基于LPV模型的间歇过程系统辨识及其应用

项目名称： 基于LPV模型的间歇过程系统辨识及其应用

项目编号： No.61304141

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 自动化技术、计算机技术

项目作者： 周笋

作者单位： 厦门大学

项目金额： 25万元

中文摘要： 间歇生产在整个工业生产中的地位日益重要，非线性整体模型的辨识问题是间歇过程先进控制应用的瓶颈。本项目围绕具有温和非线性的一般间歇过程，针对其无固定稳态工作点的特点，研究基于线性变参数（LPV）模型的辨识方法。提出一种不需稳态工作点实验的，且能避免非线性搜索中中间模型不稳定问题的基于模型插值的LPV模型的辨识寻优技术；提出一种高斯调度函数，可用于工作点非等间隔分布情形；针对实际冶铁高炉热风炉的典型间歇过程，研究辨识中两级模型确认、工作点优化等特殊问题，分别建立基于模型插值的和基于标准正交基函数的LPV模型，分析模型准确度及算法鲁棒性，确立适合控制计算的非线性整体模型；按多目标分层优化模型预测控制（MPC）策略设计高炉热风炉控制算法，通过控制器的控制品质最终检验所研究的辨识方法。本项目的研究有利于合理解决间歇过程先进控制应用的瓶颈，以及促进冶铁业提高高炉风温，实现节约焦炭、降低CO2排放的效益

中文关键词： 间歇过程；非线性系统辨识；LPV模型；模型预测控制；

英文摘要： Batch production has become more and more important in industrial production. Identification of the whole nonlinear model of a batch process is the bottleneck of advanced control application. For general batch processes that has "mild" nonlinearity, according to the characteristics that the process parameters vary continuously and there is no stable working point, this project studies the identification approaches based on Linear Parameter-Varying (LPV) model. An identification optimation technology is proposed by which stable working point experiments are not necessary and the problem that the middle models may be unstable during the nonlinear searching procedure can be avoided. It can be applied to the LPV model identification of general batch processes including non-stable ones; a Gaussian scheduling function form is proposed for the cases when the working points are distributed with unequal intervals; for a typical batch process, a real hot blast stove of blast furnace of iron-making industry, study a series of basic problems like working point optimation, etc. during the practical identification process, establish the model-interpolation based LPV model and the orthonormal-basis-function based LPV model separately; along with the Model Predictive Control (MPC) strategy where the optimization is implemented

英文关键词： batch process；nonlinear system identification；LPV model；model predictive control；