Wind speed prediction based on wavelet analysis and time series method

In order to improve the accuracy of wind speed prediction, the paper proposed a brand new prediction model, called the nonlinear least squares autoregressive moving average model, which based on the wavelet decomposition and reconstruction. Firstly, this model introduces the wavelet transform, which decomposes the original non-stationary wind speed sequence into a relatively stable sequence; Then it gets established the nonlinear least squares autoregressive moving average model and makes single step prediction for each sequence decomposed by the wavelet, during the prediction update the value of history sequence constantly; Finally, for the sake of the outcome of the original wind speed series, it sums up the prediction result of each layer decomposed by the wavelet. The results of actual calculation show that the time series forecasting method based on wavelet decomposition is better than the traditional method, and has higher prediction precision in wind forecasting.