The Study Of Forecasting Models On Electric Characteristics About Crystallization Process Of Cu-Zr-Al Amorphous Alloys

Material formation process is a great complex dynamics.It is time-consuming and difficult to accurate model by mechanism.By combining materials science and information technology,the evolution process of characteristics in material is considered as a non-stationary process in this research.Targeted to crystallization process of Cu-Zr-Al amorphous alloys,the model of material formation process and property forecasting are studied mainly in this thesis.In order to make LSSVM more suitable for online applications,The concept of fuzzy membership is introduced to incremental LSSVM(ILSSVM)in the research so as to reduce the sensitivity of regression function for isolated points and improve the accuracy of the forecast.The results of simulation show the algorithm is more accurate, faster with its own recursive feature.To a non-stationary time series prediction,the key is how to extract its low and high frequency components,and to avoid the over-fitted for high frequency signals. Considering the adaptability of the wavelet decomposition,the wavelet transform is applied to time series forecasting.The algorithm using the wavelet analysis and BP-ILSSVM is proposed in this thesis.By using wavelet decomposition and reconstruction,the non-stationary time series with tendency are decomposed into a low frequency component and several high frequency components.The high frequency signals are predicted using BP models,and the low frequency is predicted using ILSSVM.The results of simulations show this new prediction method avoids the over-fitted for high frequency signals,and effectively fits the low frequency signal of the non-stationary time series.