Recent Work and Publications
Reinforcement learning algorithms still struggle to handle nonlinear set point control loop design. Three new techniques inspired by classical feedback control that significantly improve on the present state of the art are presented in
- R. Zhang, P. Mattsson and T. Wigren, "Robust nonlinear set-point control with reinforcement learning", to appear in Proc. ACC, San Diego, USA, May 31- June 2, 2023.
An new algorithm for recursive identification of a nonlinear continuous time state space ODE model with unknown nonlinear measurement equation is described in the open access publication:
- T. Wigren, "Recursive identification of a nonlinear state space model", to appear in Int. J. Adaptive Contr. Signal Processing, 2023.
The software package that implements the algorithm is available for free downlad at http://www.it.uu.se/katalog/tw/software , including a user manual in the report:
- T. Wigren, "MATLAB software for recursive identification and scaling using a structured nonlinear black-box model – revision 7", Technical Reports from the Department of Information Technology, 2021-008, Uppsala University, Uppsala, Sweden, December, 2021.
The previous software package that handles identification of a delay in combination with a continuous time ODE model, has also been augmented with an unknown output equation. The new revision of this SW package is described by the report:
- T. Wigren, "MATLAB software for nonlinear and delayed recursive identification - revision 2", Technical Reports from the Department of Information Technology, 2022-002, Uppsala University, Uppsala, Sweden, January, 2022.
Free download is available from http://www.it.uu.se/katalog/tw/software.
My Ph.D thesis from 1990 has been digitized by the university library:
- T. Wigren, "Recursive identification based on the nonlinear Wiener model", Ph.D. thesis, Acta Universitatis Upsaliensis, Uppsala Dissertations from the Faculty of Science 31, Uppsala University, Uppsala, Sweden, December, 1990. Available: http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-118290