Nonlinear regression methods for simultaneous property measurement

A general procedure for the statistical estimation of thermal properties from experimental data is reviewed. The procedure, based on the techniques of nonlinear regression, includes all traditional methods of property determination and removes four specific restrictions frequently imposed on the design of property measurement experiments. This article surveys nonlinear-regression techniques and applications. A new convergence criterion for halting the iterative calculations is presented. The criterion Tnorm is based on the Euclidean norm of the scaled normal-equation residuals, equaling zero at a minimum of the root-mean-square deviation Trms and being scaled never to exceed Trms.