Short-term Wind Speed Prediction Based On WD-ARIMA-LSTM

At present,wind power is the most important form of comprehensive utilization of wind energy in my country,and it has attracted more and more scholars’ attention.Accurate wind speed prediction can optimize power distribution,increase power generation,avoid or reduce losses,and also reduce operating costs and increase the benefits of wind farms.However,wind speed is affected by many factors and presents non-stationary and non-linear characteristics.This paper proposes a new combined forecasting model for the randomness and volatility of wind speed scries.The original wind speed time series are usually disturbed by noise,which affects modeling and analysis.Traditional short-term wind speed forecasting methods often predict the original wind speed series and lack the analysis of the error series.Therefore,this article first uses the wavelet threshold denoising method to preprocess the original data to eliminate the adverse effects of noise based on the characteristics of historical wind speed data.,Due to the relevance in the time dimension,according to the dependence between the sequence data,the ARIMA model based on linear recursion is established for preliminary prediction.Taking into account the useful information in the residual sequence,the nonlinear recursive LSTM model is further used to analyze the residual The sequence is analyzed and predicted,and the ARIMA model and the LSTM model are combined to obtain the final wind speed prediction value.Using traditional methods and artificial intelligence,giving full play to the advantages of time series analysis and neural networks,building a combined forecasting model,which is more systematic and comprehensive,can achieve the purpose of improving forecasting accuracy and enhancing the practicability of the model.Use the collected wind speed data to verify the model,select multiple indicators to comprehensively evaluate the performance of the model,and compare it with other models.After verification,the WD-ARIMA-LSTM model proposed in this paper has satisfactory effects and stronger applicability.