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%.
出处
《中国安全科学学报》
CAS
CSCD
北大核心
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)
