CN113988360A - A wind power prediction method and device based on wind speed fluctuation feature classification - Google Patents

Abstract

The invention discloses a wind power prediction method and device based on the classification of wind speed fluctuation characteristics. The method divides the fluctuation process of wind speed and wind power by analyzing the fluctuation characteristics of wind speed and wind power, based on the degree of wind speed and wind power fluctuation. Based on the correlation of weather fluctuation process, the weather fluctuation process is classified. On this basis, prediction models suitable for different weather fluctuation processes are established. Finally, the prediction results output by different prediction models are rearranged in time series to obtain the combined prediction value of wind power. The invention takes into account the correlation between the wind speed fluctuation feature and the prediction model to perform classification prediction, establishes a combined model to improve the versatility of the model, and further improves the prediction accuracy of wind power.

Description

A wind power prediction method and device based on wind speed fluctuation feature classification

technical field

The invention relates to a wind power prediction method and device based on wind speed fluctuation feature classification, and belongs to the technical field of wind power prediction.

Background technique

The strong volatility, randomness and intermittency of wind energy determine the strong volatility of wind power. After the large-scale integration of wind power into the power grid, this fluctuation characteristic will inevitably bring great challenges to the safe and stable operation of the power grid. However, the fluctuation of wind speed Changes are not completely random, but still have certain regularity. Therefore, high-precision prediction of wind speed and wind power is very important.

At present, most of the main methods of short-term wind power forecasting in China are based on time series methods, artificial intelligence algorithms, etc. These methods build models based on the linear or nonlinear relationship between historical wind speed and power data, and then forecast wind power. However, they ignore the The correlation between the wind speed fluctuation characteristics and the prediction model leads to a large prediction error of a single prediction model.

SUMMARY OF THE INVENTION

In order to solve the problems of low accuracy of the prediction model due to the neglect of the correlation between the fluctuation characteristics of wind speed and the prediction model in the existing wind power prediction methods, the invention provides a wind power prediction method based on the classification of wind speed fluctuation characteristics. And the device, considering the correlation between the wind speed fluctuation characteristics and the prediction model to carry out classification prediction, establish a combined model to improve the versatility of the model, and then improve the prediction accuracy of wind power.

The technical scheme adopted in the present invention is as follows:

The present invention provides a wind power prediction method based on wind speed fluctuation feature classification, comprising:

Classification of wind speed fluctuation characteristics based on historical wind speed data of meteorological forecasts; and classification of wind power fluctuation characteristics based on historical wind power data of wind farm units; wherein, a fluctuation process is: from a local minimum value less than a certain wind speed/wind power threshold It starts from the value of the wind speed, passes through at least one wave peak in sequence, and then returns to the local minimum value less than the wind speed/wind power threshold value and ends;

Divide different types of weather processes based on the matching between wind speed fluctuation feature types and wind power fluctuation feature types;

Establish wind power prediction models for different types of weather processes, and train the established models based on historical data;

Input the real-time wind speed data into the wind power prediction model corresponding to the weather process according to the type of wind speed fluctuation characteristics;

The wind power forecast results of different types of weather processes are reorganized in time series to obtain the combined forecast value of wind power.

further,

Normalize historical wind speed data and historical wind power data.

Further, the classification of the wind speed fluctuation characteristics based on the historical wind speed data of the meteorological forecast includes:

为单个风速波动过程内的最大峰值，λ＝1,2,…，为波峰序列，

为第λ个波峰的峰值，ε_v,0为归一化后的风电场切入风速，ε_v,1和ε_v,2分别是不同风速波动过程类型的判断阈值，W_v,0，W_v,1，W_v,2和W_v,3分别为零输出风速波动过程，风速小波动过程，风速中波动过程和风速大波动过程。in,

is the maximum peak value in a single wind speed fluctuation process, λ=1,2,…, is the wave peak sequence,

is the peak value of the λth wave peak, ε _v,0 is the normalized cut-in wind speed of the wind farm, ε _v,1 and ε _v,2 are the judgment thresholds for different wind speed fluctuation process types, W _v,0 , W _{v ,1} , W _v,2 and W _v,3 are respectively zero output wind speed fluctuation process, small wind speed fluctuation process, medium wind speed fluctuation process and large wind speed fluctuation process.

Further, the ε _v,1 ∈ [0.2, 0.4], ε _v,2 ∈ [0.4, 0.6].

Further, the classification of wind power fluctuation characteristics based on historical wind power data of wind farm units includes:

为单个功率波动过程内的波峰值，α＝1,2,…，α为波峰序列，

为第α个波峰的峰值，ε_p,1和ε_p,2分别是不同功率波动过程类型的判断阈值，W_p,1，W_p,2和W_p,3分别为功率小波动过程，功率中波动过程和功率大波动过程。in,

is the wave peak value in a single power fluctuation process, α=1,2,..., α is the wave peak sequence,

is the peak value of the αth peak, ε _p,1 and ε _p,2 are the judgment thresholds for different power fluctuation process types, respectively, W _p,1 , W _p,2 and W _p,3 are the small power fluctuation processes, respectively. Medium fluctuation process and large power fluctuation process.

Further, the ε _v,1 ∈ [0.2, 0.4], ε _v,2 ∈ [0.4, 0.6].

Further, different types of weather processes are divided based on the matching between the wind speed fluctuation feature type and the wind power fluctuation feature type, including:

The small fluctuation process of wind speed matches the small fluctuation process of power, and is divided into small fluctuation weather process;

The fluctuating process in wind speed matches the fluctuating process in power, and is divided into moderate fluctuation weather process;

The large fluctuation process of wind speed matches the large fluctuation process of power, and is divided into large fluctuation weather process.

Further, the wind power prediction models are respectively established for different types of weather processes, including:

For the small fluctuation weather process, the autoregressive moving average method is used to establish a prediction model based on the wind speed data of the small fluctuation process of wind speed and the wind power data of the small power fluctuation process;

For the moderate fluctuation weather process and the large fluctuation weather process, the least squares support vector machine method is used based on the wind speed data of the wind speed fluctuation process and the wind power of the power fluctuation process, and the wind speed data of the wind speed fluctuation process and the power fluctuation process. Wind power data to build a forecasting model.

The present invention also provides a wind power prediction device based on wind speed fluctuation feature classification, comprising:

The classification module is used to classify the wind speed fluctuation characteristics based on the historical wind speed data of meteorological forecast; and classify the wind power fluctuation characteristics based on the historical wind power data of the wind farm units; wherein, one fluctuation process is: from less than a certain wind speed/wind power Start from the local minimum value of the power threshold, pass through at least one peak in sequence, and then return to the local minimum value smaller than the wind speed/wind power threshold value and end;

a matching module for dividing different types of weather processes based on the matching between wind speed fluctuation feature types and wind power fluctuation feature types;

The model module is used to establish wind power prediction models for different types of weather processes, and train the established models based on historical data;

as well as,

The recombination module is used to reorganize the wind power prediction results of different types of weather processes in time series to obtain the combined forecast value of wind power.

Further, the classification module is specifically used for,

The wind speed fluctuation characteristics are classified as follows:

为单个风速波动过程内的最大峰值，λ＝1,2,…，为波峰序列，

为第λ个波峰的峰值，ε_v,0为归一化后的风电场切入风速，ε_v,1和ε_v,2分别是不同风速波动过程类型的判断阈值，W_v,0，W_v,1，W_v,2和W_v,3分别为零输出风速波动过程，风速小波动过程，风速中波动过程和风速大波动过程；in,

is the maximum peak value in a single wind speed fluctuation process, λ=1,2,…, is the wave peak sequence,

is the peak value of the λth wave peak, ε _v,0 is the normalized cut-in wind speed of the wind farm, ε _v,1 and ε _v,2 are the judgment thresholds for different wind speed fluctuation process types, W _v,0 , W _{v ,1} , W _v,2 and W _v,3 are respectively zero output wind speed fluctuation process, small wind speed fluctuation process, medium wind speed fluctuation process and large wind speed fluctuation process;

The characteristics of wind power fluctuations are classified as follows:

为单个功率波动过程内的波峰值，α＝1,2,…，α为波峰序列，

为第α个波峰的峰值，ε_p,1和ε_p,2分别是不同功率波动过程类型的判断阈值，W_p,1，W_p,2和W_p,3分别为功率小波动过程，功率中波动过程和功率大波动过程。in,

is the wave peak value in a single power fluctuation process, α=1,2,..., α is the wave peak sequence,

is the peak value of the αth peak, ε _p,1 and ε _p,2 are the judgment thresholds for different power fluctuation process types, respectively, W _p,1 , W _p,2 and W _p,3 are the small power fluctuation processes, respectively. Medium fluctuation process and large power fluctuation process.

Further, the matching module is specifically used for,

Matching the small fluctuation process of wind speed with the small fluctuation process of power, it is divided into small fluctuation weather process;

Match the fluctuating process in wind speed with the fluctuating process in power, and divide it into moderate fluctuation weather process;

The large fluctuation process of wind speed is matched with the large fluctuation process of power, and it is divided into large fluctuation weather process.

Further, the model module is specifically used for,

For the small fluctuation weather process, the autoregressive moving average method is used to establish a prediction model based on the wind speed data of the small fluctuation process of wind speed and the wind power data of the small power fluctuation process;

For the moderate fluctuation weather process and the large fluctuation weather process, the least squares support vector machine method is used based on the wind speed data of the wind speed fluctuation process and the wind power of the power fluctuation process, and the wind speed data of the wind speed fluctuation process and the power fluctuation process. Wind power data to build a forecasting model.

The beneficial effects of the present invention are:

Based on the analysis of wind speed fluctuation characteristics, the invention establishes wind power prediction models for different types of weather processes respectively, and reorganizes the wind power prediction results of different types of weather processes in time series to obtain a combined prediction value of wind power. The invention takes into account the correlation between the wind speed fluctuation feature and the prediction model to perform classification prediction, establishes a combined model to improve the versatility of the model, and further improves the prediction accuracy of wind power.

Description of drawings

Fig. 1 is the overall implementation flow chart of wind power prediction based on wind speed fluctuation feature classification according to the present invention;

Fig. 2 is ARMA model establishment flow chart among the present invention;

FIG. 3 is a flowchart of the LSSVM algorithm in the present invention.

Detailed ways

The present invention is further described below. The following examples are only used to illustrate the technical solutions of the present invention more clearly, and cannot be used to limit the protection scope of the present invention.

The invention provides a wind power prediction method based on wind speed fluctuation feature classification. This method takes into account the influence of the fluctuation characteristics of the weather on the prediction accuracy of wind power. By analyzing the fluctuation characteristics of wind speed and wind power, the fluctuation process of wind speed and wind power is divided. Based on the correlation between wind speed and wind power fluctuation, The weather fluctuation process is classified, and on this basis, a prediction model suitable for different weather fluctuation processes is established to reduce the prediction error. The specific implementation steps are shown in Figure 1, including:

Step 1: Select wind speed, one of the main meteorological factors in the meteorological characteristic factors of numerical weather forecast, as the basic meteorological factor for weather fluctuation process division, analyze the periodicity and regularity of historical wind speed fluctuations, and identify the fluctuation characteristics of historical wind speed based on the identification of historical wind speed. , to classify the types of wind speed fluctuations;

And, based on the historical operation data of the wind farm, the fluctuation characteristics of wind power are studied, and the fluctuation types of wind power are classified.

Specifically, first normalize the historical wind speed/wind power data to facilitate subsequent identification of fluctuation features;

Secondly, a fluctuation process of wind speed/wind power is defined as: starting from a local minimum value less than a certain wind speed/wind power threshold, passing through single or multiple peaks in turn, and returning to the local minimum value less than the wind speed/wind power threshold The value is over, taking the wind speed fluctuation process as an example, the mathematical expression is as follows:

和

分别为一个波动过程的始末值，

为

和

中的较大值，且三者都小于等于归一化后的风电场切入风速ε_v,0，

为单个风速波动过程中的波峰值，λ为峰值序列，n(·)为峰值数量统计函数，一个完整的风速波动过程至少含有一个波动峰值。in,

and

are the beginning and end values of a fluctuation process, respectively,

for

and

The larger value of , and the three are less than or equal to the normalized wind farm cut-in wind speed ε _v,0 ,

is the wave peak value in a single wind speed fluctuation process, λ is the peak value sequence, n( ) is the peak number statistical function, and a complete wind speed fluctuation process contains at least one fluctuation peak value.

The amplitude, duration and number of peaks of the wind speed/wind power fluctuation process divided by the above rules are quite different. Therefore, the wind speed/wind power fluctuation process needs to be classified according to specific parameters. Taking the division of wind speed fluctuation process as an example, it can be It is expressed as follows:

为单个风速波动过程内的最大峰值，ε_v,1和ε_v,2分别是不同风速波动过程类型的判断阈值，W_v,0，W_v,1，W_v,2，W_v,3分别为零输出风速波动过程、风速小波动过程、风速中波动过程，风速大波动过程。参考相关文献，ε_v,1∈[0.2，0.4]，ε_v,2∈[0.4，0.6]，考虑风电场具体情况，可选取ε_v,1＝0.25，ε_v,2＝0.5。in,

is the maximum peak value in a single wind speed fluctuation process, ε _v,1 and ε _v,2 are the judgment thresholds for different wind speed fluctuation process types, W _v,0 , W _v,1 , W _v,2 , W _v,3 respectively Zero output wind speed fluctuation process, small wind speed fluctuation process, medium wind speed fluctuation process, and large wind speed fluctuation process. Referring to relevant literature, ε _v,1 ∈ [0.2, 0.4], ε _v,2 ∈ [0.4, 0.6], considering the specific conditions of wind farms, ε _v,1 =0.25, ε _v,2 =0.5 can be selected.

The division principle of wind power fluctuation process is the same as above, which can be expressed as:

为单个功率波动过程内的波峰值；α为峰值序列，与风速波峰序列相似，一个完整的功率波动过程至少含有一个波动峰值，实际功率波动过程中往往不仅只包含一个波峰；ε_p,1和ε_p,2分别是不同功率波动过程类型的判断阈值，W_p,1，W_p,2，W_p,3分别为功率小波动过程、功率中波动过程，功率大波动过程。参考相关文献，ε_p,1∈[0.2，0.4]，ε_p,2∈[0.4，0.6]，考虑风电场具体情况，可选取ε_p,1＝0.25，ε_p,2＝0.5。in,

is the peak value in a single power fluctuation process; α is the peak sequence, similar to the wind speed peak sequence, a complete power fluctuation process contains at least one fluctuation peak, and the actual power fluctuation process often contains more than one peak; ε _p,1 and ε _p,2 are the judgment thresholds for different power fluctuation process types, W _p,1 , W _p,2 , and W _p,3 are the small power fluctuation process, the medium power fluctuation process, and the large power fluctuation process, respectively. Referring to relevant literature, ε _p,1 ∈ [0.2, 0.4], ε _p,2 ∈ [0.4, 0.6], considering the specific conditions of the wind farm, ε _p,1 =0.25, ε _p,2 =0.5 can be selected.

Step 2: Statistical correlation between wind speed fluctuation characteristics and wind power fluctuation characteristics at the same time period of the wind farm, and classify different types of weather processes according to the matching of fluctuation laws;

Specifically, the wind power fluctuation types under different wind speed fluctuation processes in the selected wind farm are statistically analyzed within a period of time, and the zero output power section corresponding to the zero output wind speed fluctuation process is not considered according to the statistical results. The process of small fluctuation mainly matches the process of small fluctuation of power, the process of fluctuation in wind speed mainly matches the process of fluctuation in power, and often also matches the process of large fluctuation of power. The process of large fluctuation of wind speed mainly matches the process of large fluctuation of power. Therefore, Based on the matching of different wind speed fluctuation processes and different power fluctuation processes in the above analysis, the division results of wind speed fluctuation processes are used as the basis for the division of different types of weather processes, and different types of weather processes are obtained. basis.

Step 3: According to the division result, establish a wind power prediction method suitable for different weather processes;

Specifically, the fluctuation characteristics under different weather fluctuation types are analyzed respectively, and the corresponding prediction models are established. For the types of small weather fluctuations corresponding to the types of small wind speed fluctuations, the Autoregressive moving average model (ARMA) is selected for prediction. The main reason is that the autoregressive moving average method has a simple model and high calculation efficiency. , the prediction of wind power under the weather type with small fluctuation range; for the type of weather fluctuation corresponding to the fluctuation type of wind speed and the type of large weather fluctuation corresponding to the type of large wind speed fluctuation, the least squares support vector machine (least squares support vector machine) LSSVM) for prediction, the main reason is that the least squares support vector machine has fast training speed and strong generalization ability, and is suitable for weather fluctuation types with large fluctuation amplitude and large sample size.

The autoregressive moving average model ARMA(n, m) is established for the output power data of the wind turbine as follows:

和θ都是不为零的待定系数，

为自回归参数，θ_j(j＝1,2,…,n)为移动平均参数；{α_t}为单独的误差项序列，其均值为零；x_t是平稳的时间序列，在此模型中，x_t为t时刻的风速，x_t-i为t-i时刻的风速。Among them, the model contains n autoregressive terms and m moving average terms, and n and m are the autoregressive order and moving average order of the model, respectively;

and θ are both non-zero undetermined coefficients,

is the autoregressive parameter, θ _j (j=1,2,…,n) is the moving average parameter; {α _t } is the sequence of separate error terms, and its mean is zero; x _t is the stationary time series, in this model where x _t is the wind speed at time t, and x _ti is the wind speed at time ti.

The specific prediction process of ARMA is shown in Figure 2, as follows:

(1) Judging whether the input time series is stable, if not, perform differential processing to make it tend to zero mean and stabilize;

(2) Establish an ARMA model, gradually increase the model order, fit the ARMA(n,n-1) model, use the nonlinear least squares method to estimate the model parameters, select the model corresponding to the minimum variance of the residual sequence as the primary selection model, and determine Order;

(3) Model adaptability test, judge whether the model passes the parameter significance test and residual test, if not, then re-determine the difference order and repeat step (2).

LSSVM maps samples into a high-dimensional feature space through a nonlinear mapping function, and records the nonlinear relationship between the input and output of training samples. Therefore, when the training data and prediction data are of the same type, the prediction accuracy can be effectively improved. The nonlinear function of LSSVM is:

Among them, y(x) is the output vector of the model, that is, the wind power value, x _i (i=1,2,...,t) is the correlation vector of the prediction result, which is the wind speed value in this model, and t is the training set time point The total number; a _i is the Lagrange multiplier, b is the deviation, and K(x, x _i ) is the kernel function, all three need to solve the optimal value during the training process.

The specific prediction steps of LSSVM are shown in Figure 3, as follows:

The first step is to normalize the historical wind speed and wind power data, and normalize all the values with different dimension indicators to [-1, 1];

The second step is to select appropriate LSSVM kernel functions and parameters for training according to the fluctuation law of wind speed;

In the third step, according to the error size obtained from the training data, select the appropriate kernel function and the corresponding parameters according to the cross-validation method;

The fourth step is to re-train with the selected kernel function and parameters, and test the prediction results. If the test is passed, the model establishment is completed, and if it fails, the third step is repeated;

The fifth step is to input the real-time wind speed data into the trained model for prediction, and obtain the final prediction result.

Step 4: Reorganize the forecast results of different models in time series to obtain the combined forecast value of wind power;

Based on the classification of the wind speed time series, different prediction models are selected according to the fluctuation law of wind speed, and finally the prediction values of the separated models are combined in time series to obtain the final output result.

Another embodiment of the present invention provides a wind power prediction device based on wind speed fluctuation feature classification, including:

The classification module is used to classify the wind speed fluctuation characteristics based on the historical wind speed data of meteorological forecast; and classify the wind power fluctuation characteristics based on the historical wind power data of the wind farm units; wherein, one fluctuation process is: from less than a certain wind speed/wind power Start from the local minimum value of the power threshold, pass through at least one peak in sequence, and then return to the local minimum value smaller than the wind speed/wind power threshold value and end;

a matching module for dividing different types of weather processes based on the matching between wind speed fluctuation feature types and wind power fluctuation feature types;

The model module is used to establish wind power prediction models for different types of weather processes, and train the established models based on historical data;

as well as,

The recombination module is used to reorganize the wind power prediction results of different types of weather processes in time series to obtain the combined forecast value of wind power.

In the embodiment of the present invention, the classification module is specifically used to:

The wind speed fluctuation characteristics are classified as follows:

为单个风速波动过程内的最大峰值，λ＝1,2,…，为波峰序列，

为第λ个波峰的峰值，ε_v,0为归一化后的风电场切入风速，ε_v,1和ε_v,2分别是不同风速波动过程类型的判断阈值，W_v,0，W_v,1，W_v,2和W_v,3分别为零输出风速波动过程，风速小波动过程，风速中波动过程和风速大波动过程；in,

is the maximum peak value in a single wind speed fluctuation process, λ=1,2,…, is the wave peak sequence,

is the peak value of the λth wave peak, ε _v,0 is the normalized cut-in wind speed of the wind farm, ε _v,1 and ε _v,2 are the judgment thresholds for different wind speed fluctuation process types, W _v,0 , W _{v ,1} , W _v,2 and W _v,3 are respectively zero output wind speed fluctuation process, small wind speed fluctuation process, medium wind speed fluctuation process and large wind speed fluctuation process;

The characteristics of wind power fluctuations are classified as follows:

为单个功率波动过程内的波峰值，α＝1,2,…，α为波峰序列，

为第α个波峰的峰值，ε_p,1和ε_p,2分别是不同功率波动过程类型的判断阈值，W_p,1，W_p,2和W_p,3分别为功率小波动过程，功率中波动过程和功率大波动过程。in,

is the wave peak value in a single power fluctuation process, α=1,2,..., α is the wave peak sequence,

is the peak value of the αth peak, ε _p,1 and ε _p,2 are the judgment thresholds for different power fluctuation process types, respectively, W _p,1 , W _p,2 and W _p,3 are the small power fluctuation processes, respectively. Medium fluctuation process and large power fluctuation process.

In the embodiment of the present invention, the matching module is specifically used to:

Matching the small fluctuation process of wind speed with the small fluctuation process of power, it is divided into small fluctuation weather process;

Match the fluctuating process in wind speed with the fluctuating process in power, and divide it into moderate fluctuation weather process;

The large fluctuation process of wind speed is matched with the large fluctuation process of power, and it is divided into large fluctuation weather process.

In the embodiment of the present invention, the model module is specifically used to:

For the small fluctuation weather process, the autoregressive moving average method is used to establish a prediction model based on the wind speed data of the small fluctuation process of wind speed and the wind power data of the small power fluctuation process;

For the moderate fluctuation weather process and the large fluctuation weather process, the least squares support vector machine method is used based on the wind speed data of the wind speed fluctuation process and the wind power of the power fluctuation process, and the wind speed data of the wind speed fluctuation process and the power fluctuation process. Wind power data to build a forecasting model.

It is worth noting that this apparatus embodiment corresponds to the foregoing method embodiment, and the implementation manners of the foregoing method embodiment are all applicable to this apparatus embodiment, and can achieve the same or similar technical effects, so they are not repeated here.

As will be appreciated by those skilled in the art, the embodiments of the present application may be provided as a method, a system, or a computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present application. It will be understood that each flow and/or block in the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to the processor of a general purpose computer, special purpose computer, embedded processor or other programmable data processing device to produce a machine such that the instructions executed by the processor of the computer or other programmable data processing device produce Means for implementing the functions specified in a flow or flow of a flowchart and/or a block or blocks of a block diagram.

These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory result in an article of manufacture comprising instruction means, the instructions The apparatus implements the functions specified in the flow or flow of the flowcharts and/or the block or blocks of the block diagrams.

These computer program instructions can also be loaded on a computer or other programmable data processing device to cause a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process such that The instructions provide steps for implementing the functions specified in the flow or blocks of the flowcharts and/or the block or blocks of the block diagrams.

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 them. Although the present invention has been described in detail with reference to the above embodiments, those of ordinary skill in the art should understand that: the present invention can still be Modifications or equivalent replacements are made to the specific embodiments of the present invention, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention shall be included within the protection scope of the claims of the present invention.

Claims (12)

1. a wind power prediction method based on wind speed fluctuation characteristic classification, is characterized in that, comprises: Classification of wind speed fluctuation characteristics based on historical wind speed data of meteorological forecasts; and classification of wind power fluctuation characteristics based on historical wind power data of wind farm units; wherein, a fluctuation process is: from a local minimum value less than a certain wind speed/wind power threshold It starts from the value of the wind speed, passes through at least one wave peak in sequence, and then returns to the local minimum value less than the wind speed/wind power threshold value and ends; Divide different types of weather processes based on the matching between wind speed fluctuation feature types and wind power fluctuation feature types; Establish wind power prediction models for different types of weather processes, and train the established models based on historical data; Input the real-time wind speed data into the wind power prediction model corresponding to the weather process according to the type of wind speed fluctuation characteristics; The wind power forecast results of different types of weather processes are reorganized in time series to obtain the combined forecast value of wind power. 2. a kind of wind power prediction method based on wind speed fluctuation characteristic classification according to claim 1, is characterized in that, also comprises: Normalize historical wind speed data and historical wind power data. 3. a kind of wind power prediction method based on wind speed fluctuation feature classification according to claim 2, is characterized in that, described based on meteorological forecast historical wind speed data to classify wind speed fluctuation feature, comprising:

in,

is the maximum peak value in a single wind speed fluctuation process, λ=1,2,…, is the wave peak sequence,

is the peak value of the λth wave peak, ε _v,0 is the normalized cut-in wind speed of the wind farm, ε _v,1 and ε _v,2 are the judgment thresholds for different wind speed fluctuation process types, W _v,0 , W _{v ,1} , W _v,2 and W _v,3 are respectively zero output wind speed fluctuation process, small wind speed fluctuation process, medium wind speed fluctuation process and large wind speed fluctuation process. 4. A wind power prediction method based on wind speed fluctuation feature classification according to claim 3, wherein the ε _v,1 ∈ [0.2, 0.4], ε _v,2 ∈ [0.4, 0.6] . 5. A wind power prediction method based on wind speed fluctuation feature classification according to claim 1, wherein the classification of wind power fluctuation features based on historical wind power data of wind farm units comprises:

in,

is the wave peak value in a single power fluctuation process, α=1,2,..., α is the wave peak sequence,

is the peak value of the αth peak, ε _p,1 and ε _p,2 are the judgment thresholds for different power fluctuation process types, respectively, W _p,1 , W _p,2 and W _p,3 are the small power fluctuation processes, respectively. Medium fluctuation process and large power fluctuation process. 6. A wind power prediction method based on wind speed fluctuation feature classification according to claim 5, wherein the ε _v,1 ∈ [0.2, 0.4], ε _v,2 ∈ [0.4, 0.6] . 7 . A wind power prediction method based on wind speed fluctuation feature classification according to claim 1 , wherein the weather process of different types is divided based on the matching between wind speed fluctuation feature types and wind power fluctuation feature types. 8 . ,include: The small fluctuation process of wind speed matches the small fluctuation process of power, and is divided into small fluctuation weather process; The fluctuating process in wind speed matches the fluctuating process in power, and is divided into moderate fluctuation weather process; The large fluctuation process of wind speed matches the large fluctuation process of power, and is divided into large fluctuation weather process. 8. A wind power prediction method based on wind speed fluctuation feature classification according to claim 7, wherein the wind power prediction models are respectively established for different types of weather processes, comprising: For the small fluctuation weather process, the autoregressive moving average method is used to establish a prediction model based on the wind speed data of the small fluctuation process of wind speed and the wind power data of the small power fluctuation process; For the moderate fluctuation weather process and the large fluctuation weather process, the least squares support vector machine method is used based on the wind speed data of the wind speed fluctuation process and the wind power of the power fluctuation process, and the wind speed data of the wind speed fluctuation process and the power fluctuation process. Wind power data to build a forecasting model. 9. A wind power prediction device based on wind speed fluctuation feature classification, characterized in that it comprises: The classification module is used to classify the wind speed fluctuation characteristics based on the historical wind speed data of meteorological forecast; and classify the wind power fluctuation characteristics based on the historical wind power data of the wind farm units; wherein, one fluctuation process is: from less than a certain wind speed/wind power Start from the local minimum value of the power threshold, pass through at least one peak in sequence, and then return to the local minimum value smaller than the wind speed/wind power threshold value and end; a matching module for dividing different types of weather processes based on the matching between wind speed fluctuation feature types and wind power fluctuation feature types; The model module is used to establish wind power prediction models for different types of weather processes, and train the established models based on historical data; as well as, The recombination module is used to reorganize the wind power prediction results of different types of weather processes in time series to obtain the combined forecast value of wind power. 10. A wind power prediction device based on wind speed fluctuation feature classification according to claim 9, wherein the classification module is specifically used to: The wind speed fluctuation characteristics are classified as follows:

in,

is the maximum peak value in a single wind speed fluctuation process, λ=1,2,…, is the wave peak sequence,

is the peak value of the λth wave peak, ε _v,0 is the normalized cut-in wind speed of the wind farm, ε _v,1 and ε _v,2 are the judgment thresholds for different wind speed fluctuation process types, W _v,0 , W _{v ,1} , W _v,2 and W _v,3 are respectively zero output wind speed fluctuation process, small wind speed fluctuation process, medium wind speed fluctuation process and large wind speed fluctuation process; The characteristics of wind power fluctuations are classified as follows:

in,

is the wave peak value in a single power fluctuation process, α=1,2,..., α is the wave peak sequence,

is the peak value of the αth peak, ε _p,1 and ε _p,2 are the judgment thresholds for different power fluctuation process types, respectively, W _p,1 , W _p,2 and W _p,3 are the small power fluctuation processes, respectively. Medium fluctuation process and large power fluctuation process. 11. The wind power prediction device based on wind speed fluctuation feature classification according to claim 10, wherein the matching module is specifically used to: Matching the small fluctuation process of wind speed with the small fluctuation process of power, it is divided into small fluctuation weather process; Match the fluctuating process in wind speed with the fluctuating process in power, and divide it into moderate fluctuation weather process; The large fluctuation process of wind speed is matched with the large fluctuation process of power, and it is divided into large fluctuation weather process. 12 . The wind power prediction device based on wind speed fluctuation feature classification according to claim 11 , wherein the model module is specifically used for: 12 . For the small fluctuation weather process, the autoregressive moving average method is used to establish a prediction model based on the wind speed data of the small fluctuation process of wind speed and the wind power data of the small power fluctuation process; For the moderate fluctuation weather process and the large fluctuation weather process, the least squares support vector machine method is used based on the wind speed data of the wind speed fluctuation process and the wind power of the power fluctuation process, and the wind speed data of the wind speed fluctuation process and the power fluctuation process. Wind power data to build a forecasting model.

Images