arXiv:2611.14715v1 Announce Type: new Abstract: Internal model principle-based controllers regulate polynomial references and disturbances exactly, but only asymptotically, and their transient input can peak and oscillate, increasingly so as the degree of the exogenous signals grows. A filter placed in series with the plant smooths the command but alters the closed-loop spectrum and can destroy stability. We propose an architecture that keeps the filter outside the loop the regulator stabilizes: the internal model is driven by the error of an ideal, unfiltered copy of the plant, the physical plant receives the filtered command, and an auxiliary plant copy absorbs the mismatch between the two inputs. If the filter satisfies blocking conditions at the origin, any gains that stabilize the unfiltered loop can be combined with any such filter without losing regulation of the true error. Under additive plant uncertainty, we derive a small-gain condition that guarantees robust stability and regulation. On a scalar plant, the proposed controller produces a smoother input and a larger input-delay margin than the unfiltered one, and it maintains regulation under a large loss of actuator effectiveness. On a three-mass chain, its tolerance to unmodeled actuator dynamics and delay stays essentially constant as the regulator is made more aggressive.
Continue to arXiv Mathematics →You are being forwarded in 6s — stay on this page