作者: Carlos Alberto Claumann , André Wüst Zibetti , Ariovaldo Bolzan , Ricardo A.F. Machado , Leonel Teixeira Pinto
DOI: 10.1016/J.CHROMA.2015.10.073
关键词:
摘要: For this work, an analysis of parameter estimation for the retention factor in GC model was performed, considering two different criteria: sum square error, and maximum error absolute value; relevant statistics are described each case. The main contribution work is implementation initialization scheme (specialized) estimated parameters, which features fast convergence (low computational time) based on knowledge surface criterion. In application to a series alkanes, specialized resulted significant reduction number evaluations objective function (reducing estimation. obtained happened between one orders magnitude, compared with simple random initialization.