基于MSPCA-KECA的冷水机组故障监测及诊断

发布时间：2018-04-27 15:44

本文选题：故障诊断 + 多尺度主元分析　；参考：《化工学报》2017年04期

【摘要】：针对冷水机组同类型不同等级故障的变量变化存在差异会造成误诊断的问题,提出一种基于多尺度主元分析-核熵成分分析(MSPCA-KECA)的故障诊断策略。MSPCA提取故障特征,其输出作为KECA分类器的输入,实现故障的实时监测与自动诊断。首先,改进的MSPCA算法通过将小波多尺度分析与主元分析相结合,筛选故障信息可能存在的尺度直接重构并采用PCA提取故障特征,获取不同类型故障之间差异的同时也保留了同类型但不同等级故障之间的相似性,提高故障诊断的可靠性。之后建立KECA非线性分类器并引入一种新的监测统计量——散度测度统计量,使降维后不同特征信息之间呈现显著的角度差异,易于分类。最后,采用支持向量数据描述(SVDD)算法确定新统计量的控制限,以克服无法获知统计量分布的问题。通过对冷水机组数据的仿真研究,验证了MSPCA-KECA方法的可行性及有效性。
[Abstract]:In view of the problem that the variation of variables of the same type and different grade faults of chillers will lead to misdiagnosis, a fault diagnosis strategy based on multi-scale principal component analysis-kernel entropy component analysis (MSPCA-KECA-KECA-MSPCA-KECA-based) is proposed to extract fault features. The output is used as the input of KECA classifier to realize real-time fault monitoring and automatic diagnosis. Firstly, the improved MSPCA algorithm combines wavelet multiscale analysis with principal component analysis to screen the possible scale reconstruction of fault information and extract fault features using PCA. In order to improve the reliability of fault diagnosis, the differences between different types of faults are obtained and the similarities between the same types but different grades of faults are retained. Then the KECA nonlinear classifier is established and a new monitoring statistic, the divergence measure statistic, is introduced, which makes the difference of angle between the different feature information after dimensionality reduction is significant, and it is easy to classify. Finally, the support vector data description (SVDDD) algorithm is used to determine the control limit of the new statistics, so as to overcome the problem that the distribution of statistics can not be obtained. The feasibility and effectiveness of the MSPCA-KECA method are verified by the simulation of the chiller data.
【作者单位】：内蒙古工业大学电力学院;北京工业大学电子信息与控制工程学院;
【基金】：国家自然科学基金项目(61364009,21466026,61640312) 内蒙古自治区自然科学基金项目(2015MS0615)~~
【分类号】：TP277

【参考文献】