论文部分内容阅读
针对过程工业数据中所含的噪声和干扰信号、过程工业的非线性及基于主元分析(Principal Component Analysis,PCA)的统计性能监控法由于不用过程机理模型的信息从而对故障诊断问题难以在理论上作系统分析的缺陷,提出基于小波变换核主元分析和多支持向量机的过程监控方法,该方法首先采用基于小波变换的收缩阈值去噪法对建模数据进行预处理,以有效抑制过程数据中所含的噪声和干扰信号,然后利用核主元分析来进行故障特征的提取,从而提高非线性统计过程监控的准确性;最后提出多支持向量机用来对故障的来源进行分类,以避免求解核主元空间到原始空间的逆映射.将该方法应用到对TE(Tennessee Eastman,TE)过程的监控,表明了所提出方法的有效性,为过程的监控和故障诊断提供了一个新的方法.
In view of the noise and interference signals contained in the process industry data, the non-linearity of the process industry and the statistical performance monitoring method based on Principal Component Analysis (PCA), it is difficult for the fault diagnosis problem to be theoretically solved without the information of the process mechanism model The author puts forward a process monitoring method based on kernel principal component analysis (PCA) and multi-support vector machine (SVM) based on wavelet transform. This method first uses the shrinkage threshold denoising method based on wavelet transform to preprocess the modeling data to effectively suppress the process Data contained in the noise and interference signals, and then use the kernel principal component analysis of the fault feature extraction, thereby enhancing the accuracy of nonlinear statistical process monitoring; Finally, a multi-support vector machine is used to classify the source of the fault to Avoiding the inverse mapping of the space of the principal component space to the original space.This method is applied to the monitoring of TE (Tennessee Eastman, TE) process, which shows the effectiveness of the proposed method and provides a new method for process monitoring and fault diagnosis Methods.