摘要
针对扭力冲击钻滚动轴承易多发故障问题提出了一种改进支持向量机的故障特征提取方法,并结合多维时态关联规则来判断轴承是否出现故障。主要借助抽样算法来形成向量集合,并且在此基础上提升识别效率,找到异常信号,并且对其来源进行判定,通过多维时态关联规则找出异常信号与故障类别之间的关系。通过不平衡转子动力学模型与搭建实验平台试验验证关联规则的准确性和可靠性,再根据各信号的实时状态通过已建立的时态关联规则实时预测下一时间段的信号状态,从而达到实时预测的目的。实验表明,本故障诊断预测方法有效,能够识别和预测滚动轴承的90%的故障。
Aiming at the problem that the torsion impact drill rolling bearing is prone to multiple faults,this paper proposes an improved fault feature extraction method of support vector machine,and combines the multidimensional temporal association rules to judge whether the bearing has faults.The vector set is formed mainly by means of sampling algorithm,and on this basis,the identification efficiency is improved,the abnormal signals are found,and the source is determined.The relationship between the abnormal signals and fault categories is found out through multi-dimensional temporal association rules.The accuracy and reliability of association rules were verified by the unbalanced rotor dynamics model and the experimental platform,and then the signal state of the next period was predicted by the established temporal association rules according to the real-time status ofeach signal,so as to achieve the purpose of real-time prediction.Experimental results show that the proposed fault diagnosis and prediction method is effective and can identify and predict 90%faults of roling bearings.
作者
胡景松
樊军
马冉
HU Jing-song;FAN Jun;MA Ran(School of Mechanical Engineering,Xinjiang University,Xinjiang Urumqi 830049,China)
出处
《机械设计与制造》
北大核心
2023年第9期17-21,共5页
Machinery Design & Manufacture
基金
国家自然科学基金-扭力冲击器双模射流自适应控制最佳匹配机理研究(11462021)。
关键词
SVM
故障特征提取
多维时态关联规则
实时预测
SVM
Fault Feature Extraction
Multidimensional Temporal Association Rules
Real-Time Prediction
