Maciej Wasilewski, M.Sc., Eng.

Department of Intelligent Technologies (ZTI)
Division of Control and System Dynamics (PSiDU)
position: doctoral student
telephone: (+48) 22 826 12 81 ext.: 267
room: 437
e-mail: mwasil

Recent publications
1.Wasilewski M., Pisarski D., Bajer C.I., Adaptive stabilization of partially damaged vibrating structures, Machine Dynamics Research, ISSN: 2080-9948, Vol.40, No.1, pp.65-82, 2016
Abstract:

In this paper, an online adaptive continuous-time control algorithm will be studied in the vibration control problem. The examined algorithm is a Reinforcement Learning based scheme able to adapt to the changing system’s dynamics and providing control converging to the optimal control. Firstly, a brief description of the algorithm is provided. Then, the algorithm is studied by the numeric simulation. The controlled model is a simple conjugate oscillator with a sudden change of its rigidity. The effectiveness of the adaptation of the algorithm is compared to the simulation results of controlling the same object by the traditional Linear Quadratic Regulator. Because of the lack of constraints for a system size or its linearity, this algorithm is suitable for optimal stabilization of more complex vibrating structures.

Keywords:

Vibration control, Adaptive control, Optimal control, Policy iterations, Hamilton-Jacobi-Bellman equation

Affiliations:
Wasilewski M.-IPPT PAN
Pisarski D.-IPPT PAN
Bajer C.I.-IPPT PAN

Conference papers
1.Wasilewski M., Pisarski D., Adaptive optimal control algorithm for vibrational systems under nonlinear friction, MMAR 2017, 22nd International Conference on Methods and Models in Automation and Robotics, 2017-08-28/08-31, Międzyzdroje (PL), pp.107-112, 2017
Abstract:

In this paper, a novel control algorithm for vibration attenuation is presented. The proposed scheme is developed to control linear systems with a presence of an external disturbance. The goal of the control is to steer the system to prescribed reference trajectory by minimizing associated quadratic performance index.
The synthesis of the control law consists of two steps. At the first step, past measures of disturbance are used to develop a local linear approximation of dynamics of the disturbance signal. Weights of the associated auto-regressive model are calculated by the least-square algorithm. At the second step, the calculated model is used to obtain a linear time-invariant approximation of the control system. The receding horizon control law is then calculated by using finite horizon Linear Quadratic Regulator. The algorithm is verified numerically for a torsional vibrating system under nonlinear, time-varying friction. The results of the simulation are compared to a standard Linear Quadratic Gaussian control.

Affiliations:
Wasilewski M.-IPPT PAN
Pisarski D.-IPPT PAN

Conference abstracts
1.Wasilewski M., Pisarski D., Konowrocki R., Bajer C.I., New efficient adaptive control of torsional vibrations induced by sudden nonlinear disturbances, CMM-2017, 22nd International Conference on Computer Methods in Mechanics, 2017-09-13/09-16, Lublin (PL), pp.1-2, 2017
Abstract:

The aim of this paper is to present a novel adaptive technique of control of the vibrating drilling systems. The algorithm constitutes an adaptive linear quadratic regulator that uses direct measurements of the disturbance to synthesize its linear dynamic approximation. This approach allows generating control law that includes the impact of the friction on the system dynamics and optimally steers the system to the desired trajectory. The effectiveness of the algorithm is validated via comprehensive numerical simulations of the control of the simplified drilling model. The results are compared to these obtained with the use of the Linear Quadratic Gaussian regulator.

Keywords:

vibration control, drillstring, adaptive control, auto-regressive model

Affiliations:
Wasilewski M.-IPPT PAN
Pisarski D.-IPPT PAN
Konowrocki R.-IPPT PAN
Bajer C.I.-IPPT PAN