Wind Power Prediction Based On The Random Fuzzy Theory

To change the energy structure, development and utilization of new energy is the most effective way to solve two big problems, the world energy crisis and environmental pollution. Wind power, widely distributed, is a kind of clean and cheap renewable green energy with considerable total resource amount. In recent years, with the rapid growth of wind power installed capacity, national policy support for wind power, developing of wind power generation technology, reduction of wind power cost and enhancement of its competitiveness in the electricity market, wind power has become the world’s hottest renewable energy power generation. Due to the bad effects on the economic dispatch, safe and stable operation of power grid caused by wind randomness and volatility, research on wind power prediction is increasingly going deeper.Wind has randomness, volatility, intermittent and other significant characteristics. Wind speed is the biggest influence factor of wind power. Most of researches on wind power prediction in the past are deterministic. At present, there are few researches in the field of wind power prediction on uncertainty. In this paper, from NWP forecasted wind speed to actual wind speed to output, wind power prediction is described as a random fuzzy process including double uncertainties. And a wind power prediction method based on the random fuzzy theory is also presented. By the method, point value and prediction interval of wind power under different probability and credibility levels are obtained.Wind power prediction, based on historical operating data of a wind farm, NWP forecasted wind speed and other data, uses physical model or statistical model through wind power curve to predict wind power and meet dispatching requirements. So the uncertainty of wind power prediction comes from two aspects, NWP and wind power curve. In this paper, the relationship between NWP forecasted wind speed and actual wind speed is studied. Three kinds of wind power curve modeling methods are compared, and the measured wind power curve,which reflects the actual operation of wind turbine, is drawn by method of bins based on recorded field data. Wind speed is classified through partition fitting because the global wind power distribution does not satisfy a particular distribution. A non-parametric confidence interval estimation method is applied to establish a probability density function model for wind power in each wind speed level, and on the basis of point estimation, the uncertainty estimation interval of wind power curve is obtained.Wind power prediction is described as a random fuzzy process. Cauchy distribution is used as the membership function of forecasted wind speed error and non-parametric estimation is adopted to calculate the probability density function of wind power. By the approach of random fuzzy simulation, finally, the point value and fluctuation interval of wind power prediction are obtained. Numerical results verify the efficiency and feasibility of the proposed method which enriches the information of wind power prediction and has a good reference value.