On optimization strategies for parameter estimation in models governed by partial differential equations

摘要

Extraction problems governed by systems of partial differential equations appear in several branches of Engineering. Parameter estimation involves discretization and modeling in a finite dimensional setting. A model that arises in the supercritical extraction area is analyzed in this paper. Numerical difficulties of the nonlinear programming formulation of the estimation process are discussed and a satisfactory procedure based on unconstrained derivative-free optimization is suggested.