项目名称: 生物反应过程的在线支持向量机建模与优化
项目编号: No.61273131
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 潘丰
作者单位: 江南大学
项目金额: 78万元
中文摘要: 生物反应过程由于其反应机理复杂、高度非线性和不确定性等特点,相对于其它工业过程,物理建模困难、其自动化程度还很低,已经成为当今国际过程控制领域研究的热点。本课题针对以上问题,研究生物反应过程的在线支持向量机(OSVM)建模与优化方法,主要研究内容包括:(1)提出多输出OSVM学习算法,同时优化OSVM算法在线学习步骤;(2)提出生物反应过程OSVM建模策略,实现模型参数在线校正,并且优化模型的参数;(3)将在线支持向量机和生物反应过程机理模型结合,提出串联、串并联、并联三种不同结构在线自校正混合建模策略;(4)开发用于实际生物反应过程的建模软件,实现先进的建模方法的实际应用;(5)运用所建模型,实现生物反应过程的多目标协同优化控制,并设计优化控制器。本项目属于应用基础研究,研究成果在生物反应工程等高附加值的间歇过程工业领域有广泛的应用前景。
中文关键词: 软测量建模;在线支持向量机;预测函数控制;优化控制;生化反应过程
英文摘要: Comparing with other industrial processes, bioprocess has complicated mechanism, high nonlinear and uncertainty issues. It is difficult to build physics model, and also the automation degree is low, so bioprocess has become hot point of international process control research. Aiming to solve these issues, bioprocess modeling and optimization strategies based on online support vector machines (OSVM) are studied, the main contents include: i) Multi-output OSVM learning strategy is proposed, and optimize the online learning steps of OSVM; ii) Bioprocess modeling based on OSVM is studied, the model parameters can be adjusted online, and optimize model parameters; iii) Combining OSVM and bioprocess mechanism models, three different structural (series, series-parallel and parallel) online self-adjust hybrid modeling strategies are proposed; iv) Bioprocess modeling software is explored, advanced modeling strategies could be used in field study; v) Multi-object optimization control of bioprocess is studied with these models, and design optimized controller. This project is belonging to basic application research; the results of this research can be widely applied to bioprocess engineer etc. high-added-value intermittent industrial processes.
英文关键词: Soft Sensor Modeling;Online Support Vector Regression;Model Predictive Control;Optimization Control;Bioprocess