Wind power plant long-term wind speed prediction method and system

Abstract

The invention provides a wind power plant long-term wind speed prediction method and system, and the method comprises the steps: obtaining the wind measurement data of a target station and a reference station of a specific wind power plant in a first time period, and obtaining the reanalysis data of the reference station in a second time period; obtaining related characteristics of the wind measurement data of the target station and the reference station in a first time period, and establishing a correlation model; and a method, a device, equipment and a system for predicting the long-term wind speed and the wind speed distribution condition of the target station by using the correlation model and the reanalysis data of the reference station in the second time period. According to the technical scheme provided by the invention, a single method combination application mode of the wind speed correlation model algorithm and the wind speed distribution fitting algorithm is provided, so that the long-term wind speed and wind speed distribution prediction of the target station is better and more accurate.

Description

Technical Field

[0001] This document relates to the technical field of wind speed prediction, and in particular to a method and system for long-term wind speed prediction in a wind farm. Background Art

[0002] Wind speed prediction of wind farms is the main indicator of wind energy resource assessment. Accurate prediction of wind speed is directly related to the assessment level of wind energy resources and the economic benefits of wind farms. It is a key factor in wind farm construction decision-making. It can be said that the accuracy of wind speed prediction plays a decisive role in the success of a wind power project. Based on this, scholars at home and abroad have invested a lot of time and energy to study wind speed prediction of wind farms and find suitable evaluation methods to improve the accuracy of prediction, including MCP methods based on statistics, MCP methods based on machine learning, and MCP methods based on physical models. Some of these evaluation methods were eventually eliminated, while accurate and effective evaluation methods were widely adopted. The MCP method was first proposed for the prediction of long-term average wind speed of the site to be tested, and gradually developed into a method for predicting the long-term wind speed and wind direction time series or frequency distribution of the site to be tested. The least squares method can accurately predict the long-term average wind speed, but the predicted wind speed distribution varies greatly, and it cannot guarantee the accurate prediction of power output; the vector method can simultaneously predict wind speed and direction, but it performs poorly in the prediction of long-term average wind speed and wind speed distribution; the joint probability density distribution method can obtain a better wind speed frequency distribution of the target station and can adapt to more complex terrain environments, but it is difficult to predict the long-term wind speed time series of the target station; Weibull distribution is the most commonly used distribution function for fitting wind speed distribution, which solves the problem that when the parameters of the Weibull distribution of wind speed between stations are different, the linear regression has a relatively poor fitting effect on the wind speed correlation model, but when the wind speed distribution has a bimodal distribution or the frequency of zero wind speed is high, it will greatly affect the goodness of fit of the Weibull distribution function to the wind speed distribution, thereby increasing the prediction error of the Weibull scaling method for the long-term wind speed frequency distribution of the target station.

[0003] Based on this, the present invention provides a method and system for long-term wind speed prediction in a wind farm, which improves the correlation of wind data by studying the variation law of the average power density and the average time period of a specific wind farm, thereby improving the prediction accuracy of the MCP model. Summary of the invention

[0004] The present invention provides a method and system for predicting long-term wind speed in a wind farm, aiming to solve the above-mentioned problem.

[0005] The embodiment of the present invention provides a method for long-term wind speed prediction in a wind farm, comprising: obtaining wind measurement data of a target station and a reference station in a specific wind farm within a first time period, and obtaining reanalysis data of the reference station within a second time period;

[0006] Obtaining relevant features of wind measurement data of the target station and the reference station within a first time period, and establishing a correlation model; and,

[0007] The association model and the reanalysis data of the reference station in the second time period are used to predict the long-term wind speed and wind speed distribution of the target station.

[0008] An embodiment of the present invention provides a long-term wind speed prediction system for a wind farm, comprising:

[0009] A data acquisition module is used to acquire wind measurement data of a target station and a reference station of a specific wind farm within a first time period, and to acquire reanalysis data of the reference station within a second time period;

[0010] A correlation model module is used to obtain relevant features of wind measurement data of the target station and the reference station in a first time period and establish a correlation model; and

[0011] The wind speed prediction module is used to predict the long-term wind speed and wind speed distribution of the target station by using the association model and the reanalysis data of the reference station in the second time period.

[0012] The embodiment of the present invention improves the correlation of wind data by studying the changing rules of the average power density and the average time period of a specific wind farm, thereby improving the prediction accuracy of the MCP model. In order to make the MCP method adapt to my country's complex resource environment, the method of the present invention proposes a combined application mode of a single method of a wind speed correlation model algorithm and a wind speed distribution fitting algorithm, so as to better obtain a better and more accurate effect of predicting the long-term wind speed and wind speed distribution of the target station in the process of MCP prediction and analysis. The method of the present invention can analyze and obtain the best MCP algorithm and average time period application mode for wind energy resource assessment suitable for my country's geographical distribution characteristics and regional differences in engineering practice. The present invention helps enterprises improve the prediction accuracy of long-term wind energy resource reserves in wind farms and reduce the risk of mismatch of wind turbine models, filling the gap in my country's lack of localized resource assessment models for wind farm resource simulation and assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate one or more embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0014] Figure 1 This is a flow chart of a method for predicting long-term wind speed in a wind farm according to an embodiment of the present invention;

[0015] Figure 2A flowchart of a specific implementation of the method for predicting long-term wind speed in a wind farm according to an embodiment of the present invention;

[0016] Figure 3 This is a diagram showing the effect of the long-term wind speed prediction accuracy of a wind farm in an embodiment of the present invention relative to a traditional average time period;

[0017] Figure 4 This is a diagram showing the effect of the long-term wind speed prediction error of a wind farm relative to a traditional average time period according to an embodiment of the present invention;

[0018] Figure 5 Schematic diagram of a long-term wind speed prediction system for a wind farm according to an embodiment of the present invention. DETAILED DESCRIPTION

[0019] In order to enable those skilled in the art to better understand the technical solutions in one or more embodiments of this specification, the following will be combined with the drawings in one or more embodiments of this specification to clearly and completely describe the technical solutions in one or more embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on one or more embodiments of this specification, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this document.

[0020] Method Embodiment

[0021] According to an embodiment of the present invention, a method for long-term wind speed prediction in a wind farm is provided. Figure 1 The flow chart of the method for predicting long-term wind speed in a wind farm according to the embodiment of the present invention is as follows: Figure 1 As shown, the long-term wind speed prediction method for a wind farm according to an embodiment of the present invention specifically includes:

[0022] S1, obtaining wind measurement data of a target station and a reference station of a specific wind farm within a first time period, and obtaining reanalysis data of the reference station within a second time period;

[0023] A specific implementation of S1 in the embodiment of the present invention is:

[0024] The European Centre for Medium-Range Weather Forecasts reanalysis data at an altitude of 100 m are used as reference station data, with a temporal resolution of 1 hour, a spatial resolution of 0.1° × 0.1°, and a time span of 10 years.

[0025] The measured data from wind towers at a height of more than 80 meters are used as the target station data, with a time resolution of 1 hour and a time span of 1 full year.

[0026] S2, obtaining relevant features of wind measurement data of the target station and the reference station within a first time period, and establishing a correlation model;

[0027] A specific implementation of S2 in the embodiment of the present invention is to respectively use the linear least squares method (LLS algorithm), the variance ratio algorithm (VR algorithm) and the Weibull fitting algorithm (WBL algorithm) to calculate the relevant features, and use the maximum likelihood method with unbiasedness and minimum statistical error to perform wind speed distribution fitting analysis, as follows:

[0028] The linear least squares method (LLS algorithm) calculation formula is:

[0029]

[0030] Among them: y represents the average wind speed measured by the wind tower, x represents the average wind speed of the reference station during the same period, and i represents the number of time steps in the corresponding average time period.

[0031] The calculation formula of the variance ratio algorithm (VR algorithm) is:

[0032]

[0033] Where: y represents the average wind speed measured by the wind tower, and x represents the average wind speed of the reference station during the same period. The VR algorithm is a variant of the simple linear regression model, which aims to maintain a simple linear mapping of the variance of the target data.

[0034] The calculation formula of Weibull fitting algorithm (WBL algorithm) is:

[0035]

[0036] Among them: y represents the average wind speed measured by the wind tower, x represents the average wind speed of the reference station during the same period, c represents the Weibull scale parameter, which reflects the degree of dispersion of the data, and k represents the shape parameter, which reflects the shape and tail characteristics of the Weibull distribution.

[0037] The present invention adopts the maximum likelihood method with unbiasedness and minimum statistical error to perform wind speed distribution fitting analysis, and its expression is:

[0038]

[0039] Where: c represents the Weibull scale parameter, reflecting the degree of dispersion of the data; k represents the shape parameter, reflecting the shape and tail characteristics of the Weibull distribution; U represents the average wind speed; N represents the number of time steps ... i represents the average wind speed corresponding to the i-th time step.

[0040] S3. Predict the long-term wind speed and wind speed distribution of the target station by using the correlation model and the reanalysis data of the reference station in the second time period.

[0041] A specific implementation of S3 in the embodiment of the present invention determines the prediction accuracy by using the average wind power density, the determination coefficient and the distribution error, as follows:

[0042] The average wind power density (WPD) is calculated as:

[0043] Where: n j Indicates that the average time period or time step is t j t represents the number of time steps. n1 represents the number of records with a time step of 1 hour in a year. j represents the average time period, i.e., the time step. R represents the gas constant (287 J / kg·K). ix Indicates that the average time period or time step is t j At time T, the recorded value of the xth average pressure within the ith average pressure. ix Indicates that the average time period or time step is t j When , the recorded value of the xth average temperature within the ith average temperature. V ix Indicates that the average time period or time step is t j At , the recorded value of the xth average wind speed within the ith average wind speed.

[0044] The coefficient of determination (R) was used to judge the fit of the model to the data. 2 )measure:

[0045] Where: y i represents the measured wind speed value of the target station i, represents the average wind speed measured at the target station, Represents the predicted value of the i-th data point of the target station. The coefficient of determination between the actual target wind speed value and the target wind speed predicted by the relevant algorithm during the same period, R 2 The closer the value is to 1, the better the model fits the data, the stronger the correlation between the actual value and the predicted result, and the higher the ability of the model to explain the data variation. 2 The closer the value is to 0, the poorer the model fits the data, the weaker the correlation between the predicted value and the actual value, and the lower the model's ability to explain data variation.

[0046] The calculation formula of the distribution error DE, which measures the matching degree between the predicted wind speed distribution and the actual wind speed distribution, is:

[0047] The closer the predicted distribution is to the actual distribution, the smaller the distribution error is. N represents the total number of wind speed groups at the target station, with an interval of 0.5 m / s; F i Represents the frequency of the ith group in the actual distribution; represents the frequency of the ith group of the predicted distribution.

[0048] The main technical solution of the method of the present invention is to perform correlation analysis on short-term wind measurement data and long-term reference data in a wind farm to establish a correlation relationship between the two, and then apply the long-term reference data to the relationship to predict the long-term average wind resource level at the short-term wind measurement tower.

[0049] The key to establishing a wind speed correlation model is that the actual wind measurement data of the target station during the same period and the long-term wind measurement data of the reference station show a strong similarity in the time series. Therefore, the data between the two wind measurement stations are required to be representative, accurate, stable and reliable. For example, the distance between the two stations cannot exceed 15 kilometers, the terrain is relatively similar, the climate characteristics are not much different or basically the same, and the distribution of wind resources is relatively similar.

[0050] Figure 2 The flowchart of the specific implementation of the method for predicting long-term wind speed in a wind farm according to the embodiment of the present invention is as follows: Figure 2 It can be seen that the execution steps of a specific example of the embodiment of the present invention are as follows:

[0051] 1. Measure: mainly collect and analyze wind measurement data, including measurement data of the same period and long-term data of reference stations; the accuracy, completeness and consistency of the data are important reference factors.

[0052] 2. Correlate: It is mainly to find the relationship between the measured wind data and the reference wind data, and to establish a correlation model by extracting the relevant features between wind speeds to facilitate prediction and analysis.

[0053] 3. Predict: It mainly uses the correlation model and long-term reference data to predict the long-term wind speed and wind speed distribution of the target station. The main factors affecting the prediction accuracy include the complexity of the model, the quality of the data and the selection of the algorithm.

[0054] Step 1 is as follows:

[0055] The wind speed time series of the target station wind tower and the reference station reanalysis data for a whole year are obtained. The average time period and the corresponding number of time steps are shown in Table 1.

[0056] Table 1 Analysis of average time period and number of steps

[0057]

[0058] Step 2 is as follows:

[0059] The relevant features between the wind measurement data of the target station and the reference station are extracted, and a wind speed correlation model is established.

[0060] The linear least squares method (LLS algorithm) calculation formula is:

[0061]

[0062] Among them: y represents the average wind speed measured by the wind tower, x represents the average wind speed of the reference station during the same period, and i represents the number of time steps in the corresponding average time period.

[0063] The calculation formula of the variance ratio algorithm (VR algorithm) is:

[0064]

[0065] Where: y represents the average wind speed measured by the wind tower, and x represents the average wind speed of the reference station during the same period. The VR algorithm is a variant of the simple linear regression model, which aims to maintain a simple linear mapping of the variance of the target data.

[0066] The calculation formula of Weibull fitting algorithm (WBL algorithm) is:

[0067]

[0068] Among them: y represents the average wind speed measured by the wind tower, x represents the average wind speed of the reference station during the same period, c represents the Weibull scale parameter, which reflects the degree of dispersion of the data, and k represents the shape parameter, which reflects the shape and tail characteristics of the Weibull distribution.

[0069] The present invention adopts the maximum likelihood method with unbiasedness and minimum statistical error to perform wind speed distribution fitting analysis, and its expression is:

[0070]

[0071] Where: c represents the Weibull scale parameter, reflecting the degree of dispersion of the data; k represents the shape parameter, reflecting the shape and tail characteristics of the Weibull distribution; U represents the average wind speed value; N represents the number of time steps; U represents the average wind speed value; N represents the number of time steps; U represents the average wind speed value; i represents the average wind speed corresponding to the i-th time step.

[0072] Step 3 is as follows:

[0073] The long-term wind speed time series and wind speed frequency distribution of the target station are predicted by using the correlation model and the 10-year reanalysis data of the reference station. The average wind power density WPD, the determination coefficient R 2 Indicators such as mean absolute error MAE, mean deviation error MBE, and distribution error DE quantify the prediction accuracy of various models and test the accuracy and reliability of the prediction.

[0074] The average wind power density (WPD) is calculated as:

[0075]

[0076] Where: n j Indicates that the average time period or time step is t j t represents the number of time steps. n1 represents the number of records with a time step of 1 hour in a year. j represents the average time period, i.e., the time step. R represents the gas constant (287 J / kg·K). ix Indicates that the average time period or time step is t j At time T, the recorded value of the xth average pressure within the ith average pressure. ix Indicates that the average time period or time step is t j When , the recorded value of the xth average temperature within the ith average temperature. V ix Indicates that the average time period or time step is t j At , the recorded value of the xth average wind speed within the ith average wind speed.

[0077] The coefficient of determination (R) was used to judge the fit of the model to the data. 2 )measure:

[0078]

[0079] Where: y i represents the i-th measured wind speed value of the target station, y represents the average measured wind speed of the target station, Represents the predicted value of the i-th data point of the target station. The coefficient of determination between the actual target wind speed value and the target wind speed predicted by the relevant algorithm during the same period, R 2 The closer the value is to 1, the better the model fits the data, the stronger the correlation between the actual value and the predicted result, and the higher the ability of the model to explain the data variation. 2 The closer the value is to 0, the poorer the model fits the data, the weaker the correlation between the predicted value and the actual value, and the lower the model's ability to explain data variation.

[0080] The calculation formula of the distribution error DE, which measures the matching degree between the predicted wind speed distribution and the actual wind speed distribution, is:

[0081] The closer the predicted distribution is to the actual distribution, the smaller the distribution error is. N represents the total number of wind speed groups at the target station, with an interval of 0.5 m / s; F i Represents the frequency of the ith group in the actual distribution; represents the frequency of the ith group of the predicted distribution.

[0082] Table 2 WPD and R changing with the average time period 2 、DE change analysis

[0083]

[0084]

[0085] Through experiments on wind farms with completely different topography and climate environments, it was found that when applying the linear least squares (LLS) method, variance ratio (VR) algorithm, and Weibull (WBL) fitting algorithm for research and analysis, a new method to improve the correlation of wind data is to change the traditional method of 1h averaging time period application mode: the 12h averaging time period application mode is not affected by the topography and climate environment; horizontal comparative analysis shows that the variance ratio (VR) algorithm is better than the linear least squares (LLS) method and the Weibull (WBL) fitting algorithm.

[0086] In the 12h average time period application model obtained by the new method research, the relative deviation of the average power density predicted by the least squares (LLS) method decreased the most, and the relative deviation of the average power density predicted by the Weibull (WBL) fitting algorithm decreased the second most; the variance ratio (VR) algorithm performed relatively stable, and the relative deviation of the predicted average power density was smaller.

[0087] Mean Absolute Error (MAE) is the average of the absolute errors between the predicted value and the actual value, which can help companies understand the degree of dispersion and deviation of the model's prediction results. The smaller the MAE value, the smaller the deviation between the model's prediction results and the actual value, and the higher the model's prediction accuracy. Its expression is:

[0088]

[0089] Where: y i represents the measured wind speed value of the target station i, It represents the predicted value of the i-th data point of the target station, and n represents the number of time steps of the measured data of the target station.

[0090] The mean bias error (MBE) is the average of the predicted value minus the actual value. The closer its value is to 0, the more accurate the model prediction is. It measures the average deviation in the prediction. Its expression is:

[0091]

[0092] Where: y i represents the measured wind speed value of the target station i, It represents the predicted value of the i-th data point of the target station, and n represents the number of time steps of the measured data of the target station.

[0093] Table 3 Error analysis of wind speed prediction data at target station

[0094]

[0095]

[0096] By analyzing the error of predicting the average wind speed through three different test data of the linear least squares (LLS) method, the variance ratio (VR) algorithm, and the Weibull (WBL) fitting algorithm, the following analysis is made: In general, the 12h average time period application mode obtained by the method of the present invention is significantly better than the traditional 1h average time period application mode; the relative reduction of the average absolute error of the average wind speed predicted by the three algorithms is basically the same, and the difference is not large. In horizontal comparison analysis, the variance ratio (VR) algorithm is better than the linear least squares (LLS) method and the Weibull (WBL) fitting algorithm. Figure 3 This is a diagram showing the effect of the long-term wind speed prediction accuracy of a wind farm in an embodiment of the present invention relative to the traditional average time period. Figure 4 This is a diagram showing the effect of the long-term wind speed prediction error of a wind farm relative to a traditional average time period according to an embodiment of the present invention.

[0097] By adopting the embodiments of the present invention, the following beneficial effects are achieved:

[0098] The embodiment of the present invention improves the correlation of wind data by studying the changing rules of the average power density and the average time period of a specific wind farm, thereby improving the prediction accuracy of the MCP model. In order to make the MCP method adapt to my country's complex resource environment, the method of the present invention proposes a combined application mode of a single method of a wind speed correlation model algorithm and a wind speed distribution fitting algorithm, so as to better obtain a better and more accurate effect of predicting the long-term wind speed and wind speed distribution of the target station in the process of MCP prediction and analysis. The method of the present invention can analyze and obtain the best MCP algorithm and average time period application mode for wind energy resource assessment suitable for my country's geographical distribution characteristics and regional differences in engineering practice. The present invention helps enterprises improve the prediction accuracy of long-term wind energy resource reserves in wind farms and reduce the risk of mismatch of wind turbine models, filling the gap in my country's lack of localized resource assessment models for wind farm resource simulation and assessment.

[0099] System Example

[0100] According to an embodiment of the present invention, a long-term wind speed prediction system for a wind farm is provided. Figure 5 The schematic diagram of the long-term wind speed prediction system for a wind farm according to an embodiment of the present invention is shown in FIG. Figure 5 As shown, the long-term wind speed prediction system for a wind farm according to an embodiment of the present invention specifically includes:

[0101] A data acquisition module 50 is used to acquire wind measurement data of a target station and a reference station of a specific wind farm within a first time period, and to acquire reanalysis data of the reference station within a second time period;

[0102] A correlation model module 52 is used to obtain the relevant features of the wind measurement data of the target station and the reference station in the first time period and establish a correlation model; and

[0103] The wind speed prediction module 54 is used to predict the long-term wind speed and wind speed distribution of the target station by using the correlation model and the reanalysis data of the reference station in the second time period.

[0104] This system embodiment is a device embodiment that corresponds one-to-one to the above method embodiment. For specific implementation, please refer to the above method embodiment, which will not be repeated here.

[0105] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. 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 replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for long-term wind speed prediction in a wind farm, characterized in that include: Obtain wind measurement data of a target station and a reference station of a specific wind farm within a first time period, and obtain reanalysis data of the reference station within a second time period; Obtaining relevant features of wind measurement data of the target station and the reference station within a first time period, and establishing a correlation model; and, The association model and the reanalysis data of the reference station in the second time period are used to predict the long-term wind speed and wind speed distribution of the target station.

2. The method according to claim 1, characterized in that The method further comprises: The prediction accuracy of the association model is quantified through preset indicators.

3. The method according to claim 2, characterized in that The preset indicators specifically include: Mean wind power density, coefficient of determination, distribution error, mean absolute error, and mean bias error.

4. The method according to claim 3, characterized in that: The average wind power density is obtained by formula 1: Among them, n j Indicates that the average time period or time step is t j The number of time steps corresponding to the time step; n1 represents the number of records with a time step of 1 hour in a year; t j represents the average time period, i.e., the time step; R represents the gas constant; P ix Indicates that the average time period or time step is t j The recorded value of the xth average pressure within the ith average pressure at time T ix Indicates that the average time period or time step is t j The recorded value of the xth average temperature within the ith average temperature at time V ix Indicates that the average time period or time step is t j The recorded value of the xth average wind speed within the ith average wind speed at time ; The coefficient of determination is obtained by formula 2: Among them, y i represents the measured wind speed value of the target station i; Indicates the average wind speed measured at the target station; represents the predicted value of the i-th data point at the target station; The distribution error is obtained by formula 3: Where N represents the total number of wind speed groups at the target station, with an interval of 0.5 m / s; F i Represents the frequency of the ith group in the actual distribution; represents the frequency of the i-th group of the predicted distribution; The mean absolute error is obtained by formula 4: Among them, y i represents the measured wind speed value of the target station i, represents the predicted value of the i-th data point at the target station, and n represents the number of time steps of the measured data at the target station; The mean deviation error is obtained by formula 5: Among them, yi represents the measured wind speed value of the target station i, It represents the predicted value of the i-th data point of the target station, and n represents the number of time steps of the measured data of the target station.

5. The method according to claim 1, characterized in that The obtaining of relevant features of the wind measurement data of the target station and the reference station within the first time period and establishing the correlation model specifically includes: The linear least square method, the variance ratio algorithm and the Weibull algorithm are respectively used to obtain the relevant characteristics of the wind measurement data of the target station and the reference station in the first time period; The Weibull fitting was used to establish the correlation models.

6. The method according to claim 5, characterized in that The linear least squares calculation formula is: Where y represents the average wind speed measured by the wind tower; x represents the average wind speed of the reference station during the same period; i represents the number of time steps in the corresponding average time period; b represents the intercept of the linear least squares algorithm, and m represents the slope of the linear least squares algorithm. represents the average of the observed reference wind speeds, represents the average of the observed target (measured) wind speed, X i represents the reference wind speed corresponding to the i-th time step, Y i represents the target (measured) wind speed corresponding to the i-th time step, S XX represents the deviation of the observed reference wind speed, S XY Indicates the deviation from the observed target (measured) wind speed.

7. The method according to claim 5, characterized in that The variance ratio algorithm is obtained by formula 10: Among them, y represents the average wind speed measured by the wind tower, and x represents the average wind speed of the reference station during the same period. represents the average of the observed target (measured) wind speed, represents the average of the observed reference wind speeds, represents the average of the predicted target wind speed, m represents the slope of the variance ratio algorithm, σ y represents the standard deviation of the observed target (measured) wind speed, σ x Represents the standard deviation of the observed reference wind speeds.

8. The method according to claim 5, characterized in that The Weibull algorithm is obtained by formula 11 to formula 13: Where y represents the average wind speed measured by the wind tower, x represents the average wind speed of the reference station during the same period, c represents the Weibull scale parameter, k represents the shape parameter, β represents the ratio of the Weibull shape factor of the reference station to the Weibull shape factor of the target station, α represents the ratio of the Weibull scale factor of the target station to the Weibull scale factor of the reference station, and k x represents the reference station Weibull shape factor, k y represents the target station Weibull shape factor, represents the predicted target wind speed, c y represents the Weibull scaling factor of the target station, represents the Weibull scaling factor of the reference station raised to exponent β.

9. The method according to claim 1, characterized in that: The long-term wind speed prediction method for a wind farm performs wind speed distribution fitting analysis by using the maximum likelihood method, specifically including: Where c represents the Weibull scale parameter, k represents the shape parameter, U represents the average wind speed, N represents the number of time steps, and U i represents the average wind speed corresponding to the i-th time step.

10. A long-term wind speed prediction system for a wind farm, characterized in that include: A data acquisition module is used to acquire wind measurement data of a target station and a reference station of a specific wind farm within a first time period, and to acquire reanalysis data of the reference station within a second time period; A correlation model module is used to obtain relevant features of wind measurement data of the target station and the reference station in a first time period and establish a correlation model; and The wind speed prediction module is used to predict the long-term wind speed and wind speed distribution of the target station by using the association model and the reanalysis data of the reference station in the second time period.

Images