基于数据驱动的多模态工业过程监控理论及应用

项目名称： 基于数据驱动的多模态工业过程监控理论及应用

项目编号： No.61074079

项目类型： 面上项目

立项/批准年度： 2011

项目学科： 轻工业、手工业

项目作者： 侍洪波

作者单位： 华东理工大学

项目金额： 10万元

中文摘要： 本项目利用基于数据驱动的过程监控方法，通过系统分析化工过程中生产数据的实际特征，深入研究了如何对实际过程进行更为精确有效的监控方法，并取得了一些具有创新性的研究成果。一方面为了提高单个模型对非线性生产过程的监控效果，针对连续过程和间歇过程的不同特点，发掘新的监控策略并设计新的监控指标，同时研究利用新的优化算法与建模方式来提高模型精度；另一方面针对复杂生产过程中多个运行模态频繁切换的实际情况，通过对过程数据局部信息进行分析并加以利用，并结合多模型融合策略，提出新颖的多模态过程监控方案。相关研究成果总结整理后已分别在国内外专业期刊、学术会议发表。

中文关键词： 过程监控；故障检测；多模态过程；数据驱动；智能优化

英文摘要： In this project, after a fully analysis about the characteristics of the operating data in chemical industry processes, we put great research effort on how to monitor the real process accurately and effectively by using data-based monitoring methods, and obtained several innovative results. On the one side, based on the difference of continuous process and batch process, some new monitoring objects and monitoring statistics were designed to increase the efficiency of the single monitoring model in the nonlinear process. Furthermore, several novel optimum algorithms and modeling manner were proposed to improve the model accuracy. On the other side, by applying the way of utilizing the local information and the way of integrating multiple models, some novel multimode monitoring strategies were proposed. The summary of relevant research works have already published or accepted by the domestic or international journals and academic conferences.

英文关键词： Process monitoring; Fault detection; Multimode process; Data-driven;Intelligent optimization