上海科技大学 | English | 地图
首页> 新闻信息> 活动
Numerical Methods for Fast Optimal Control of Mechatronic Systems
Date:2014/2/16     Browse:855

Speaker: Boris Houska, Shanghai Jiaotong Univ.

Time: 2/19, Wed., 2014, 4:50~5:50pm

Location: Room 220, Building 8#

Abstract:

In the last 15 years computation times of real-time algorithms for nonlinear model predictive control have decreased by a factor of 100 000. This tremendous speed-up is not only due to faster computer hardware, but also due to algorithm development and novel software implementations. In this talk we review these algorithmic and software developments of numerical methods for real-time model predictive control. We demonstrate that nowadays receding horizon control problems with 20-30 states and several time-varying control variables can be solved within a few milliseconds. This allows us to apply nonlinear real-time optimal control to challenging applications such as mechatronic systems with fast sampling times. We present the software environment ACADO Toolkit as well as its extensions towards automatic code generation. It features an optimal control code that is based on a real-time variant of a multiple-shooting based Gauss-Newton method. The nonlinear feedback control algorithm and its potential will be illustrated with challenging real-world applications including the control of fast flying tethered airplanes for wind power generation. We conclude the talk with a vision of future research directions in real-time optimal control touching the fields economic model predictive control as well as distributed optimal control algorithms which have the potential to experience a similar speed-up in the coming 15 years.

Bio:

Boris Houska is a lecturer at the Department of Automation at Shanghai Jiao Tong University, where he has been since April 2013. He obtained his Ph.D. in 2011 in Electrical Engineering at the Optimization in Engineering Center (OPTEC) at KU Leuven (Belgium) after studying mathematics and physics at the university of Heidelberg (Germany) in 2003-2008. From 2012 to 2013 he was a postdoctoral researcher at the Centre of Process Systems Engineering at Imperial College London (UK). His research interests include numerical optimization and optimal control, robust and global optimization, as well as fast model predictive control algorithms.  

SIST-Seminar Series-14001