Research On Short Term Wind Speed Prediction Of Wind Farms Based On Chaotic Time Series

In recent years, as the global economy continues developing, and the problem of environment pollution caused by the burning of fossil fuels is getting more and more serious, all countries in the world pay more attention to the research, development and utilization of new energy. Wind energy reserves are abundant, clean, and widely distributed, with good growth potential. At present, wind power field grid operation is an effective method in actual application of wind energy. Wind energy has the characteristics of randomness and volatility, which can cause the wind power random and unstable, and thus a large number of wind turbines access to the power system will affect the security and stability of the whole system. Accurate wind speed prediction helps to alleviate the adverse effects, and this play an important role in the supervision and management of wind power field grid operation. In this paper, with the measured wind speed time series as the research object, based on the analysis of its chaotic characteristics, the maximum entropy principle, support vector machine theory and chaos theory are combined to propose a short term wind speed prediction model based on support vector machine using maximum entropy of chaotic time series. The main content of the work is as follows:(1)The computation results of embedding dimension and delay time has a significant impact on improving the quality of phase space reconstruction and reducing the prediction error of time series; at the same time, the two parameters are the bases of the further calculation of other parameters. In this paper, using a variety of methods to discuss the calculation of embedding dimension and delay time respectively, and they are compared by examples.(2)Only prove that the wind speed time series is chaotic, can the theory and analysis methods associated with the chaos be used in the construction of wind speed short-term forecasting model. In this paper, using the method of calculating the saturation correlation dimension and the largest Lyapunov exponent to determine the chaotic characteristics of historical wind speed data.(3)For the comparison with the prediction model proposed in this paper, using the weighted one rank local region method multi step prediction model for multi step prediction of the actual measured data, the results show that the prediction accuracy of the model is relatively low, the prediction error is relatively large.(4)Based on the least squares support vector machine under the Bayesian framework and maximum entropy principle, the short term wind speed prediction model based on support vector machine using maximum entropy of chaotic time series is established, and using this model to predict the actual measured wind speed data. The prediction results show that the prediction model proposed in this paper has obviously improved in terms of forecast effect and forecast accuracy.