基于核主元分析与支持向量机的监控诊断方法及其应用

摘要

In order to monitor the imperial smelting furnace (ISF) state in time and accurately diagnose the faults，a fault diagnosis approach based on kernel principal component analysis (KPCA) and multi-class classifiers of support vector machine (SVM) was proposed. The principle of the method was as follows: Firstly, the KPCA approach was adopted to extract the feature and the monitoring model was established. Secondly, the SVM multi-class classifiers with 'one to other' algorithm was used for classification with the input of the feature. The experimental results show that, compared with the features extracted by principal component analysis (PCA), the proposed method increases the separability of the data set, performs better recognition ability, and it can be used in the imperial smelting furnace(ISF) fault diagnosis.%为了及时反映密闭鼓风炉冶炼过程状态,实现对密闭鼓风炉炉况的监控与诊断,提出核主元分析和多支持向量机分类的相结合的过程监控与故障诊断方法.其原理是:首先,用核主元分析方法提取过程数据特征,建立核主元分析的监控模型;然后,将代表过程特征的核主元送入多支持向量机分类器中,利用"一对其余"算法对故障进行诊断与分类.实验结果表明,所提出的方法与传统的主元分析方法相比,整个样本集的可分性变大,分类正确率提高,能更准确地诊断炉子的各种故障,可有效地用于密闭鼓风炉冶炼过程的故障诊断.