改进KICA带钢热连轧过程故障检测方法研究  

Research on Fault Detection Method of Improved KICA Hot Strip Mill Process

作　　者：张瑞成 裴然 ZHANG Rui-cheng;PEI Ran(College of Electrical Engineering,North China University of Science and Technology,Hebei Tangshan 063210,China)

机构地区：[1]华北理工大学电气工程学院,河北唐山063210

出　　处：《机械设计与制造》2021年第12期178-182,共5页Machinery Design & Manufacture

基　　金：河北省自然科学基金资助项目(F2018209201)。

摘　　要：针对传统的FastICA算法对初始权值的敏感性,以及带钢热连轧工业过程数据间呈现非线性和非高斯性的问题,提出一种基于改进的快速核独立元分析(FastKICA)方法进行故障检测。该方法首先对原始数据进行小波包去噪,利用核方法对数据进行特征重构,使数据线性结构化,引入超松弛因子改进FastICA初始权值的选择机制,提取更合理的独立元信号,并且计算相应的统计量和控制限。选用带钢热连轧工业过程实测数据进行仿真,与传统的FastKICA方法相比较,结果证明,基于超松弛因子改进的FastKICA故障检测的方法准确率更高,验证了该方法的有效性和优越性。Aiming at the sensitivity of the traditionalFastICA algorithm to the initial weight and the nonlinear and non-Gaussian data between the hot strip mill process,a fast kernel independent component analysis(FastKICA)method is proposed to solve the fault detection. The method firstly performs wavelet packet denoising on the original data,uses the kernel method to reconstruct the data,linearizes the data,introduces the over-relaxation factor to improve the selection mechanism of the initial weight of FastICA,and extracts a more reasonable independent component signal.Then calculate the corresponding statistics and control limits. The simulation data of strip hot strip mill industrial process is selected for simulation. Compared with the traditional FastKICA method,the results show that the method based on over-relaxation factor improved FastKICA fault detection has higher accuracy,which verifies the effectiveness and superiority of the method.

关 键 词：故障检测 小波包分析 核独立元分析 超松弛因子 快速固定点算法 

分 类 号：TH16[机械工程—机械制造及自动化] TP277[自动化与计算机技术—检测技术与自动化装置]