Structure-Preserving Data-Driven Identification of Port-Hamiltonian Differential-Algebraic Systems (Q7345260)

We propose a data-driven identification method for linear index-1 differential-algebraic port-Hamiltonian systems that preserves their structure and minimizes the difference between simulated and measured input-output data. This is achieved by formulating an adjoint-based optimization problem using implicit midpoint discretization of the pH-DAE dynamics, which enables efficient gradient computation and guarantees convergence to a minimum.