A parameterization method based on multi-method collaboration

Due to the inherent advantages, limitations, as well as the complementarity of different parameterization methods, a collaborative mechanism of multi-parameterization method is proposed in this paper, that is, a collaborative parameter analysis model is developed on the basis of multi-parameterization method analysis results (such as regression analysis method, improved neural network method, granularity analysis method, etc.) to improve the accuracy and reliability of analysis result. At the end of this paper, accuracy and effectiveness of the proposed method are verified through a practical analysis of failure prediction for the heating heads of semiconductor test equipments.