Space-Dependent Oblique Projection-Based Iterative Learning Control for the Rejection of Unknown Periodic Disturbances of Continuously Rotary Systems

Abstract

Automation systems are often subject to multiple components of unknown periodic signals, especially disturbances that behave not only time-dependent but essentially position-dependent. Dedicated to approximately identifying and attenuating these disturbances with unknown and arbitrary frequencies, a space-dependent oblique projection-based iterative learning control (SOBP-ILC) approach is proposed for continuously rotary systems. The framework of oblique projection in spatial domain is formulated using Bernstein polynomials as a universal approximator. Position-dependent memory is implemented to facilitate the controller design. The order of Bernstein polynomials and the spatial sampling numbers are discussed in consideration of tracking accuracy and computational complexity. Moreover, position-dependent information extracted from space-dependent oblique basis functions is effectively utilized. The projected estimations are introduced into the SOBP-ILC law at each iteration, making it easier and faster to calculate, improving the rejection capability, and guaranteeing better tracking performance. The proposed approach is computational due to the limited size of the learning matrices. Simulation results and experimental comparisons are conducted to highlight the practical effectiveness and superiority of the proposed approach.