Novel nonlinear multivariate calibration methods

Abstract: The interest fornonlinear modeling has increasedconcomitantly with the increased awareness of the ubiquity ofnonlinear systems. The traditional solution in analyticalchemistry, to linearize the system by some sort of samplepretreatment, is not always an option in process analyticalchemistry. Sample pretreatment merely introduces unwantedsources of error. The analysis should be performed in-line, orat least on-line, and facilitate a rapid prediction of keyprocess parameters.Keywords: multivariate calibration, nonlinear modeling,PLS, neural networks, linearing transform