摘要
在复杂的现场环境中,识别模型必须具有实时数据处理能力、对不同状态的准确识别能力和良好的抗噪声能力。针对上述问题,提出一种基于混合域特征和VMD-LSTM的钢轨表面状态识别方法,用于识别钢轨表面的5种状态,即:道岔、接缝、损伤、道岔和损伤的混合状态以及正常状态。该方法主要包括3个阶段,即:混合域特征提取、特征选择和状态识别。通过仍处使用期的列车现场试验采集的振动加速度数据进行验证,测试准确率达到98.26%。实验结果表明,该方法能准确识别钢轨表面状态,可用于实际现场。
For complex on-site environments,recognition models must have real-time data processing capabilities,accurate recognition of different states,and good noise resistance.To address the aboved issues,a rail surface state recognition method based on mixed domain features and VMD-LSTM is proposed to identify five states of the rail surface,which are the mixed state of turnout,joint,damage,turnout and damage and normal state.This method mainly includes three stages,namely mixed domain feature extraction,feature selection and state recognition.The vibration acceleration data collected through on-site testing of inservice trains is verified,and the testing accuracy reaches 98.26%.The experimental results show that this method can accurately identify the surface state of steel rails and can be used in practical fields.
作者
胡康康
崔桂艳
HU Kangkang;CUI Guiyan(School of Materials and Chemistry,University of Shanghai for Science and Technology,Shanghai 200093,China;Shanghai Institute of Microsystem and Information Technology,Chinese Academy of Sciences,Shanghai 200050,China;School of Urban Rail Transportation,Shanghai University of Engineering Science,Shanghai 201620,China)
出处
《智能计算机与应用》
2024年第5期130-135,共6页
Intelligent Computer and Applications
基金
中国科学院关键技术人才项目(E07YRA1)。
