论文信息 - AN ALGORITHM FOR AUTOMATIC SELECTION AND ESTIMATION OF MODEL PARAMETERS

AN ALGORITHM FOR AUTOMATIC SELECTION AND ESTIMATION OF MODEL PARAMETERS

Abstract An algorithm for automatic selection and estimation of model parameters is presented. The algorithm uses a sensitivity matrix based calculation of the parameters effects on the measured outputs and of a linear-independence metric. A predictability degradation index and a parameter correlation degradation index are used as stop criteria and the method is extended to dynamic models and multiple operating points. The applicability of the developed algorithm is illustrated through a hypothetical nonlinear input-output model and through the analysis of data from an experimental isothermal batch bioreactor. The obtained results show the effectiveness of the algorithm.

[1] D. Himmelblau,et al. Optimization of Chemical Processes , 1987 .

[2] Per Olof Gutman,et al. Dominant parameter selection in the marginally identifiable case , 2004, Math. Comput. Simul..

[3] Michael J. Kurtz,et al. Selection of model parameters for off-line parameter estimation , 2004, IEEE Transactions on Control Systems Technology.

[4] A. Secchi,et al. A growth kinetic model of Kluyveromyces marxianus cultures on cheese whey as substrate , 2004, Journal of Industrial Microbiology and Biotechnology.

[5] Richard J. Finno,et al. Selecting parameters to optimize in model calibration by inverse analysis , 2004 .

[6] M. A. Akanbi,et al. Numerical solution of initial value problems in differential - algebraic equations , 2005 .

[7] Stefan Weijers,et al. A procedure for selecting best identifiable parameters in calibrating activated sludge model no. 1 to full-scale plant data , 1997 .

[8] Peter Reichert,et al. Practical identifiability of ASM2d parameters--systematic selection and tuning of parameter subsets. , 2002, Water research.