Abstract

A definition for the measure of observability is introduced. It is illustrated by two examples of nonlinear systems defined by ordinary and partial differential equations. Using computational dynamic optimization, the concept of observability is numerically implementable for a wide spectrum of problems. Some topics addressed in this paper include the observability based on user-knowledge, the determination of strong observability vs. weak observability, the optimal sensor locations, and the partial observability of complex systems.

Keywords: Observability, optimal sensor placement, dynamic optimization, numerical computation.