Loading...
Search
Search in this resource
sort by
Send to Friend
 Digital Object List
 Bookmark
 Review List
  1. کاهش بار محاسباتی در کنترل پیش‌بین سیستم‌های غیرخطی چندورودی

Computational Load Reduction in Model Predictive Control of Multi-Input Nonlinear Systems

Adelipour, Saeed | 2020

461 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 52836 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Haeri, Mohammad
  7. Abstract:
  8. In this thesis, computational load reduction in linear matrix inequality based model predictive control for nonlinear multi-input systems is studied. The main idea is decomposing the system into several smaller interacting subsystems and computing each control input separately. The main challenge is how to consider the effects of interactions among subsystems in the control design. Three approaches are investigated in this thesis. In the first approach, a static approximation of the effects of other subsystems is augmented in the local model of each subsystem, to improve the closed loop performance than fully decentralized methods. In the second approach, considering the mutual interactions as external disturbances, each local robust decentralized model predictive controller is designed as a state feedback control law minimizing the local infinite horizon objective function while robust stability for each subsystem is attained using the concept of robust positively invariant sets. The attenuation of the amount of coupling disturbance on each subsystem is established at each sample time to improve the decentralized performance. Moreover, the extension of the proposed results to tackle uncertain nonlinear systems and its ability to deal with reconfigurable large-scale systems, where subsystems can leave or join the system, are also discussed. Next, for the non-sparse systems with several inputs, which cannot be properly decomposed into some non-overlapping subsystems or networks of nonlinear systems with strong interactions where robust decentralized methods produce conservative results, a cooperative distributed approach is proposed. In this approach, each control law is obtained separately as a state feedback of all system’s states, exploiting the state and input information of other subsystems, while a global infinite horizon objective function is optimized by all subsystems. To further reduce the computational load, only one input can be optimized at each sample time while keeping the other inputs at the previous feasible values
  9. Keywords:
  10. Large Scale System ; Linear Matrix Inequality (LMI) ; Decentralized Controller ; Distributed Control ; Nonlinear Predictive Control ; Computational Load Reduction ; Multi-Input Systems

 Digital Object List

 Bookmark

No TOC