基于高效复合多尺度模型的微纳过程控制方法研究

项目名称： 基于高效复合多尺度模型的微纳过程控制方法研究

项目编号： No.61473287

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 自动化技术、计算机技术

项目作者： 倪东

作者单位： 浙江大学

项目金额： 81万元

中文摘要： 纳米技术因其广泛的应用而引起科技界的高度重视。近年来随着纳米技术的发展，其微细化逼近物理极限，微观结构日趋复杂，对纳米加工制造技术的要求日益严苛，特别是大规模批量加工制造的难度急剧增加。核心问题之一是缺乏微纳尺度控制的系统方法，即用宏观尺度的外力去驱动微纳尺度系统的动态，实现有效控制，并且能保证其可重复性和稳定性。本研究提出一种切实可行的复合多尺度模型预测控制方法来有效控制微纳尺度系统。该方法运用随机偏微分方程模型（sPDE）描述微纳尺度动态，并结合宏观尺度的偏微分方程模型（PDE）构建封闭形式的复合多尺度模型。在保留微纳尺度随机离散特性的前提下，该封闭形式模型可进一步被降阶为常微分方程组（ODE）/随机常微分方程组(sODE)系统，使计算复杂度满足实时计算需求，并且易于应用模型预测控制理论进行控制器设计。本研究还将提出闭环系统中多尺度模型误差的修正方案并进行控制系统的实验验证。

中文关键词： 多尺度系统；模型降阶；微纳尺度；随机系统；分布参数

英文摘要： Nanotechnology has received tremendous attentions due to its numerous applications in many critical fields, however, its recent developments have driven its critical dimensions to physical limits and microstructures to extreme complexity. The bar for nanofabrication performance requirements, especially for mass production, has rapidly rising to sky high particularly in the last decade and posed a huge challenge to the industry. One of the most fundamental problems for nanofabrication is the lack of practical solutions to use macroscopic driving force to steer and control microscopic dynamics with good repeatability and stability. This research proposed a innovative and practical solution to address this problem. A hybrid multi-scale model, with stochastic partial differential equation (sPDE) depicting the microscopic dynamics, and deterministic partial differential equation (PDE) describing the macroscopic dynamics, is used to model a general nanofabrication process, so that the stochastic and discrete characteristics could be retained, while the computational complexity are drastically reduce to meet real-time application needs because the closed-form solution of the model and model reduction method to further reduce it to an ordinary differential equations (ODE)/stochastic ordinary differential equations (sODE) system. Furthermore, classical model predictive control (MPC) methods could thus be applied. The research also proposed a modelling error reduction method for the closed-loop system, as well as conducted real world validation experiments of the proposed control system.

英文关键词： Multiscale;Model Reduction;Microscale;Stochastic Process;Distributed Parameters