【作者】 王晓华；

【导师】 邵诚；

【作者基本信息】 大连理工大学 ， 检测技术与自动化装置， 2008， 硕士

【摘要】 在工业过程中,有效的过程监控是保证生产安全的关键。通过监测生产过程的运行状态,及时有效地检测故障发生,从而来保证生产过程安全运行和提高产品质量一致性,是进行工业过程监控的目的。传统的用于过程监控的方法有基于数学模型和基于知识的方法。然而,准确详细的数学模型往往很难得到,而且经验知识有限,限制了这两种方法在过程监控中的应用。随着计算机控制技术的快速发展,大量的过程数据被采集并存储下来。如何充分利用这些存储的数据信息,提高过程监控能力,是过程控制领域的研究热点之一。因此,基于数据驱动的过程监控方法应运而生。常用的基于数据驱动的过程监控方法有主元分析方法和多向主元分析方法等。而这些方法受限于工业数据是否服从特定分布并依赖于预估的未来输出。本文针对这些局限性,提出了一种基于独立成分分析理论和步进多向主元分析理论的过程监控与故障诊断的新方法,即ICA_MPCA方法。该方法通过将观测变量转换为互不相关的独立成分,克服了工业数据需要服从特定分布的限制;同时利用步进多向主元分析方法建立了一系列主元分析模型,避免了传统方法在线监控时需要预测过程未来输出而产生的漏报错报现象。本文首先概述了工业过程监控的研究现状,介绍了传统的过程监控和故障诊断方法。其次,详细介绍了独立成分分析和多向主元分析的相关理论。再次,建立了基于独立成分分析理论和多向主元分析理论的过程监控与故障诊断的新方法。最后,利用新的方法给出了实现过程监控与故障诊断的一种具体方案。仿真试验验证了该方法的有效性。更多还原

【Abstract】 In process industry,the effective process monitoring is the key to ensure safety.The object of process monitoring is to ensure safety and provide products with consistent quality, through monitoring the state of the production process,detecting the fault promptly.Traditional methods include model and knowledge.It is difficult to achieve the exact mathematical model,which limits the application of them.With the rapid development of the computer technology,a great amount of process data can be sampled and collected.How to fully utilize this information to improve the performance of the process monitoring has been gradually becoming one of the focuses in the field of process control.Therefore,the process monitoring based on data-driven method comes out.Traditional data-driven methods are principal component analysis(PCA) and multi-way PCA methods.The effects of these methods are limited by the specific distribution data and they depend on the pre-estimated data.In view of these constraints,the new ICA_MPCA method,which is based on independent component analysis(ICA) theory and step-by-step MPCA theory,is established in this thesis.The ICA_MPCA method combines the advantages of ICA theory and step-by-step MPCA theory.The ICA theory transforms the observed variables to the independent components,which overcomes the specific distribution restriction.The step-by-step MPCA algorithm can avoid the wrong alarm through establishing a series of PCA to avoid pre-estimating the unknown part of the process variable trajectory deviation.The main research questions are as follows:First of all,a current situation overview of the process monitoring and fault diagnosis is introduced.The traditional methods of the process monitoring and malfunction diagnosing are also proposed in detail.Secondly,the relevant theories of ICA and MPCA are presented in detail.Thirdly,the ICA_MPCA method of the process monitoring and fault diagnosis is introduced.The new method includes ICA theory and step-by-step MPCA theory.Finally,a kind of specific industrial process monitoring and fault diagnosis step is proposed in this thesis.The method is evaluated on the industrial process.The results show the feasibility and superiority of the new method.更多还原