Tracking Control of Non-minimum Phase Systems: A Kernel-based Approach
| dc.contributor.author | Mehrabi, Mohammadmahdi | |
| dc.contributor.author | Ahmadi, Keivan | |
| dc.date.accessioned | 2023-10-05T17:39:55Z | |
| dc.date.available | 2023-10-05T17:39:55Z | |
| dc.date.copyright | 2023 | en_US |
| dc.date.issued | 2023-10-05 | |
| dc.description.abstract | Feedforward control with model inversion can achieve high-accuracy output tracking, but it generates unbounded control input for non-minimum phase (NMP) models due to unstable poles. Pseudo-inversion methods use parametric regression to bound the control input based on the desired trajectory and the system dynamics. We propose a new non-parametric pseudo-inversion method that uses Bayesian inference and kernel functions to design optimal and flexible control trajectories. Compared to existing methods, the presented approach can handle arbitrary types of NMP systems and trajectories and embed desirable features in the control input. We also develop a recursive limited-preview version that is computationally efficient and suitable for online and adaptive applications. We present closed-form equations for both versions and compare their performances with existing methods in benchmark examples. | en_US |
| dc.description.reviewstatus | Unreviewed | en_US |
| dc.description.scholarlevel | Faculty | en_US |
| dc.description.sponsorship | This work was supported by the National Research Council Canada under Grant number DHGA-108-1. | en_US |
| dc.identifier.uri | http://hdl.handle.net/1828/15478 | |
| dc.language.iso | en | en_US |
| dc.subject | Feedforward Control | |
| dc.subject | Nonminimum Phase System | |
| dc.subject | Bayesian Estimation | |
| dc.subject | Model Inversion | |
| dc.subject | Kernel Estimation | |
| dc.subject.department | Department of Mechanical Engineering | |
| dc.title | Tracking Control of Non-minimum Phase Systems: A Kernel-based Approach | en_US |
| dc.type | Preprint | en_US |