A Hybrid Optimization Approach to Parameter Estimation

Many parameter estimation problems in chemical or biochemical engineering lead to ill-conditioned and nonconvex optimization problems. For bad starting values the use gradient based result in local optimal solutions. To overcome this drawback, a global optimization approach, Simulated Annealing, has been coupled with a gradient-based SQP approach. To improve the accuracy of the parameter estimates, sensitivity information has been included into the objective function by iteratively adjusting the weighting matrix with the variancecovariance matrix of the model prediction. The hybrid approach has been applied to a case study of biochemical nonlinear parameter estimation problem.