Parameter identification in dynamic systems using the homotopy optimization approach (Q649540)

The authors consider the nonlinear parameter identification problem for systems governed by ordinary differential equations. It is assumed that only partial state measurement corrupted by noise is available from experiments. The identification problem is posed as the minimization problem for the integral of the squared difference between the experimental and simulated states. To solve the minimization problem, it is proposed to apply the homotopy optimization technique. This technique is based on using the objective function modified by adding another function whose optimum is known, and a morphing parameter is used here to transform the modified function into the original objective function. To demonstrate the effectiveness of this methodology for obtaining global minima, several examples are presented including a multibody vehicle model.