一种基于改进核Fisher的故障诊断方法

发布时间：2019-01-04 16:56

【摘要】：针对化工过程故障数据呈非线性分布,数据类别复杂,难以进行故障诊断的问题,提出一种改进核Fisher的故障诊断方法。对于原始样本数据投影后,样本出现因类间距离存在很大差异性而导致部分类别样本存在混叠的现象,以及不同类别的边界数据归类模糊问题,给出了统一的解决办法。该方法首先采用改进类间距的方法来改变样本投影空间的分布,使得样本具有较好的投影效果,然后通过定义阈值参数来筛选出边界数据,对于边界数据,采用改进的K近邻(K-NN)算法来分类,对于非边界数据,采用马氏距离来分类。最后在TE过程中进行了仿真验证,结果表明方法在兼顾了故障诊断时间的同时,有效提高了故障诊断精度。
[Abstract]:Aiming at the problem of nonlinear distribution of fault data in chemical process, complicated data categories and difficulty in fault diagnosis, an improved fault diagnosis method based on kernel Fisher is proposed. After the original sample data is projected, a unified solution is proposed to solve the problem that some of the samples are aliasing due to the great differences in the distance between classes, and the boundary data of different categories are classified vaguely. In this method, the improved class spacing method is used to change the distribution of the sample projection space, so that the sample has a better projection effect, and then the boundary data is filtered by defining the threshold parameters. The improved K-nearest neighbor (K-NN) algorithm is used to classify the unbounded data, and the Markov distance is used to classify the unbounded data. Finally, the simulation results in TE show that the method not only takes into account the time of fault diagnosis, but also improves the accuracy of fault diagnosis effectively.
【作者单位】：北京理工大学自动化学院;
【基金】：国家自然科学基金项目(61503027)~~
【分类号】：TQ050.7;TP277

【相似文献】