主题: Model Predictive Control and Estimation of Linear Transport-reaction Systems
主讲人: Stevan Dubljevic
地点: 松江校区二号学院楼224室
时间: 2018-12-11 10:30
组织单位:信息科学与技术学院数字化纺织服装技术教育部工程研究中心
报告人简介：
Stevan Dubljevic，加拿大阿尔伯塔大学副教授，研究方向包括分布式参数系统控制和模型预测控制、材料和化学过程的动力学和优化。在Automatica、JPC等控制领域顶级期刊上发表论文50余篇，2012年以来承担12项科研项目，包括加拿大NSERC项目10项，与企业联合开发项目2项。2005年在加州大学洛杉矶分校获博士学位。曾在加州大学洛杉矶分校David Geffen医学院心脏病学系担任独立博士后研究员（2006-2009）。曾获美国心脏协会（AHA）西方国家联盟博士后奖学金（2007-2009），2007年获得美国自动控制委员会（AACC）O. Hugo Schuck应用技术获，2012年获得加拿大化学工程师学会的DG Fisher奖。
报告简介:
Distributed parameter systems (DPS) are ubiquitouslypresent as models of fundamental conservation laws and in process control,manufacturing, transport systems and/or human society. The major drawback ofDPS models is that they take form of partial differential equations containinghigher order derivatives in space and time. The complexity of a partialdifferential equation (PDE) in the case of linear PDE models lies in necessityof modelers to account for model spatial characteristics by an approximatingunderlying model through some spatial approximation arriving to a finitedimensional model representation amenable for subsequent control, observerand/or monitoring device design.
This work provides foundation for systematicdevelopment of modelling framework for a linear DPS system which uses a finiteand low dimensional setting for the controller/observer/estimator designwithout application of any spatial approximation or order reduction. Inparticular, we are interested in formulating control design methodology for ageneral class of linear DPS systems which in this work account for an optimalconstrained optimization-based setting.
In addition to classical chemical process systems,we also address wave and beam equation system which accounts for a large classof distributed parameter systems. In this work, the discrete model of adistributed parameter system is obtained by using energy preserving Cayley-Tustindiscretization. Discrete DPS models are low dimensional, energy preserving and donot dissipate numerically. In particular, discrete setting is amenable to anexplicit, economic and/or a classical model predictive control settingrealization, with emphasize on the different slight variations in realizationof constrained finite dimensional controllers. Having this in mind, the modelpredictive control is designed by utilizing standard optimal control law with inputor/and state/output constraints. The issues of stabilization, optimality andconstrained stabilization are addressed for an infinite-dimensional system inthis work. In addition, we also address the state estimation in this setting whichallows practitioners to extend freely finite dimensional concepts to the PDE models. Finally, thecontroller performance is assessed by numerical simulation with application ondifferent distributed parameter systems.