Oskina & Bobtsov's new observer method handles unmatched disturbances in linear descriptor systems
A novel parameterization technique enables linear regression for nonlinear disturbance parameters, solving a long-standing control challenge.
The paper addresses a critical problem in control theory: designing observers for linear descriptor systems affected by unknown unmatched disturbances. Traditional methods often assume disturbances are matched or measurable, limiting their applicability. Oskina and Bobtsov introduce structural assumptions that allow construction of an observer for the dynamic part of the state vector. Once the state is estimated, the disturbance signal is reconstructed and its unknown parameters identified—even when the disturbance depends nonlinearly on those parameters.
The key innovation is a new parameterization method for disturbance input signals that depend nonlinearly on unknown parameters. This clever transformation yields a linear regression in the unknowns, making the parameter identification tractable. The approach is validated through numerical simulations that demonstrate accurate state and parameter estimation. This work is particularly relevant for systems like power grids, autonomous vehicles, and industrial robotics, where unmeasured disturbances can degrade performance or cause instability.
- Observer design for linear descriptor systems (DAEs) affected by unknown unmatched disturbances.
- Novel parameterization converts nonlinear disturbance parameters into a linear regression problem.
- Numerical simulations confirm accurate estimation of both state vector and disturbance parameters.
Why It Matters
Opens the door to robust control in safety-critical systems like autonomous vehicles and power grids, improving reliability under real-world uncertainty.