Efficient Tracking of Behavior in Complex Hybrid Systems via Hybrid Bond Graphs

For many real-world systems, which exhibit complex, nonlinear, and hybrid behavior, it is important to accurately track and monitor the state and health of these systems. The continuous state estimation problem has been well studied, and a number of extensions of the Kalman filter to nonlinear systems have been proposed. Hybrid state estimation poses an additional challenge, because the model must be quickly updated during a mode change to facilitate accurate, real time tracking. This paper discusses an approach to minimize the amount of equation regeneration necessary when the system undergoes hybrid mode changes. These equations are used as the state update equation for an Unscented Kalman Filter that tracks the system’s state. We demonstrate the effectiveness of our approach by tracking the hybrid behaviors of NASA’s ADAPT test bench. Results show that our algorithm scales well for tracking large nonlinear and hybrid systems.