基于威布尔分布的故障预测与诊断方法研究  

Research on fault prediction and diagnosis method based on Weibull distribution

作　　者：杨喆[1] 冯靖祎[1] YANG Zhe;FENG Jing-yi(Department of Medical Engineering and Materials,The First Affiliated Hospital of Zhejiang University School of Medicine,Hangzhou 310003,Zhejiang,China)

机构地区：[1]浙江大学医学院附属第一医院医学工程与物资部,浙江杭州310003

出　　处：《生物医学工程与临床》2025年第2期256-261,共6页Biomedical Engineering and Clinical Medicine

摘　　要：目的 为减少医疗设备故障对医院工作的影响,提高医疗设备使用率和可靠性,降低维修与保养成本,从而提升医院医疗服务质量。方法 实验数据来源于浙江大学医学院附属第一医院的云物资管理系统,其中包含所有医疗设备自安装启用以来所有的维修保养记录,由分管工程师维护和管理。基于MATLAB平台,提出了一种结合威布尔分布的医疗设备故障预测与诊断方法。通过分析某类医疗设备的历史故障数据,建立学习模型,实现故障的预测与诊断。结果 通过计算医疗设备故障数据的中位秩F (ti),绘制了威布尔概率图,并采用KS检验法验证散点是否符合威布尔分布。结果显示,KS统计量为0.130,P值为0.369,均大于显著性水平,因此可以认定该实验数据符合威布尔分布。随后,根据公式计算得出形状参数β为1.543,尺度参数η为237.69,进一步推导出威布尔分布的累计分布函数,并预测未来1年内可能出现的故障数量N及故障发生时间间隔(T2-T1)。结论 该方法符合实际情况,可以进行医疗设备故障诊断,为医疗设备管理人员提供辅助信息和数据分析,帮助识别和排除潜在的故障,可以有针对性地实施预防性维护,减少损失和维修成本,从而实现对医院医疗设备的精细化管理。Objective To reduce the impact of medical equipment failure,improve utilization rate and reliability of medical equipment,reduce cost of maintenance and repair,and improve the quality of medical services in hospitals.Methods The experimental data was drawn from cloud material management system of the First Affiliated Hospital of Zhejiang University School of Medicine,and included maintenance records of all medical equipment since its installation and inception,which maintained and managed by in-charge engineers.Based on MATLAB platform,the fault prediction and diagnosis method of medical equipment combined with Weibull distribution was proposed.By examining the historical failure data of a particular type of medical equipment,the learning model was developed to achieve fault diagnosis and prediction.Results The Weibull probability diagram was drawn by calculating median rank F(ti)of fault data of medical equipment,and scatter points conformance to Weibull distribution was checked using KS test method.The results showed that the KS statistic was 0.130 and P was 0.369,both of which were greater than significance level.Therefore,the experimental data conformed to Weibull distribution and identified.Then,according to the formula,the shape parameterβwas 1.543,scale parameterηwas 237.69,and cumulative distribution function of Weibull distribution was further derived,the number of possible faults N and fault occurrence interval(T2-T1)within 1-year were predicted.Conclusion It is demonstrated that the method is consistent with actual situation,which could diagnose medical equipment faults,provide auxiliary information and data analysis for medical equipment managers,and help identify and eliminate potential faults,implement preventive maintenance in a targeted manner to reduce losses and maintenance costs,so as to achieve fine management of hospital medical equipment.

关 键 词：医疗设备 数据分析 威布尔分布 故障预测与诊断 

分 类 号：R197.32[医药卫生—卫生事业管理]