Forecasting Model And Applied Research Of Based On Time Series Arch

As one of the branches of statistic, time series is the objective records of the history behavior of the system studied. Time series analysis provides a suit of methods dealing with dynamic data, the most important measure of this method is to approximately depict many kinds of data by relevant mathematical models. Through the analysis of the corresponding model, we can further know the structure character and the change law of data, so we can achieve the purpose of forecasting its development trend and putting up essential control.This paper studies the optimization estimation method of additive GARCH model and spectral analysis test method of the time series model.First of all, we mainly show the purposes of time series analysis, the development history and current state of time series analysis, and analyze the development foreground of time series analysis and expound the development history of the conjugate gradient method.Secondly, we introduce the relevant models of additive GARCH model and model selection criteria and the optimization estimation method of model, and give the test method for time series models and the spectral theoryThirdly, Based on the non-parametric GARCH model, additive GARCH model is proposed。We make up a new estimation algorithm and using new algorithm method to estimate the additive GARCH model. In this chapter, we compared the additive GARCH model with other time series models. It shows that the additive GARCH model is significant. And we compared the new estimation algorithm with the backward-fitting algorithm. The results show that the new estimation algorithm is better than the backward-fitting algorithm in the estimation of additive GARCH model.Fourthly, we make up another new kind of conjugate gradient method and the application of it in the additive GARCH model. Compared the backward-fitting algorithm, the effect of the new algorithm is good.Lastly, Based on the window spectral estimation and spectrum estimation, the new spectrum analysis test method is proposed. And by example show that the spectrum analysis test method is more effective model validation than the traditional residual analysis method usually used by time series models.