摘要
为提高故障预测与健康管理PHM模型预警的准确率以有效支撑动车组现场处置工作,在系统或模型报出诊断结果后,模型研发或运用单位等通常需要投入大量人力和精力对预警结果进行人工核验才能更为准确、有效地指导现场工作,效率较低,影响生产组织。为此,研究提出采用K最近邻和卷积神经网络等两种监督机器学习算法对PHM预警结果进行自动核验,并选取某型动车组牵引系统温度类PHM预警模型对算法进行了测试和验证。结果表明,该算法检验的准确率等指标较为理想,可以在一定程度上代替人工核验,提高模型结果的可用性。
In order to improve the accuracy of PHM(Prognostics and Health Management)model early warning and effectively support the on-site disposal of EMU,after the diagnosis results are reported by the system or model,the model research and development or application units usually need to invest a lot of manpower and energy to manually verify the early warning results to guide the on-site work more accurately and effectively,which is inefficient and affects the production organization.Therefore,two supervised machine learning algorithms are used to verify the PHM early warning results,and the method is studied and verified by selecting a typical PHM temperature model of EMU traction system.The results show that the accuracy of the algorithm is ideal,which can replace manual verification to a certain extent.
作者
董光磊
刘冰
李时偕
徐小明
杨伟君
杨宁
陆航
付昱飞
刘典
DONG Guanglei;LIU Bing;LI Shixie;XU Xiaoming;YANG Weijun;YANG Ning;LU Hang;FU Yufei;LIU Dian(Locomotive&Car Research Institute,China Academy of Railway Sciences Corporation Limited,Beijing 100081,China;Beijing Zongheng Electro-Mechanical Technology Co.,Ltd.,Beijing 100094,China;State Key Laboratory for Traction and Control of EMU and Locomotives.Beijing 100081,China)
出处
《铁道机车车辆》
北大核心
2024年第3期72-76,共5页
Railway Locomotive & Car
基金
国家自然科学基金高铁联合基金(U1934204)
中国铁道科学研究院集团有限公司重点课题(2021YJ285)。
关键词
动车组
故障预测与健康管理
机器学习
神经网络
EMU
Prognostics and Health Management(PHM)
machine learning
neural network
