Direct input synthesis control to output tracking of linear systems: A data-driven approach

In this paper, a data-driven direct input synthesis control (DD-DISC) approach is proposed for output tracking of multi-input-multi-output linear time invariant systems. Recently, data-driven approach has attracted more and more attention to address the limitations and challenges in model-based control. Unlike many existing data-driven control (DDC) techniques that are often focused on the design and analysis of a general-purpose feedback/feedforward controller, we propose to directly synthesize the control input to track a finite-previewed desired trajectory using past input–output data, thereby, avoiding the complexity and constraints in designing a general-purpose controller, particularly for non-minimum phase (NMP) systems. Specifically, the proposed approach alleviates the limitations of existing DDC methods in requiring the system state to be measurable and/or the past data to be noise-free, and compensates for the NMP zero effect on the tracking precision through pre-actuation. The robustness and performance of the proposed technique under the output noise/disturbance effect are analyzed, and the approach is extended to completely reject periodic output disturbance—even when the disturbance is different from that in the past data. The proposed method is illustrated through the output tracking on a multi-input multi-output system in simulation.