Research On Multi-step Forecasting And Predictability Of Ultra-short Term Wind Power In Wind Farm

With the increasing consumption of global fossil fuels and the growing demand for electricity,the development of renewable energy for power production has become the focus of attention around the world.Compared with other energy sources,wind energy has a significant advantage in the context of large-scale commercial development.In recent years the most rapid development of new energy power generation technology.However,due to the influence of the natural wind,the wind power is random,intermittent and uncertain.The large-scale wind power grid integration has a negative effect on the safe and stable operation of the power system.Therefore,it is of great practical significance to forecast the wind power.Based on the wind power data collected by the wind farm in real-time,this paper studies the wind power fluctuation characteristics,the short-term wind power forecasting method,the wind power forecasting error and the wind power predictability.Based on the time-space distribution of the output power fluctuation of large-scale wind farm,a method of describing the probability distribution of wind power fluctuation characteristics based on mixed distribution model is proposed.The wind power fluctuation of the same unit number and the wind power fluctuation of the same sampling interval are described.On the basis of the analysis,the cumulative probability of wind power fluctuation rate is analyzed,which is of great significance to wind power forecasting.Based on the correlation analysis and K-nearest neighbor algorithm,a new multi-output model of ultra-short term wind power forecasting method is proposed to realize the ultra-short term wind power forecasting of high-precision,and the influence of wind turbine assembly on the forecasting results is analyzed.At the same time,in order to enhance the information complementarity of the forecasting method,a single forecasting method is proposed to establish the combined forecasting model based on the adaptive neuro-fuzzy inference system to optimize the single wind power forecasting results.Taking the measured data of the wind farm as an example,the simulation analysis is carried out to verify the effectiveness of the two wind power forecasting models,and the influence of the convergence effect on the forecasting results is analyzed,which is favorable to the forecasting accuracy of wind power.Based on the mixed distribution model to describe the distribution characteristics of wind power forecasting error,the evaluation indexes of various distribution models are compared respectively.Due to the existence of significant time-dependent structure of the wind power data,the wind power level is divided by the difference of the forecasted power level.The probability density of the forecasting error distribution is calculated by dividing the forecasting error in thesegment into statistical samples,and then the cumulative probability of the distribution model is solved.The probability analysis of wind power forecasting error can provide the basis for the improvement of wind power forecasting accuracy and uncertainty analysis.In view of the objective fact that the wind power prediction method can not achieve the prediction of no error at all,the concept of predictability of wind power time series is proposed.The approximate entropy and the predictable coefficients are used to quantitatively analyze the predictability,and the variation law of the approximate entropy of the number of different units is analyzed.The wind power predictability research can objectively evaluate the advantages and disadvantages of the forecasting method on the same platform,and can also provide the basis for determining the practical accuracy of the forecasting accuracy of different wind farms.