主元分析在电厂故障诊断中的应用
发布时间：2018-11-12 12:59
【摘要】：在过程监测中,随着对系统稳定性与安全性需求的增加,故障检测与诊断也越发重要。及时、有效、准确的检出故障对工业过程的经济性与安全性都有非常大的意义,及时检出故障,则有助于降低维修成本,并且避免长期故障引起的设备损坏,若不能及时检出故障,不仅在经济方面造成损失,严重的甚至会威胁现场工作人员的生命安全。 作为一种基于数据驱动的故障诊断方法主元分析分析方法,需要稳态工况下的数据进行建模。传统主元分析方法,其T2统计量与SPE统计量控制限均为固定的,然而,在过渡工况下,使用固定的控制限,会引发大量的误报,这严重影响主元分析的检测性能。同时,从系统获取的数据含有大量的噪声,导致主元分析检测性能下降。 本文首先介绍了故障诊断的基础知识及故障分类、故障诊断方法及其分类,深入研究了主元分析的故障诊断机理及方法。针对主元分析在过渡工况中误报率较高的问题,研究了T2统计量方差自适应控制限。然而,采用方差自适应控制限增加了故障的漏报率,因此,本文就T2统计量方差自适应控制限与EWMA滤波展开研究,将方差自适应控制限的主元分析方法首次用于火电厂故障诊断中,有效地降低了主元分析T2统计量在过渡工况中的误报情况；并且方差自适应控制限与EWMA滤波相结合,在不增加误报的情况下有效地提高了主元分析检出微小故障的能力。 主要研究工作如下： 1、对故障诊断基础进行调研,研究了故障的分类及故障诊断类型的分类。本文介绍了两类故障诊断方法：1)基于模型的故障诊断方法；2)基于数据的故障诊断方法,并分析对比了其方法的优缺点。 2、深入研究了经典主元分析方法,为了解决经典主元分析误报的问题,研究了统计量用自适应控制限的方法,并且深入研究了主元分析与EWMA滤波相结合的方法,提出方差自适应控制限与EWMA相结合的故障诊断方法,在降低误报率的同时,提高了主元分析检出微小故障的性能。 3、采用数值仿真实验验证方差自适应控制限与EWMA相结合的方式可以有效降低误报,同时提高了主元分析检出微小故障的性能,并且将这种方法用于火电厂燃烧故障诊断过程中。
[Abstract]:In process monitoring, with the increasing demand for system stability and security, fault detection and diagnosis become more and more important. Timely, effective and accurate detection of faults is of great significance to the economy and safety of industrial processes. Timely detection of faults will help to reduce maintenance costs and avoid equipment damage caused by long-term failures. If failure is detected in time, it will not only cause economic losses, but also threaten the safety of workers. As a data-driven principal component analysis method for fault diagnosis, it is necessary to model the data under steady condition. The control limits of T2 and SPE statistics of traditional principal component analysis methods are fixed. However, under transient conditions, the use of fixed control limits will lead to a large number of false positives, which seriously affect the detection performance of principal component analysis. At the same time, the data obtained from the system contain a lot of noise, which leads to the degradation of the performance of principal component analysis (PCA) detection. In this paper, the basic knowledge and classification of fault diagnosis, the method and classification of fault diagnosis are introduced, and the principle and method of fault diagnosis based on principal component analysis (PCA) are studied. Aiming at the problem of high false alarm rate of PCA in transient condition, the adaptive control limit of variance of T2 statistic is studied. However, the error rate is increased by using variance adaptive control limit. Therefore, the T2 statistic variance adaptive control limit and EWMA filter are studied in this paper. The principal component analysis method of variance adaptive control limit is applied to fault diagnosis of thermal power plant for the first time, which effectively reduces the misinformation of T2 statistic in transient condition. And the combination of variance adaptive control limit and EWMA filter can effectively improve the ability of principal component analysis (PCA) to detect small faults without adding false positives. The main research work is as follows: 1. The basic of fault diagnosis is investigated, and the classification of fault diagnosis and fault diagnosis is studied. This paper introduces two kinds of fault diagnosis methods: (1) model-based fault diagnosis and (2) data-based fault diagnosis, and analyzes and compares their advantages and disadvantages. 2. The classical principal component analysis method is deeply studied. In order to solve the problem of false positives in classical principal component analysis, the method of adaptive control limit is studied, and the method of combining principal component analysis with EWMA filter is studied. A fault diagnosis method based on variance adaptive control limit and EWMA is proposed, which can reduce the false alarm rate and improve the performance of principal component analysis (PCA) to detect small faults. 3. Numerical simulation experiments show that the combination of variance adaptive control limit and EWMA can effectively reduce false positives and improve the performance of principal component analysis (PCA) to detect small faults. And this method is used in the process of combustion fault diagnosis in thermal power plant.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM62
本文编号：2327175
[Abstract]:In process monitoring, with the increasing demand for system stability and security, fault detection and diagnosis become more and more important. Timely, effective and accurate detection of faults is of great significance to the economy and safety of industrial processes. Timely detection of faults will help to reduce maintenance costs and avoid equipment damage caused by long-term failures. If failure is detected in time, it will not only cause economic losses, but also threaten the safety of workers. As a data-driven principal component analysis method for fault diagnosis, it is necessary to model the data under steady condition. The control limits of T2 and SPE statistics of traditional principal component analysis methods are fixed. However, under transient conditions, the use of fixed control limits will lead to a large number of false positives, which seriously affect the detection performance of principal component analysis. At the same time, the data obtained from the system contain a lot of noise, which leads to the degradation of the performance of principal component analysis (PCA) detection. In this paper, the basic knowledge and classification of fault diagnosis, the method and classification of fault diagnosis are introduced, and the principle and method of fault diagnosis based on principal component analysis (PCA) are studied. Aiming at the problem of high false alarm rate of PCA in transient condition, the adaptive control limit of variance of T2 statistic is studied. However, the error rate is increased by using variance adaptive control limit. Therefore, the T2 statistic variance adaptive control limit and EWMA filter are studied in this paper. The principal component analysis method of variance adaptive control limit is applied to fault diagnosis of thermal power plant for the first time, which effectively reduces the misinformation of T2 statistic in transient condition. And the combination of variance adaptive control limit and EWMA filter can effectively improve the ability of principal component analysis (PCA) to detect small faults without adding false positives. The main research work is as follows: 1. The basic of fault diagnosis is investigated, and the classification of fault diagnosis and fault diagnosis is studied. This paper introduces two kinds of fault diagnosis methods: (1) model-based fault diagnosis and (2) data-based fault diagnosis, and analyzes and compares their advantages and disadvantages. 2. The classical principal component analysis method is deeply studied. In order to solve the problem of false positives in classical principal component analysis, the method of adaptive control limit is studied, and the method of combining principal component analysis with EWMA filter is studied. A fault diagnosis method based on variance adaptive control limit and EWMA is proposed, which can reduce the false alarm rate and improve the performance of principal component analysis (PCA) to detect small faults. 3. Numerical simulation experiments show that the combination of variance adaptive control limit and EWMA can effectively reduce false positives and improve the performance of principal component analysis (PCA) to detect small faults. And this method is used in the process of combustion fault diagnosis in thermal power plant.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM62
【参考文献】
相关期刊论文 前6条
1 王海清,宋执环,王慧;PCA过程监测方法的故障检测行为分析[J];化工学报;2002年03期
2 邱天;白晓静;郑茜予;朱祥;;多元指数加权移动平均主元分析的微小故障检测[J];控制理论与应用;2014年01期
3 徐涛;王祁;;基于小波包的多尺度主元分析在传感器故障诊断中的应用[J];中国电机工程学报;2007年09期
4 张曦;阎威武;刘振亚;邵惠鹤;;基于核主元分析和邻近支持向量机的汽轮机凝汽器过程监控和故障诊断[J];中国电机工程学报;2007年14期
5 邱凤翔;司风琪;徐治皋;;电站关联规则的主元分析挖掘方法及传感器故障检测[J];中国电机工程学报;2009年05期
6 邱天;刘吉臻;牛玉广;;电站锅炉主元分析建模中的数据选取[J];中国电机工程学报;2009年08期
，本文编号：2327175
本文链接：https://www.wllwen.com/kejilunwen/dianlilw/2327175.html
教材专著
