BOUNDARY CONTROL OF BURGERS-EQUATION - A NUMERICAL APPROACH

We describe a methodology for solving boundary control problems for th e viscous Burgers' equation. The aim is to identify boundary forcing i n order to ensure the ''best'' fit between data and model results, by minimizing a functional which measures model and data discrepancies. A continuous variational formulation involving the adjoint technique is used, and its counterpart discretized version is obtained with a matr ix approach as a guideline. A particular discretization of the nonline ar term of the equation is performed in order to insure, for the gradi ent of the functional to be minimized, a discretized expression which can be directly deduced from its continuous counterpart. Numerical exp eriments validate the proposed optimization algorithm.