| Abstract: | The efficiency of two deterministic optimization strategies are discussed in order to solve the arising least squares minimization problem. A sequential quadratic programming method is compared with a trust region enhanced simplex method. In particular we focus on the effort to determine the necessary gradients for the sqp‐method and compare the computational costs of the identification process. In the later development the identification problem is extended to account for inhomogeneous deformations using the finite element method for simulation purpose. Here, displacement fields are determined via an optical measurement technique. The accompanying modification of the solution procedure results in an extended least squares functional. The sensitivity analysis, needed for the gradient based method, is embedded in a distinct format in the finite element context. |
