IDENTIFICATION OF NONLINEAR MODEL PARAMETERS FOR RAPID THERMAL PROCESSES
Read the full article
A problem of parameters identification is considered for a nonlinear model of rapid thermal processes. A hybrid
approach is proposed for parameter estimation combining both analytical solution and numerical optimization. At the first step, the rate of temperature change is estimated from experimental data, which makes it possible to rewrite the considered nonlinear model as a linear regression and estimate the parameters by the least-squares method. Further, this estimation is used as an initial guess for numerical optimization of prediction error minimization problem, thus the optimal predictor is obtained. The proposed approach was verified at an experimental setup for vapor deposition processing; the resulting estimates provide high-quality temperature prediction.