摘要
依托于固定约束条件,进行离心血浆分离机故障容错诊断,使得诊断方法呈现出较低的AUC值。因此,提出基于局部性修复的离心血浆分离机故障容错诊断研究。深入分析离心血浆分离机的运行特征,构建设备故障模型。采用奇异值分解技术,完成特征数据的提取与选择。再结合深度森林算法,设计一种可用于故障容错诊断的改进级联森林框架。最后,以局部性修复理念为基础,设计离心血浆分离机重调度操作算法,建立故障容错分析涉及的动态约束条件,得出更加准确的诊断结果。实验结果表明:所提方法的AUC值为0.98,与基于SVM的诊断方法、基于RF的诊断方法相比,AUC值提升了13%和16%,有效降低了故障容错诊断误差。
Based on the fixed constraints,the fault-tolerant diagnosis of centrifugal plasma separator is carried out,which makes the diagnosis method show a low AUC value.Therefore,the fault-tolerant diagnosis of centrifugal plasma separator based on local repair is proposed.The operation characteristics of centrifugal plasma separator are deeply analyzed,and the equipment fault model is constructed.The singular value decomposition technology is used to extract and select the feature data.Combined with the deep forest algorithm,an improved cascaded forest framework for fault tolerance diagnosis is designed.Finally,based on the concept of local repair,the rescheduling operation algorithm of centrifugal plasma separator is designed,the dynamic constraints involved in fault tolerance analysis are established,and more accurate diagnosis results are obtained.The experimental results show that the AUC value of the proposed method is 0.98.Compared with the SVM based diagnosis method and RF based diagnosis method,the AUC value is increased by 13%and 16%,which effectively reduces the fault-tolerant diagnosis error.
作者
陈菊
周梅玲
张琼
何圆圆
CHEN Ju;ZHOU Meiling;ZHANG Qiong;HE Yuanyuan(Department of Hematology,Taihe Hospital,Shiyan Hubei 442000,China)
出处
《自动化与仪器仪表》
2023年第1期276-280,共5页
Automation & Instrumentation
关键词
局部性修复
离心血浆分离机
运行故障
容错
特征数据
深度森林
local repair
centrifugal plasma separator
operation failure
fault tolerance
characteristic data
deep forest
