Control of rotary double inverted pendulum using neural network based adaptive linear quadratic regulator | |
| Author | Viroch Sukontanakarn |
| Call Number | AIT Diss. no.ISE-12-01 |
| Subject(s) | Pendulum Rotational motion (Rigid dynamics) |
| Note | A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Engineering in Mechatronics, School of Engineering and Technology |
| Publisher | Asian Institute of Technology |
| Series Statement | Dissertation ; no. ISE-12-01 |
| Abstract | This dissertation introduces the development and control of a rotary double inverted pendulum (RDIP) system. Rotary double inverted pendulum system is a nonlinear and unstable system. The control of a nonlinear system is a challenging topic for control engineers. This dissertation presents an optimization method for balancing control of a rotary double inverted pendulum system. The design of the controller of this system is difficult and challenging because it has under-actuated control input. The input, torque of a motor, is used to control three links of the system; the arm position is maintained in horizontal plane, and the two pendulums are maintained at the upright position. The main controller used in this dissertation is the linear quadratic regulator (LQR). However, the RDIP system cannot be controlled with good results when using only LQR because the system model derived as a linear model is not the exact the same as the actual system. Moreover, system parameters are subjected to change due to varying of operating conditions. The difference between the linear model and the actual model is called modeling error which creates system uncertainty. Therefore, in the controller design, robustness must be taken into account for the uncertainty. The LQR provides optimal performance for the nominal system but not for the system with model uncertainty. LQR cannot guarantee the stability and the optimal performance for the system subjected to parametric uncertainty. Neural networks based adaptive LQR is proposed to improve the performance of the balancing of the conventional LQR. Finally, the performance analysis of the proposed control algorithm from simulation and experiment are conducted. |
| Year | 2012 |
| Corresponding Series Added Entry | Asian Institute of Technology. Dissertation ; no. ISE-12-01 |
| Type | Dissertation |
| School | School of Engineering and Technology (SET) |
| Department | Department of Industrial Systems Engineering (DISE) |
| Academic Program/FoS | Industrial Systems Engineering (ISE) |
| Chairperson(s) | Manukid Parnichkun; |
| Examination Committee(s) | Afzulpurkar, Nitin V.;Vilas Wuwongse;Chevallereau, Christine; |
| Scholarship Donor(s) | Rajamangala University of Technology Isan KhonKaen Campus; |
| Degree | Thesis (Ph. D.) - Asian Institute of Technology, 2012 |