Wind speed prediction model along railway based on a data decomposition and deep belief networks

Abstract

Wind is one of the most important natural factors that should be paid attention to when train running safely. The prediction of wind speed along the railway is of great significance to the safe running, dispatching and riding comfort of trains. A new wind speed prediction model along railway is proposed based on VMD-BGA-DBN. Firstly, variational modal decomposition is used to preprocess the original time series and obtain the decomposed sub-series. Then, binary-coded genetic algorithm selects the features of the sub-series for the deep belief network predictor. Finally, the optimized sub-series after feature selection processing are put into the deep belief network models to obtain the prediction results of each wind speed sub-series, and the prediction results of all sub-series are combined to generate the final wind speed prediction results. According to the prediction results, it can be seen that (a) among all decomposition strategies, the variational modal decomposition method adopted in this paper can better deal with the non-stationarity and randomness of wind speed timing data; (b) the proposed hybrid model has significant research potential.