Cascading metamodels from different sources for performance analysis of a power module

During the development process of a semiconductorbased product several types of results are generated, often in large volumes, e.g. simulation or test measurements. These have to be processed and can then be used as a reusable knowledge base for further experiments/developments. Hence, to manage such knowledge from various data sources, it is not sufficient to use classical data analysis methods. A compressed representation of this information, showing only what is important with respect to the systems performance, is desirable. We develop a method to support the combination of information from different sources and to represent it. The concept is based on cascading metamodels: The outputs of metamodels become inputs to subsequent metamodels, and mathematical composition operators can be generated for this concatenating procedure. This method is applied to a power module in order to perform sensitivity analysis on the combined metamodel.

[1]  Peter Buchholz,et al.  Optimization of Process Chain Models with Response Surface Methodology and the ProC/B Toolset , 2005 .

[2]  Thomas J. Santner,et al.  Design and analysis of computer experiments , 1998 .

[3]  Timothy W. Simpson,et al.  Sampling Strategies for Computer Experiments: Design and Analysis , 2001 .

[4]  Angela M. Dean,et al.  Design and analysis of experiment , 2013 .

[5]  Joaquim P. Marques de Sá,et al.  Applied statistics : using SPSS, STATISTICA, and MATLAB , 2003 .

[6]  Paul G. Maropoulos,et al.  Design verification and validation in product lifecycle , 2010 .

[7]  Jack P. C. Kleijnen,et al.  A methodology for fitting and validating metamodels in simulation , 2000, Eur. J. Oper. Res..

[8]  Lubos Buzna,et al.  Modelling of cascading effects and efficient response to disaster spreading in complex networks , 2008, Int. J. Crit. Infrastructures.

[9]  D. Ruppert The Elements of Statistical Learning: Data Mining, Inference, and Prediction , 2004 .

[10]  Matteo Matteucci,et al.  Robust Estimation of Natural Gradient in Optimization by Regularized Linear Regression , 2013, GSI.

[11]  Georg Pelz,et al.  Configurable load emulation using FPGA and power amplifiers for automotive power ICs , 2012, Proceeding of the 2012 Forum on Specification and Design Languages.

[12]  Douglas C. Montgomery,et al.  Response Surface Methodology: Process and Product Optimization Using Designed Experiments , 1995 .

[13]  Masashi Sugiyama,et al.  Active Learning with Model Selection in Linear Regression , 2008, SDM.

[14]  V. Barnett,et al.  Applied Linear Statistical Models , 1975 .

[15]  Yuhong Yang,et al.  Combining Linear Regression Models , 2005, Journal of the American Statistical Association.

[16]  Christopher P. Saunders,et al.  Computationally and statistically efficient model fitting techniques , 2017 .

[17]  Masoud Rais-Rohani,et al.  A comparative study of metamodeling methods for multiobjective crashworthiness optimization , 2005 .

[18]  Gertrude M. Cox,et al.  Experimental Design , 2019, Simulation and Computational Red Teaming for Problem Solving.

[19]  G. Winskel What Is Discrete Mathematics , 2007 .

[20]  Conrad A. Fung,et al.  An explanation and critique of taguchi's contributions to quality engineering , 1988 .

[21]  Timothy W. Simpson,et al.  Sampling Strategies for Computer Experiments , 2001 .

[22]  Derek J. Pike,et al.  Empirical Model‐building and Response Surfaces. , 1988 .

[23]  Achille Messac,et al.  Metamodeling using extended radial basis functions: a comparative approach , 2006, Engineering with Computers.