the abnormal operation data in each interval are segmented and identified, and the identified data are deleted, and the entire wind speed-power data set is divided into two parts: normal data and abnormal data, that is,

[0084]

[0085] Where: D is the entire wind speed-power data set; D n is the normal wind speed-power data set; D0 is the abnormal wind speed-power data set; N is the number of all wind speed-power data pairs.

[0086] By identifying and eliminating abnormal wind speed-power operation data, normal data may be deleted by mistake, and some power data may be missing, which destroys the integrity of the data. Therefore, it is necessary to reconstruct the data after cleaning.

[0087] Step 4: According to the abnormal points of the first abnormal point cleaning and the second abnormal point cleaning, the offshore wind power data is divided into normal data and abnormal data, and the abnormal data is reconstructed to complete the prediction.

[0088] Power is proportional to the cube of wind speed, and output power is highly sensitive to wind speed changes. Wind turbines themselves have a large amount of historical data, which can be used as a reference for reconstruction of abnormal and missing data. The four-point interpolation subdivision algorithm uses four adjacent points to calculate new interpolation points, and uses the same calculation rules for each calculation. By finding four wind speed points close to the wind speed to be interpolated, it is theoretically feasible to use the four-point interpolation algorithm to interpolate missing data.

[0089] The specific calculation formula of the four-point interpolation algorithm is:

[0090]

[0091] Where: P′ 2i+1 is the power point to be interpolated; p i-1 , p i , p i+1 , p i+2 are the power values ​​corresponding to the four wind speeds closest to the power point to be interpolated; ω is a tensor parameter. When ω∈[0, 0.125], a more satisfactory interpolation point can be obtained.

[0092] Arranged:

[0093]

[0094] remember

[0095]

[0096] Substituting formula (13) into formula (12), we have

[0097]

[0098] For the abnormal wind speed-power dataset D0, we can find the n The four wind speed-power data with wind speeds close to the abnormal power data are used to obtain the reconstructed wind speed-power data set D of each abnormal point by using the four-point interpolation subdivision algorithm. r .

[0099] Although the present invention is described herein with reference to specific embodiments, it should be understood that these embodiments are merely examples of the principles and applications of the present invention. It should therefore be understood that many modifications may be made to the exemplary embodiments and that other arrangements may be devised without departing from the spirit and scope of the present invention as defined by the appended claims. It should be understood that the various dependent claims and features described herein may be combined in a manner different from that described in the original claims. It should also be understood that features described in conjunction with individual embodiments may be used in other described embodiments.

Claims

1. A method for predicting offshore wind power, characterized in that: The method comprises: S1. Preprocess offshore wind power data and divide the wind speed-power interval; S2, using the Thompson tau method to identify and clean the first abnormal points of the power data in the wind speed-power interval; S3, using the quartile method to perform a second abnormal point identification and a second abnormal point cleaning on the power data after the first abnormal point cleaning; S4. According to the abnormal points obtained by the first abnormal point cleaning and the second abnormal point cleaning, the offshore wind power data is divided into normal data and abnormal data, and the abnormal data is reconstructed to complete the prediction.

2. The offshore wind power prediction method according to claim 1, characterized in that: The S1 includes: Filter out wind speed-power data with wind speed greater than cut-in wind speed, less than cut-out wind speed, and power less than or equal to zero from offshore wind power data; For wind speed-power data below the rated wind speed, the wind speed is divided into n intervals according to the set size based on the amount of wind speed-power data, where n is a positive integer; The wind speed-power data above the rated wind speed and below the cut-out wind speed is divided into an interval, which is recorded as the n+1th wind speed-power interval.

3. The offshore wind power prediction method according to claim 2, characterized in that: The S2 includes: Calculate the mean of the power data within the interval for: Where: P i，j are the power values ​​in the i-th interval, and m is the number of power data in the i-th interval; The standard deviation S of the power data in the i-th interval i for: The absolute value of the deviation of each power data in the interval δ i,j for: If in the i-th interval, when δ i，j ≥τS i , then the absolute value of the deviation δ i，j The corresponding power data is an abnormal point; when δ i，j <τS i When the absolute value of the deviation δ i，j The corresponding power data is the normal point; τ is the Thompson tau value: Where t is the t distribution value of the power sample data, and α is the significance level.

4. The offshore wind power prediction method according to claim 3, characterized in that: α＝0.01。 5. The offshore wind power prediction method according to claim 2, characterized in that: The S3 includes: After the first outlier cleaning, the number of remaining power data in each interval is m′, so the power data of the i-th interval is P i ′={P i, ′1,P i, ′2,…,P i, ' m′ }, where P i, ′1≤P i, ′2≤…≤P i, ' m′ ; The second quantile M i Indicates P i The median of ′, the first and third digits represent P i The values ​​represented by the positions of the 25% data points before and after the segmentation in ′; When m′ is an even number, M i P i ′ is divided into two subsequences of the same length, namely P i_1 = {P i, ′1,P i, ′2,...,P i, ' (m′-1) / 2 } and P i_2 = = {P i, ' (m′+1) / 2 , P i, ' (m′+2) / 2 , ..., P i, ' m′ }, Q 1.i and Q 3.i Represent subsequence P i_1 and P i_2 the median; k＝0，1，…； When m′=4k+3, Q 1.i and Q 3.i The calculation formula is: When m′=4k+1, Q 1.i and Q 3.i The calculation formula is: Get P i The interquartile range of ′ i for: I i =Q 3,i -Q 1,i Calculate the upper limit W u,i and the lower limit W d,i : At the upper limit W u,i and the lower limit W d,i The power data in the middle is the normal point, and the power data outside is the abnormal point.

6. The offshore wind power prediction method according to claim 1, characterized in that: The second quantile M i for:

7. The offshore wind power prediction method according to claim 1, characterized in that: In S4, the four-point interpolation algorithm is used to reconstruct the abnormal data, and the interpolation power point P2′ i+1 : Among them, p i-1 , p i , p i+1 , p i+2 They are the power values ​​corresponding to the four wind speeds closest to the power point to be interpolated.

8. A computer-readable storage device storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the offshore wind power prediction method according to any one of claims 1 to 7 are implemented.

9. An offshore wind power prediction device, comprising a storage device, a processor, and a computer program stored in the storage device and executable on the processor, characterized in that: The processor executes the computer program to implement the steps of the offshore wind power prediction method according to any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the offshore wind power prediction method according to any one of claims 1 to 7 are implemented.

Images