项目名称: 基于数据的非高斯多模态工业过程监测
项目编号: No.61273163
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: 张颖伟
作者单位: 东北大学
项目金额: 81万元
中文摘要: 对于电熔镁炉等复杂工业流程,数据本身具有复杂的统计特性。由于过程变量的轨迹随着时间呈现很强的非线性非高斯动态变化趋势,因此,线性高斯方法应用于此存在无法解决的非线性非高斯动态带来的问题。本项目研究基于核主元、核独立元分析方法动态数据驱动方法解决故障检测方法和故障辨识方法。目前,核主元、核独立元分析方法只适用于单一静态模式,对于动态过程检测则会产生误报和漏报问题。本项目研究的动态特性分为如下三种:单一模式局部动态;多模式动态;模式间切换动态。本项目旨在: (1) 提出非线性非高斯多模态过程建模方法;(2) 非线性非高斯动态过程故障诊断方法;(3)动态故障分离统计量的定义。本项目的研究成果将提高电熔镁炉运行的安全可靠性。
中文关键词: 过程监测;数据驱动;多模态;;
英文摘要: The electro-fused magnesia furnace (EFMF) is complex industrial equipment. The operation data have complex statistical characteristics. Strong nonlinearity, non-Gaussianity and dynamic are the features of the process variables. Hence, linear and Gaossian methods are not available for nonlinear and non-Gaussian processes. Recently, kernel principle component analysis and kernel independent component analysis were proposed which belong to static approaches. There are false and missing alarms when they are applied to dynamic processes. The dynamic characteristic is divided into three kinds: local dynamic of single mode, multimode dynamic and switching dynamic between modes. Purpose of the project is: (1) nonlinear and multimode modeling methods are proposed; (2) nonlinear and multimode monitoring methods are proposed; (3) The dynamic fault isolation statistics are given. The safety and reliability of the electro-fused magnesia furnace will be improved by using the proposed approaches.
英文关键词: Process monitoring;Data driven;multi-mode;;