Research On Time Series Prediction Based On Intelligent Algorithm

Forecast is based on historical information to make inferences and expectations about the future. Accurate forecasting is the premise of people to make decision right.The forecast of time series has been the hot topic of academic and engineering fields.Usually, the development change rule of system is implied in the time series data.Through the analysis of time series adequately, the evolution of the data can be got.Then the observation of the system and predict the future behavior can be completed. It is considered important value and significance in the practical application.In practical applications, the time series data that have been obtained usually shows some dynamic characteristics of nonlinearity, non-stationary and rapid change. It is a great challenge to accurately predict this time series. Therefore, in this topic, some intelligent algorithms such as wavelet analysis, Elman neural network, echo state network and fuzzy theory was introduced on the basis of the traditional time series prediction method to improve the effect of time series prediction.In the prediction process, the predictive value is different which is obtained by different prediction methods for the same set of data, sometimes the results make a great difference. Therefore, in view of the limitations of various methods,we can adopt many methods to create a new prediction model for forecasting. The new prediction model will comprehensive utilize all kinds of useful information and improve the prediction accuracy of the data to the maximum extent.This topic proposed the following several kinds of models, the wavelet- AR model,the wavelet- AR- Elman neural network model, the wavelet echo state network model and the fuzzy echo state network model. The original time series are processed with the Mallat algorithm and Daubechies wavelet based on wavelet multi-scale analysis theory that can respectively get the details of the different layers and overview part of sequences. Then forecasting models are respectively created based on the new sequence characteristics of different to forecast and eventually add the forecast data of detail sequences and overview sequence to obtain the original time series prediction result.In this topic, the predictive simulation for the above several kinds of model has benn gaven. At last, through the simulation results and the prediction effect evaluation index, the different predictive ability of the model is analyzed.