In recent years, the time series method has been got more and more attention in financial domain. In time series analytic method, the GARCH model could analyze effectively the information of volatility clustering. When the actual model set, the economical or financial resources are always having strongly unstable information that can not be confirmed. In addition, if the sample material period too long or the sample material changing too much. These will influence the accuracy of the forecasting. Therefore, in this article, we will attempt to base on the GARCH model with the advantage of change periods and the fuzzy theory, in order to build the GARCH model that can more close to the real situation. Then, we used the simulation data and real resources to test and verify the accuracy of model forecasting.