机构地区:[1]湖南机电职业技术学院电气工程学院,湖南长沙410151 [2]中国空气动力研究与发展中心设备设计与测试技术研究所,四川绵阳621000
出 处:《铁道科学与工程学报》2022年第10期3096-3106,共11页Journal of Railway Science and Engineering
基 金:国家“十三五”重点资助项目(2016YFB1200600);国家自然科学基金资助项目(61801479);湖南省教育厅课题(20C0703)。
摘 要:悬浮系统的异常指在中低速磁浮列车的实际运行中,悬浮系统工作状态与期望状态不匹配,且系统没有发生故障。准确预测悬浮系统的异常有助于合理地分配有限的监控资源,提前安排预防性维护计划,减少计划外维修成本,降低事故率。虽然根据《中低速磁浮交通车辆悬浮控制系统技术条件CJ/T 458—2014》可获得基于悬浮间隙的经验阈值,但在工程应用中,由于悬浮系统受到额定悬浮间隙不唯一和外界扰动的影响,该方法的检测结果常会产生漏报问题。因此,考虑到悬浮系统发生异常时,悬浮系统的电流、加速度和间隙会发生一定的变化且相互之间的相关性也会发生变化,利用悬浮系统的电流、加速度和间隙等多维监测数据,提出一种基于海林格距离和相关系数的中低速悬浮系统异常检测方法。考虑到系统发生异常时,系统的部分数据间的相关性会发生一定的变化,可通过相关系数来检测数据之间的相关性变化。考虑到系统发生异常时,部分数据会发生一定的变化,可通过海林格距离来检测数据的变化。将相关系数和海林格距离的结果融合用于实现悬浮系统的异常检测。运营线数据验证了该方法受到额定悬浮间隙不唯一和外界扰动的影响较小,且与其他的方法相比,该方法能获得更高的检测率和更低的误报率。The abnormality of a maglev system refers to that in the actual operation of medium and low-speed maglev train,the working state of the suspension system does not match the expected state,and the system does not fail.Accurately predicting the abnormality of the suspension system is helpful to reasonably allocate limited monitoring resources,arrange preventive maintenance plans in advance,reduce unscheduled maintenance costs and reduce the accident rate.Although the empirical threshold based on the suspension gap can be obtained according to the technical conditions for the suspension control system of medium and low-speed maglev traffic vehicles CJ/T 458-2014,in engineering applications,the detection results of this method often produce the problem of false alarm because the suspension system is affected by the non-uniqueness of rated suspension gap and external disturbance.Therefore,considering that when the suspension system is abnormal,the current,acceleration and gap of the suspension system will change to some extent,and the correlation between them will also change,this paper proposed an anomaly detection method of medium and low-speed suspension system based on Hellinger distance and correlation coefficient by using the multidimensional monitoring data of current,acceleration and gap of the suspension system.Firstly,considering that when the system was abnormal,the correlation between some data of the system will change to some extent,and the correlation change between data could be detected by a correlation coefficient,and then considering that when the system was abnormal,some data will change to some extent,and the change of data could be detected by Hellinger distance.Finally,the results of correlation coefficient and Hellinger distance was fused to realize the anomaly detection of the suspension system.The operation line data verify that this method is less affected by the non-uniqueness of rated suspension clearance and external disturbance,and compared with other methods,this method can obtain a hig
分 类 号:TP29[自动化与计算机技术—检测技术与自动化装置]
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...