Partial least square regression on frequency shift applied to passive acoustic emission spectra

摘要: A novel approach for the regression on passive acoustic emission spectra is presented. Acoustic of fluid flow through an orifice plate partly characterized by frequency shifts in spectra. Physical properties affect and can therefore be predicted exploiting changes peak positions. The positions corresponding amplitude most distinct peaks were determined. obtained matrices then used partial least squares (PLS) regression. proposed method was validated two experimental datasets from measurements liquid plate. liquids a two-component mixture ethanol water, three-component sucrose water. resulted more parsimonious models lower root mean squared errors prediction (RMSEP), compared to PLS whole 43% RMSEP 24 27% water respectively mixture. Only minor improvement gained prediction. Copyright © 2007 John Wiley & Sons, Ltd.