Real-Time Estimation of Endogenous and Exogenous Inputs

Nearly all dynamic systems have input excitations that are either unmeasurable or unknown due to practical constraints such as feasibility or cost. Estimation of these excitations can be useful both in control applications as well as system modeling applications. The objective of this work is to expand upon an observer based approach to estimate unmeasurable or unknown inputs to a dynamic system using linear systems theory in an efficient manner that is suitable for real-time implementation. In this work, we explicitly explore two fundamental questions. How should the structure, dimensionality, and parameterization of an internal model or waveform generator model be selected for a given dynamic system? How do we determine, based on the structure of the dynamic system, whether estimation of exogenous and endogenous inputs is possible? A series of numerical simulations is performed, providing insight into these issues.