Disturbance Rejection for Periodic Scanning of Inertially Stabilized Platforms: An Iterative-Learning-Embedded Finite-Time Extended State Observer Approach

During periodic scanning, friction-dominated disturbances near motion reversals cause velocity fluctuations in an inertially stabilized platform (ISP). This paper proposes a disturbance rejection control method based on an iterative-learning-embedded finite-time extended state observer (IL-FTESO). Unlike conventional external iterative learning control (ILC), which learns directly from the tracking error, the proposed method embeds ILC in the disturbance estimation channel. Within each scanning cycle, a finite-time extended state observer (FTESO) accelerates disturbance estimation, and the resulting estimate serves as the learning signal for updating the repetitive disturbance compensation in the next cycle. This structure integrates real-time disturbance estimation within each cycle with cycle-to-cycle iterative compensation. Theoretical analysis establishes convergence of the FTESO observation error and ultimate boundedness of both the learning and closed-loop tracking errors. Experiments on a physical ISP with fiber-optic gyroscope feedback show that the proposed method improves overall angular-rate tracking accuracy and suppresses error peaks near motion reversals. Compared with conventional external ILC, it reduces the root–mean–square error by 58.8% and 61.0%, the maximum absolute error by 45.4% and 48.1%, and the mean absolute error by 53.6% and 68.5% under the 20∘/s, 0.5 Hz and 5∘/s, 2 Hz periodic commands, respectively.