Study Of Applying Time Series Prediction Based On Recurrent Neural Network And Ensemble Algorithm

Time Series Prediction is one of the applications of time series analysis. Through analysis of a large number of observational data, people can predict the future development trend of the data, in order to control events which will occur. Traditional time series analysis is analyzed in terms of pure mathematics. Because the time series in practical application has non-linear characteristics such as irregular, chaos and so on, it is very difficult to set up ideal model to the system. And the artificial neural network has self-organization, self-learning and nonlinear approximation ability. Therefore, artificial neural networks used for time series prediction can get a good prediction model.Firstly, the basic concepts, classification and prediction model of time series forecasting methods are introduced in this paper. And the analysis of artificial neural network model and advantages for the time series is focused on, while the wavelet transform used for denoising of time series is being studied. Then, based on the problem of many kinds of artificial neural network structure, the recurrent neural network is mainly studied. And then a recurrent neural network model for time series forecasting is implemented, tested with the sunspot data set, and compared with the BP neural network model. On this basis, AdaBoost ensemble learning algorithm in time series forecasting is studied, and an ensemble prediction model is implemented, using recurrent neural network model as basic prediction model. The ensemble prediction model is tested with the sunspot data set, and different integration of the model is analyzed. Finally, the ensemble prediction model is tested with the tunnel smoke concentration data. The smoke concentration level forecasted is compared with the real levels, in order to verify the performance of the ensemble prediction model. Experiment shows that the model can be used in forecasting the level of the tunnel smoke concentration and provide support for safety control.