质量相关的多工况动态间歇过程建模及故障诊断方法研究

项目名称： 质量相关的多工况动态间歇过程建模及故障诊断方法研究

项目编号： No.61473033

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： 彭开香

作者单位： 北京科技大学

项目金额： 85万元

中文摘要： 现代过程工业正逐渐倚重于生产小批量、多品种、高附加值产品的间歇工业过程。通过过程监控来保障间歇过程的安全可靠运行以及连续稳定的产品质量逐渐成为过程控制的首要任务。由于复杂间歇过程机理复杂、规模庞大、工况多变、运行环境恶劣、边界条件模糊等原因，难以建立精确的过程监控机理模型，数据驱动方法成为过程监控的主流技术。本项目以带钢热连轧这一多工况复杂动态间歇过程为背景，采用数据驱动思路研究质量相关的多工况、动态、非高斯数据特性下的建模与故障诊断理论与技术，拟研究和解决以下几方面问题：（1）质量相关的多工况动态间歇过程模态识别与划分技术；（2）质量相关的多工况复杂动态间歇过程监控技术；（3）需求改变导致的监测模型移植技术与过程及环境时变驱动的监测模型更新策略；（4）在带钢热连轧生产过程中进行验证和应用。旨在建立一套适用于多工况、动态、非高斯数据特性下质量相关的复杂动态间歇过程故障诊断方法与技术。

中文关键词： 动态间歇过程；故障诊断；多工况；质量相关；数据驱动

英文摘要： The focus of modern process industry has shifted to the production of higher-value-added products through batch processes. Thus, performing valid process monitoring (PM) is a principal task of ensuring process safety and product quality for batch processes. Due to the complex operation mechanism, chaotic environment and vague boundary conditions, it's quite tough to build an accurate physical model and implement model-based PM techniques for batch processes. As a result, data-driven approaches as alternative become more and more popular. This project takes the hot strip mill process (HSMP), which is a complex batch process, as the research background and attempts to study advanced theories and methodologies of PM for multi-mode, dynamic and non-Gaussian batch processes. The problems addressed in the project include: (1)Quality-relevant mode extraction and identification for dynamic batch processes under multi-mode running environment; (2)Quality-relevant PM technique for complex dynamic batch processes under multi-mode producing condition; (3)Process model migration technique corresponding the change of the producing demand, and the model updating strategy associated with the running environment; (4)Application of the new proposed approaches in the real hot strip mill process and testify their applicability. The project aims at establishing a set of quality-relevant PM schemes for the complex dynamic batch processes that could effectively work under the multi-mode, dynamic and non-Gaussian environment.

英文关键词： dynamic batch process;fault diagnosis;multi-mode;quality-relevant;data driven