摘要
目的提出一种基于医疗电子设备故障(简称故障)事件分布概率、用于替代人工经验传授、可持续更新的故障识别系统方法研究。方法首先设计故障特征多维矩阵,将故障事件、故障特征等信息设计成一套故障知识库,其次采用主成分分析法,将多维特征矩阵通过降维的方式输出各故障事件的分布概率,从而以最高分布概率的故障事件作为识别结果。还设计一套闭环系统,用于通过系统提高数据处理与计算的能力。最后,采集到的部分指标通过故障事件分布概率方法计算,得到婴儿培养箱和心电监护仪的故障识别效果。结果故障的婴儿培养箱采用故障事件分布概率方法识别的同时,引入了人工检测进行验证,得出同样的结论为风道故障,并且识别率为61.2%。而后通过闭环系统对监护仪的故障进行基于故障事件分布概率的方法识别,也与实际结果一致。结论基于故障事件分布概率的医疗设备故障识别方法可以将专家的经验知识转化为一套标准科学的自动化检测系统和应用故障数据库,使得故障识别技术不再虚无抽象。
Objective To propose a fault identification system method based on fault events distribution probability with sustainable update function for instead of artificial experience.Methods Firstly,a multi-dimensional matrix of fault features was designed,the fault events,fault features and other information were designed as a set of fault knowledge base.Secondly,the principal component analysis method was used to reduce dimension of multi-dimensional feature matrix and output the distribution probability of each fault event,so fault events with the highest distribution probability was used as the recognition result.The closed-loop system was also designed to improve the ability of data processing and calculation.Finally,some of collected indicators were calculated by fault events distribution probability method to obtain the fault identification effect of infant incubator and electrocardiogram monitor.Results The failure of infant incubator was identified by the method of fault events distribution probability.At the same time,manual detection was introduced for verification,the same conclusion was drawn as air duct failure,and recognition rate was 61.2%.The closed-loop system was used to identify the monitor fault based on distribution probability of fault events,which was also consistent with actual results.Conclusion It is demonstrated that the fault identification method of medical equipment based on distribution probability of fault events could transform the experience and knowledge of experts into a set of standard automatic scientific detection system and application fault database,which makes the fault identification technique concrete.
作者
杨瑾
陈晓宇
王子洪
杨睿
余潇华
苟斌
YANG Jin;CHEN Xiao-yu;WANG Zi-hong;YANG Rui;YU Xiao-hua;GOU Bin(Department of Medical Engineering,The First Hospital Affiliated to Army Medical University,Chongqing 400038,China)
出处
《生物医学工程与临床》
2025年第1期103-109,共7页
Biomedical Engineering and Clinical Medicine
基金
重庆市科卫联合医学科研面上项目(2022MSXM060)。
关键词
故障事件
故障特征
分布概率
主成分分析法
fault event
fault characteristics
distribution probability
principal component analysis
