Nonlinear Analysis And Short-term Prediction Of Wind Speed Time Series

As a clean and renewable energy,wind energy has become one of the most promising new energy generation methods.However,the characteristics of randomness,volatility and uncontrollability of wind speed make the wind speed prediction very difficult and the accuracy is low,which leads to the complex nonlinear characteristics of wind power generation such as intermittentity and instability.Therefore,in order to reveal the complex internal characteristics of the wind speed time series and realize the accurate prediction of wind speed,the wind speed time series of the M2 wind tower in the national wind energy research center of the United States is taken as the research object,and the nonlinear analysis theory was used to analyze and predict the wind speed.Firstly,starting from the nonlinear characteristics of wind speed time series,the fractal dimension and long-range correlation of wind speed time series is analyzed based on fractal theory.Secondly,on the basis of the chaos theory,the delay time and the embedding dimension are determined by C-C method,and the phase space reconstruction is carried out for the wind speed time series.On this basis,a variety of chaotic identification methods are used to identify the chaotic characteristics of wind speed time series from two different angles.Finally,on the basis of determining the chaotic characteristics of the wind speed time series,the RBF neural network and the Volterra adaptive filtering prediction model are introduced into the wind speed prediction research,and the wind speed is accurately predicted.Through the above work,the main results are as follows:1.According to the fractal theory,the basic principles of the box counting method and the correlation dimension method are studied,and the fractal dimension of the wind speed time series is calculated.At the same time,the scale-free characteristics and self similarity of the wind speed time series are analyzed.2.Taking the Fractal Gaussian Noise(FGN)as an example,the characteristics of the R/S analysis methods are compared and analyzed.The Lo method is used to calculate the Hurst exponent of the time series of wind speed,and the Hurst exponent of the time series of wind speed is calculated by this method.In addition,in order to further study the nonlinear analysis of wind speed time series,the nonlinear analysis of wind speed time series is carried out by Detrended Fluctuation Analysis(DFA)and Power Spectrum Analysis(PSA).The results shown that the Hurst exponent of the time series of wind speed are more than 0.5,indicating that the wind speed is not the time series of complete random fluctuation,and it has a clear long range positive correlation.In addition,the wind speed time series Hurst exponents are close to 1,indicating that the wind speed time series presents a ‘1/f fluctuation' feature,which is a typical non-stationary signal.3.The basic principles of phase diagram method,spectrum analysis method,maximum Lyapunov exponent method and 0-1 chaos decision method are studied.The above methods are used to determine the chaotic characteristics of wind speed time series from different angles.The results show that the phase diagram method and spectrum analysis method can distinguish the chaotic characteristics from the time series in the space or time frequency domain,and the maximum Lyapunov exponents are the important parameters of the chaotic system,and can be used to analyze the chaotic characteristics of the time series.The 0-1 chaotic test method can determine the chaotic characteristics of the wind speed time series only by whether the gradual growth rate Kc is close to 0 or 1.Compared with other chaotic identification methods,it has the advantages of low calculation cost and fast speed.4.On the basis of determining the chaotic characteristics of the wind speed time series,the RBF neural network combined with phase space reconstruction theory are used to predict the time series of wind speed and to analyze the prediction error.The prediction results show that the prediction method based on the RBF neural network has strong classification ability and approximation energy for the chaotic time series.Power and sample learning ability,and the prediction results are more accurate.5.On the basis of studying the Volterra adaptive prediction algorithm,the Volterra adaptive chaotic prediction model of wind speed time series is established,and the wind speed time is predicted and the error analysis is analyzed.The prediction results are more accurate.