A method for correcting wind power forecast data in special weather based on case-based reasoning

Abstract

The present invention discloses a method for correcting wind power forecast data for special weather conditions based on case-based reasoning. The method analyzes and describes the intermittent nature of wind speed through fractal dimension theory, divides the data into intermittent wind speed intervals and intermittent wind speed points, extracts wind speed and wind power data at all intermittent wind speed points in historical measured data of wind farms, and constructs a special weather wind speed-power case library. On this basis, all intermittent wind speed points in the wind power forecast segment data are corrected. For all intermittent wind speed points in the forecast data, the wind power forecast value at the point is weightedly corrected using the solution obtained by searching the case library to obtain the final wind power forecast output value, thereby reducing the forecast error.

Description

Technical Field

[0001] The present invention belongs to the field of wind power forecast data correction, and in particular relates to a method for correcting wind power forecast data in special weather conditions based on case-based reasoning. Background Art

[0002] Due to sudden changes in meteorological factors such as wind speed in special weather conditions such as thunderstorms and typhoons, the predicted values ​​of traditional wind power prediction models often cannot fit the actual wind power output curve well, which easily leads to large prediction errors. How to solve the problem of low wind power prediction accuracy in this scenario deserves in-depth study.

[0003] Current domestic research on wind power prediction models for exceptional weather conditions primarily focuses on analyzing the fluctuation patterns of input data and optimizing prediction model algorithms. These methods improve wind power prediction accuracy through data preprocessing and model improvements. However, existing research has not quantitatively described exceptional weather conditions and has failed to achieve satisfactory results. Based on the above analysis, we should explore the specific criteria for determining exceptional weather time points and adopt appropriate methods to adjust wind power prediction values ​​for exceptional weather conditions, thereby improving wind power prediction accuracy. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for correcting wind power prediction data in special weather conditions based on case reasoning, so as to solve the problem that the spatiotemporal correlation of wind speed series in special weather conditions is poor, the prediction value error of traditional models is large, and the forecast value cannot be well matched to the actual wind power output.

[0005] In order to solve the above technical problems, the inventors adopt the following technical solutions: In the first aspect, the present invention provides a method for correcting wind power forecast data in special weather conditions based on case-based reasoning, comprising the following steps:

[0006] 1) The fractal dimension theory in chaos theory is used to analyze the fluctuation characteristics of wind energy, determine the intermittent wind speed range and intermittent wind speed points, and approximately use the intermittent wind speed points as the basis for determining special weather time points;

[0007] 2) Based on the historical operation data of the wind farm, the wind speed at all intermittent wind speed points and the corresponding wind power output values ​​are integrated to establish a wind speed-power case library under special weather conditions;

[0008] 3) For each point in the prediction section of the wind power prediction model, determine whether it is an intermittent wind speed point. If so, correction is required. The wind speed is extracted as a feature quantity and the case library is searched to obtain the correction value. If not, the correction value is set to zero and no correction is required.

[0009] 4) The correction value obtained by searching the case library is used to correct the output value of the prediction model according to a certain weight to obtain the corrected wind power prediction value.

[0010] In a second aspect, the present invention provides a computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method described in the first aspect when executing the program.

[0011] In a third aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method described in the first aspect.

[0012] In a fourth aspect, the present invention provides a computer program product, comprising a computer program, which implements the steps of the method described in the first aspect when executed by a processor.

[0013] Compared with the prior art, the present invention has the following significant advantages:

[0014] (1) The present invention uses the fractal dimension theory in chaos theory to analyze the intermittent nature of wind speed, and uses the intermittent wind speed point as the basis for determining the time point of special weather, providing a specific quantitative indicator for special weather scenes such as thunderstorms and typhoons;

[0015] (2) The present invention applies case-based reasoning technology to the field of wind power prediction, providing a new approach to solving the problem of large wind power prediction errors in special weather conditions such as thunderstorms and typhoons. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 This is a flow chart of the data correction method based on case-based reasoning.

[0017] Figure 2 It is a schematic diagram of the division results of intermittent wind speed intervals and intermittent wind speed points.

[0018] Figure 3 This is a partial display diagram of the wind speed-power case library under special weather conditions.

[0019] Figure 4 It is the absolute error diagram of intermittent wind speed point data before and after correction.

[0020] Figure 5 This is a comparison chart of wind power forecast values ​​before and after data correction. DETAILED DESCRIPTION

[0021] The high volatility, randomness, and intermittency of wind energy determine the high volatility of wind power. While conventional wind power forecasting models can generally meet the accuracy requirements for wind power grid integration, in exceptional weather conditions such as thunderstorms and typhoons, sudden changes in meteorological factors like wind speed make it difficult for the wind power forecast curve to accurately align with the actual wind power output curve, easily leading to large wind power forecast errors.

[0022] Based on the above analysis, the present invention proposes a method for correcting wind power prediction data in special weather conditions based on case reasoning. This method uses fractal dimension theory to analyze the intermittent nature of wind speed, divides it into intermittent wind speed intervals and intermittent wind speed points, and approximates the intermittent wind speed points as the basis for determining special weather time points. For wind power prediction data, if it is judged that a certain time point is in special weather, it is necessary to search the case library to obtain the correction value, and use the weighted correction method to correct the wind power prediction value at that point. If it is not a special weather time point, the wind power prediction data is directly output. After the data correction, the prediction accuracy of the wind power prediction model can be effectively improved. Figure 1 As shown, the specific steps for implementing the correction method are as follows:

[0023] The first step is to use the box-counting dimension method in fractal dimension theory to geometrically quantify the wind speed fluctuations within a specific time period. The basic idea of ​​box-counting dimension is: if the irregular curve is placed in a uniformly divided grid with a side length of ε, at least N grids are required to completely cover the irregular curve. The mathematical expression of the box-counting dimension of the curve is as follows:

[0024]

[0025] When ε→0, the fractal dimension D S The value of is equal to the approximate slope of the curve. For the wind speed time series, the specific description of the box-counting dimension method is to decompose the historical wind speed time series of length t into N intervals, and the time length of each interval is a fixed value Δt. Then the fractal dimension D in the i-th wind speed interval is i The expression is:

[0026]

[0027] Among them, v i,max With v i,min are the maximum and minimum wind speeds in the i-th wind speed interval, respectively. Research on fractal theory shows that the larger the fractal dimension, the greater the randomness of the data, that is, the stronger the wind speed fluctuation. Therefore, by selecting an appropriate fractal dimension value D r , we can determine whether the interval is an intermittent wind speed interval by analyzing the fractal dimension value in each wind speed interval. If the fractal dimension value D in the i-th wind speed interval is i>D r , it is considered that the wind speed fluctuation in this interval is strong, and it is judged to be an intermittent wind speed interval. i With v i+1 Represent two wind speed points in the interval, then we can judge v i is the intermittent wind speed point. Combined with the actual experimental conditions of the wind farm, the fractal dimension reference value D is determined. r =1.5, for a certain wind speed time series, the division results of intermittent wind speed intervals and intermittent wind speed points are as follows Figure 2 Shown in the dotted box.

[0028] The intermittent wind speed points obtained based on the above method can be approximately regarded as special weather time points.

[0029] The second step is to use the historical measured data of wind farms as the basis, combined with the correlation analysis of factors affecting wind power output, to count and integrate the wind speeds at all intermittent wind speed points and the corresponding wind power output values, and to form a wind speed-power case library under special weather conditions. For the nth intermittent wind speed point, it is regarded as the nth special weather point, and the wind speed v at the current moment is taken. i With the characteristic vector of the case {x n}, the solution of the case {y n} is the actual wind power output value at the current moment. Then any source case can be expressed as:

[0030] C n ={x n ;y n},n=1,2,…N (3)

[0031] For the historical actual data of a wind farm, the wind speed-power case library under special weather conditions is shown as follows: Figure 3 shown.

[0032] In the third step, for the predicted output data of any wind power prediction model, the fractal dimension theory is used to determine whether the time point is an intermittent wind speed point. If so, it is determined that the time point is under special weather conditions and the predicted data needs to be corrected. Let the target case (i.e., the case to be solved) be x, and the solution to be solved be denoted as y. The similarity evaluation strategy based on Euclidean distance is used to calculate the similarity between x and the source case x. n The similarity is calculated as follows:

[0033]

[0034] When the similarity is greater than 85%, the current case is considered to match the case in the case library. The cases with a similarity greater than 85% are sorted according to the similarity, and the top 5 are taken (if there are less than 5 cases, all are selected), and their average value is calculated as the solution y for the current case. If there are no cases with a similarity greater than 85%, the similarity can be appropriately reduced to 80% or 75% (depending on the actual search situation) until a case that meets the requirements is found. The solution y obtained is the correction value in the data correction module. If it is determined that the time point is not an intermittent wind speed point, there is no need to correct the forecast data. At this time, the correction value should be set to zero, that is, the output value of the forecast model should be directly output.

[0035] For a certain time point in the wind power prediction period, the wind speed is v = 4.57, and the wind power prediction value is P w =3.94. According to the intermittent wind speed point judgment method in step 1, it is judged to be a special weather time point. The case library is searched and the top five cases are selected from high to low according to similarity: (5.557, 1.848), (5.56, 1.864), (4.718, 1.017), (4.583, 0.884), (4.474, 0.763). The correction value y = 1.28 is calculated according to formula (4).

[0036] In the fourth step, based on the correction value obtained in the previous step, the output value of the wind power prediction model is corrected using a weighted combination method. The weight is determined using the water injection method. The principle of the water injection method is that sub-channels have different attenuation characteristics, so allocating more power to sub-channels with high signal-to-noise ratios and less power to sub-channels with low signal-to-noise ratios can effectively improve signal transmission efficiency. The mathematical expression of the power allocation mechanism of the water injection method is as follows:

[0037]

[0038] Where Q is the channel capacity, P m Indicates the power allocated to the mth subchannel; is the noise variance of the mth characteristic subchannel, α m is the gain of the mth subchannel. The value of this parameter represents the importance of the mth attribute, that is, the correlation between the mth feature quantity in the case library and the output. α is calculated by the correlation coefficient m The value is calculated as follows:

[0039]

[0040] in, is the average value of the mth feature value, is the average value of all solutions in the case library, and N is the total number of source cases.

[0041] Further calculate the threshold value:

[0042]

[0043] Weight ω m The calculation formula is as follows.

[0044]

[0045] The calculated weights ω1 = 0.1, ω2 = 0.9, for the i-th time point of the predicted data, the output value of the prediction model is P wi , if it is a special weather time point, the corrected output prediction value P i for:

[0046] P i =ω1P w,i +ω2y (9)

[0047] For the forecast period in step 3, the weighted corrected wind power value is calculated using formula (9), namely:

[0048] P=0.1*3.94+0.9*1.28=1.55 (10)

[0049] The corrected value is used as the output of the prediction model. Figure 4 and Figure 5 The absolute error comparison and predicted value comparison of the prediction segment data of a certain wind power prediction model before and after data correction are given respectively.

[0050] Those skilled in the art will readily appreciate that various modifications and variations can be made to the present invention without departing from the spirit and scope of the present invention. Therefore, the scope of protection of the present invention shall be determined by the claims.

[0051] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for correcting wind power forecast data in special weather conditions based on case-based reasoning, characterized in that: The following steps are involved: 1) The fractal dimension theory in chaos theory is used to analyze the fluctuation characteristics of wind energy, determine the intermittent wind speed range and intermittent wind speed points, and approximately use the intermittent wind speed points as the basis for determining special weather time points; specifically: First, we use the box-counting dimension method in fractal dimension theory to geometrically quantify the wind speed fluctuations within a specific time period. The basic idea of ​​the box-counting dimension is: if an irregular curve is placed in a uniformly divided grid with a side length of ε, at least N grids are required to completely cover the irregular curve. The mathematical expression of the box-counting dimension of the curve is as follows: When ε→0, the fractal dimension D S The value of is equal to the approximate slope of the curve; for the wind speed time series, the specific description of the box-counting dimension method is to decompose the historical wind speed time series of length t into N intervals, and the time length of each interval is a fixed value Δt, then the fractal dimension D in the i-th wind speed interval is i The expression is: Among them, v i,max With v i,min are the maximum and minimum wind speeds in the i-th wind speed interval respectively; the larger the fractal dimension, the greater the randomness of the data, that is, the stronger the wind speed volatility, so by selecting a suitable fractal dimension reference value D r , analyzing the fractal dimension of each wind speed interval can determine whether the interval is an intermittent wind speed interval. If the fractal dimension value D in the i-th wind speed interval is i >D r , it is considered that the wind speed fluctuation in this interval is strong, and it is judged to be an intermittent wind speed interval. i With v i+1 Represent two wind speed points in the interval, then we can judge v i is the intermittent wind speed point; The intermittent wind speed points obtained based on the above method can be approximately regarded as special weather time points; 2) Based on the historical operation data of the wind farm, the wind speed at all intermittent wind speed points and the corresponding wind power output values ​​are integrated to establish a wind speed-power case library under special weather conditions; For the nth intermittent wind speed point, consider it as the nth special weather point and take the wind speed v at the current moment. i With the characteristic vector of the case {x n }, the solution of the case {y n } is the actual wind power output value at the current moment; then any source case C n Expressed as: C n {x n ;y n },n=1,2,…N (3) 3) For each point in the prediction section of the wind power prediction model, determine whether it is an intermittent wind speed point. If so, correction is required. The wind speed is extracted as a feature quantity, and the case library is searched to obtain the correction value. If not, the correction value is set to zero and no correction is required. Specifically: For the predicted output data of any wind power prediction model, the fractal dimension theory is used to determine whether the time point is an intermittent wind speed point. If so, it is determined that the time point is under special weather conditions and the predicted data needs to be corrected. Let the target case be x and the solution to be found be y. The similarity evaluation strategy based on Euclidean distance is used to calculate the similarity between x and the source case x. n The similarity S is calculated as follows: When the similarity is greater than 85%, the current case is considered to match the case in the case library. The cases with similarity greater than 85% are sorted according to the similarity, and the top 5 are taken. If there are less than 5 cases, all are selected and their average value is calculated as the solution y of the current case. If there is no case with similarity greater than 85%, the similarity is reduced until a case that meets the requirements is found. The solution y obtained is the correction value in the data correction module. If it is determined that the time point is not an intermittent wind speed point, there is no need to correct the forecast data. At this time, the correction value should be set to zero, that is, the output value of the forecast model is directly output. 4) The correction value obtained by searching the case library is used to correct the output value of the prediction model according to a certain weight to obtain the corrected wind power prediction value, specifically: Based on the correction value obtained in the previous step, the output value of the wind power prediction model is corrected by using the weighted combination method. The weight is determined by the water injection method. The calculated weights ω1 = 0.1, ω2 = 0.

9. For the i-th time point of the prediction data, the output value of the prediction model is P wi , if it is a special weather time point, the corrected output prediction value P i for: P i =ω1P wi +ω2y (5).

2. The method for correcting wind power forecast data in special weather conditions based on case-based reasoning according to claim 1, characterized in that: Fractal dimension reference value D r =1.

5.

3. The method for correcting wind power forecast data in special weather conditions based on case-based reasoning according to claim 1, characterized in that: If there is no case with a similarity greater than 85%, the similarity is reduced to 80% or 75% until a case that meets the requirements is found.

4. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the method according to any one of claims 1 to 3 are implemented.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

6. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 3 are implemented.

Images