Robust Controller Design for Systems with Constraints

该研究项目旨在开发同时处理约束和不确定性的过程控制算法。约束影响控制器的设计，因为虽然许多过程的动态行为可以用线性模型准确地描述，但系统的运行总是受制于约束，这些约束是非线性因素。控制器必须设计成能够处理这些非线性。这样的限制可能是由于执行器饱和或工艺限制(例如，某些温度和压力可能不会被超过)。不确定性是一个因素，因为控制设计者从来没有对任何实际工业过程进行完美的动态描述(模型)。系统是近似的，模型阶数和参数总是包含一定的误差，甚至可能随着运行条件的变化而变化。PI计划实现两种技术来处理这类问题，即模型预测控制(MPC)和反发条无扰转移(AWBT)。预测控制要求将控制目标表示为单一的二次目标函数，并将约束转化为线性不等式约束。AWBT控制器的设计分两步进行，首先设计一个可接受的线性鲁棒控制器，忽略所有的非线性环路元件和目标。然后，增加了一个补偿方案，用于处理约束等非线性问题。本研究将分三步进行：(1)设计一种新的鲁棒线性预测控制控制器。它将通过一个旨在满足所有过程约束的优化器以非线性方式实现，同时最大限度地减少与线性行为的偏差。(2)开发一种AWBT综合程序，在平滑开关性能、测量噪声灵敏度和稳健性之间进行权衡。(3)建立机制，以有效的方式将实际目标和约束转化为货币政策委员会或AWBT形式。

This research project is aimed at developing process control algorithms to handle both constraints and uncertainty. Constraints affect controller design because while the dynamic behavior of many processes can be accurately described by a linear model, system operation is always subject to constraints which are nonlinear elements. The controller has to be designed to handle these nonlinearities. Such constraints may arise from actuator saturation or processing limitations (certain temperatures and pressures may not be exceeded, for example). Uncertainty is a factor because the control designer never has a perfect dynamic description (model) of any real industrial process. The system is approximated and model order and parameters always contain some error, and may even change with operating conditions. The PI plans to implement two techniques for dealing with such problems, Model Predictive Control (MPC) and Anti-Windup Bumpless-Transfer (AWBT). MPC requires that the control objective is expressed in terms of a single quadratic objective function and the constraints are translated into linear inequality constraints. AWBT controller design proceeds in two steps where first an acceptable linear robust controller is designed neglecting all nonlinear loop elements and objectives. Then a compensation scheme is added which deals with the nonlinear issues such as constraints. This research will proceed in three steps: (1) A new robust linear MPC controller will be designed. It will be implemented in a nonlinear manner through an optimizer which aims at satisfying all process constraints while minimizing the deviation from linear behavior. (2) Development of an AWBT synthesis procedure that balances the trade-offs between smooth switching performance, measurement noise sensitivity and robustness. (3) Development of mechanisms to translate practical objectives and constraints into an MPC or AWBT formalism in an effective manner.