Research On Wind Power Ramps Forecasting Method

In an electric power system with high penetration of wind power,the sudden change of wind power over a short time,known as Wind Power Ramp Event(WPRE),has been an urgent problem.It leads to the imbalance of the power system directly,threatens its safe and stable operation,and even causes the serious blackouts,bringing great economic loss to the society.So,it has been necessary to grasp their occurrence characteristics,understand their main factors,so as to predict them timely and accurately.To achieve this target,WPRE was studied from three aspects: WPRE’s occurrence characteristic,WPRE’s factors analysis method and WPRE’s forecasting method.The main contributions are:(1)The periodic characteristics of wind speed was analyzedThe fluctuated wind power is caused by the fluctuated wind speed.Based on the wavelet analysis,the wind speed series was decomposed and analyzed from the periodic aspect.Besides,two indices to evaluate the strength of each periodic component in wind speed were proposed——PI and RPI.In this way,the connection between the fluctuation and the periodicity of wind speed was established.(2)The distribution characteristics and the factors of WPRE were discussedThe distribution characteristics of WPRE were analyzed,and the results showed that the occurrence features and main factors of WPRE are different in different wind farms.Based on that,an original factors analysis method for WPRE with general applicability was proposed,which can be used to determine the dominant factors of WPRE in any wind farm,so as to provide the basis for WPRE forecasting.(3)A WPRE forecasting model based on Orthogonal Test and Support Vector Machine was establishedBased on the WPRE’s factors analysis method and its connection with the various meteorological parameters,a WPRE forecasting model based on Orthogonal Test and Support Vector Machine(OT-SVM)was established.The inputs of SVM can be optimized by OT,and the results showed that the OT-SVM model can improve the forecasting accuracy of WPRE effectively and be applied in any wind farms.(4)A WPRE forecasting model based on Wavelet Transform and Auto Regressive Moving Average was establishedBased on the studies on the time series of measured wind power,a WPRE forecasting model based on Wavelet Transform and Auto Regressive Moving Average(WT-ARMA)was established.To solve the problem that the comprehensiveness and accuracy of WPRE forecasting cannot be satisfied at the same time,a ‘forecast separately and decide comprehensively’ forecasting strategy was put forward.The results showed that the WT-ARMA model can solve this problem effectively,achieving high capture rate and high accuracy of WRPE forecasting at the same time.In addition,there is no need to input the NWP(Numerical Weather Prediction)data to the model,overcoming the effect of NWP’s error on the WRPE forecasting.