Application of decomposition coordination methods to parameter identification problems in interconnected distributed parameter systems (Q1078130)

The paper studies the application of decomposition/coordination methods to parameter identification problems for a class of interconnected distributed parameter systems described by a set of coupled parabolic partial differential equations. The parameter identification problem is formulated as an optimization problem, and the solution is sought using two decomposition methods, namely, the penalty function method and the re-injection method. Local convergence results for the decomposed identification procedure are obtained from a linearized model derived by a sensitivity analysis. The results are illustrated by an example, which demonstrates some interesting features of the decomposed identification procedure.