HHT和SVM在机械安全评估与预测中的应用研究  

Research on application of HHT and SVM in safety assessment and prediction for mechanical equipment

作　　者：宣金泉[1] 王晓红[1] 陆大伟[1] 王立志[1] 

机构地区：[1]北京航空航天大学可靠性与系统工程学院,北京100191

出　　处：《中国安全科学学报》2017年第2期58-63,共6页China Safety Science Journal

基　　金：中央高校基本科研业务费专项资金(2014ZC51031);航空科学基金资助(2015ZD51044)

摘　　要：为提高大型复杂机械设备运行的安全性和可靠性,在监测机械设备振动状态的基础上,采用希尔伯特-黄变换(HHT)技术处理信号,将获得的振动频域能量值作为机械设备性能退化的特征量;进而采用网格搜索法(GS)和交叉验证法(CV),优化支持向量机模型(SVM)参数,以提高退化特征量预测精度;并据此建立一种状态空间划分法,用以评估并预测机械设备安全状态。最后,用所建立的方法评估并预测无刷直流电机振动状态和相应的安全状态,预测结果的相对误差仅为1.17%。In this paper,the vibration condition monitoring technique is adopted for mechanical equipment,based on which the HHT method is utilized to processvibration signals; the vibration frequency-domain energy value obtained is taken as the characteristic quantity to represent the performance degradation of mechanical equipment. Then the Grid Search（ GS） and Cross Validation（ CV） methods are used to optimize the parameters of SVM,so as to improve the prediction accuracy of degradation characteristic quantity. Therefore,a state space division method is developed to assess and predict the safety status of mechanical equipment. Finally,the method developed by the authors was used for assessing and predicting the vibration state and the corresponding safety status of brushless direct current motors. The results show that the prediction error is only 1.17%.

关 键 词：机械设备 振动监测 希尔伯特-黄变换(HHT) 支持向量机(SVM) 安全评估与预测 

分 类 号：X913.4[环境科学与工程—安全科学] TP206.3[自动化与计算机技术—检测技术与自动化装置]