Estimation of Non-Measurable Signals Using Adjoint Sensitivity Analysis
This paper presents an application of adjoint sensitivity analysis to estimation of unmeasurable continuous-time
signals acting in a non-linear dynamical system based on discrete-time measurements obtained from this system. The problem
is solved using an iterative gradient-based approach where in each iteration the gradient of an objective function in the space of
unknown signals is calculated. The results of the estimation for an exemplary Hammerstein model are presented.
Index Terms- Adjoint Sensitivity Analysis, Estimation, Hammerstein Model, Gradient Optimization, Sampled Systems.