CN106383932A - Wind power prediction method - Google Patents

Abstract

The invention discloses a wind power prediction method. A device comprises a combined prediction module, specifically comprises an NARX model prediction unit and an SVR model prediction unit which inputs a wind speed, a wind direction, an air temperature, an air pressure, a humidity sequence and a wind power sequence respectively, and further comprises a maximum information entropy prediction module which establishes feedback with the combined prediction module. The wind power can be predicted on the basis of determining weight values of models in the combined prediction module, so that the prediction precision is improved.

Description

A kind of Forecasting Methodology of wind power

Technical field

The present invention relates to a kind of combination forecasting method of wind power prediction.

Background technology

During wind power prediction, the data of every kind of Forecasting Methodology application is roughly the same.Individual forecast method has The not good shortcoming of precision, application is less；If but just abandoning using it is possible to lead to because certain individual event Forecasting Methodology error is larger Part useful information is lost.In combination forecasting method, the angle that distinct methods improve useful information is different, causes prediction side The pluses and minuses of method are also different.

Existing wind power forecasting method is varied, but has that Individual forecast method precision is not high, and neutral net is pre- Survey the shortcoming being easily trapped into local optimum.Combination forecasting is now into the upsurge of research, but multiple model combines Combination forecasting in be related to the problem how weight of single model determines, different determination methods have impact on prediction Precision height.

Content of the invention

The purpose of the present invention is for overcoming the problems referred to above, proposing a kind of combination forecasting method of wind power, in determination group Predict wind power on the basis of closing the weights of each model in forecast model, improve precision of prediction.

Technical scheme equipment includes

Combined prediction module, specifically includes input wind speed, wind direction, temperature, air pressure, humidity sequence and wind power sequence respectively The NARX model prediction unit of row and SVR model prediction unit, also have the maximum letter setting up feedback with combined prediction module Breath entropy prediction module；Specific prediction steps are as follows：

The first step, independent prediction：NARX model prediction unit, SVR model prediction unit reception system detecting information are simultaneously divided Other computing draws wind power prediction result, and combined prediction module calculates the average of the two result

Second step, determines centre-to-centre spacing：Combined prediction module predicts that t lights the wind-powered electricity generation in common n moment Performance numberAnd calculate its numerical characteristic e further_tm, e_tmMeetDetermine the k rank centre-to-centre spacing of the pre- power scale of wind-powered electricity generation

3rd step, combined prediction：Maximum informational entropy prediction module according to model prediction unit number n, pre- power scale each Rank centre-to-centre spacingCalculate weight coefficient p in final result for each model prediction unit result_i, meet

$p_i=\exp (-{\mathrm{\λ}}_0-\underset{k=1}{\overset{K}{\Σ}}{\mathrm{\λ}}_k{\left(\frac{x-{\stackrel{\‾}{y}}_i}{{\stackrel{\‾}{y}}_i}\right)}^2),$

Due to

${\mathrm{\λ}}_0=ln\[\underset{i=1}{\overset{N}{\Σ}}\exp (-\underset{k=1}{\overset{K}{\Σ}}{\mathrm{\λ}}_k{\left(\frac{x-{\stackrel{\‾}{y}}_i}{{\stackrel{\‾}{y}}_i}\right)}^2)\]$

And then can obtain

$\underset{i=1}{\overset{N}{\Σ}}\exp (-\underset{k=1}{\overset{K}{\Σ}}{\mathrm{\λ}}_k{\left(\frac{x-{\stackrel{\‾}{y}}_i}{{\stackrel{\‾}{y}}_i}\right)}^2)({\left(\frac{x-{\stackrel{\‾}{y}}_i}{{\stackrel{\‾}{y}}_i}\right)}^2-E\[{\left(\frac{x-{\stackrel{\‾}{y}}_i}{{\stackrel{\‾}{y}}_i}\right)}^2\])=0,$

And calculate final wind power prediction value x, meet

$\max \left(X\right(X\left)\right)=-\underset{i=1}{\overset{n}{\Σ}}{p}_i\log p_i.$

Module internal program goes out λ by above formula Equation for Calculating_k(k=1,2 ..., K), then according to λ_kλ is tried to achieve in calculating₀；According to λ₀, λ₁..., λ_kIt is calculated p_i, finally it is calculated H (X).Carry out the determination of weight by the method, the side such as averaging method making up Method determines the deficiency of weighted value, has higher generalization ability, improves precision of prediction.

Brief description

Fig. 1 is the flow chart of the present invention.

Fig. 2 is the historical power curve of the present invention and the power curve of prediction.

Specific embodiment and effect explanation

In order that technological means, creation characteristic, reached purpose and effect that the present invention realizes are easy to understand, tie below Close diagram and specific embodiment, the present invention is expanded on further.

As shown in figure 1, forecast model number n of the present invention is set to 2, respectively NARX model prediction unit, SVR mould Type predicting unit, its wind-powered electricity generation pre- power scale centre-to-centre spacing exponent number k takes 2.It is bent by the historical power under matlab simulating, verifying , as shown in Fig. 2 the mean square deviation of this combined prediction algorithm is 10.77%, the precision of prediction is higher, excellent for the power curve of line and prediction Gesture is obvious.

Ultimate principle, principal character and the advantages of the present invention of the present invention have been shown and described above.The technology of the industry , it should be appreciated that the present invention is not restricted to the described embodiments, the simply explanation described in above-described embodiment and description is originally for personnel Invention principle, without departing from the spirit and scope of the present invention the present invention also have various changes and modifications, these change Change and improvement both falls within scope of the claimed invention.Claimed scope by appending claims and its Equivalent defines.

Claims (2)

1. a kind of Forecasting Methodology of wind power, its equipment includes

Combined prediction module, specifically includes and inputs wind speed, wind direction, temperature, air pressure, humidity sequence and wind power sequence respectively A) NARX model prediction unit, b) SVR model prediction unit is it is characterised in that described Forecasting Methodology also includes and combined prediction Module sets up the maximum informational entropy prediction module of feedback；Concrete prediction steps are as follows,

1) independent prediction：NARX model prediction unit, SVR model prediction unit reception system detecting information and respectively computing draw Wind power prediction result, combined prediction module calculates the average of the two result

2) determine centre-to-centre spacing：Combined prediction module predicts that t lights the wind-powered electricity generation work(in common n moment Rate valueAnd calculate its numerical characteristic e further_tm, e_tmMeetDetermine the k rank centre-to-centre spacing of the pre- power scale of wind-powered electricity generation

3) combined prediction：Maximum informational entropy prediction module is according to each rank centre-to-centre spacing of model prediction unit number n, pre- power scaleCalculate weight coefficient p in final result for each model prediction unit result_i, meet

$p_i=\exp (-{\mathrm{\λ}}_0-\underset{k=1}{\overset{K}{\Σ}}{\mathrm{\λ}}_k{\left(\frac{x-{\stackrel{\‾}{y}}_i}{{\stackrel{\‾}{y}}_i}\right)}^2)$

$\underset{i=1}{\overset{N}{\Σ}}\exp (-\underset{k=1}{\overset{K}{\Σ}}{\mathrm{\λ}}_k{\left(\frac{x-{\stackrel{\‾}{y}}_i}{{\stackrel{\‾}{y}}_i}\right)}^2)({\left(\frac{x-{\stackrel{\‾}{y}}_i}{{\stackrel{\‾}{y}}_i}\right)}^2-E\[{\left(\frac{x-{\stackrel{\‾}{y}}_i}{{\stackrel{\‾}{y}}_i}\right)}^2\])=0,$

And calculate final wind power prediction value x, meet

$\max \left(H\left(X\right)\right)=-\underset{i=1}{\overset{n}{\Σ}}{p}_i\log p_i.$

2. a kind of wind power according to claim 1 Forecasting Methodology it is characterised in that：Forecast model number n=2, Wind-powered electricity generation pre- power scale centre-to-centre spacing exponent number k=2.