Robust extended Kalman filtering for nonlinear systems in the presence of unknown inputs and correlated noises

This study proposes a robust extended Kalman filter (REKF) for discrete-time nonlinear systems with parametric uncertainties, unknown inputs, and correlated process and measurement noises. An augmented model is proposed to estimate the unknown inputs and system states simultaneously. The designed filter guarantees an upper bound on the error covariance of the estimation. It is robust against process and measurement noises, model uncertainties, and unknown inputs. Besides, the robust performance of the designed filter is evaluated. Finally, a realistic gas pipeline is simulated by OLGA multiphase flow simulation software. REKF and extended Kalman filter are compared to detect the pipeline's leakage and location. The results show the effectiveness of the proposed REKF.

Automated summary

This study presents a robust extended Kalman filter for discrete-time nonlinear systems, capable of estimating unknown inputs and system states simultaneously while ensuring an upper bound on estimation error covariance. The filter demonstrates robustness against noise, uncertainties, and unknown inputs. The effectiveness of the REKF is validated through simulated gas pipeline leakage detection.