陀螺电机轴承健康评估隐马尔可夫模型适应性设计  

Adaptive Design of Hidden Markov Model for Gyro Motor Bearing Health Evaluation

作　　者：蔡曜 司玉辉 王玉琢 武展 付雪 阮芳 CAI Yao;SI Yuhui;WANG Yuzhuo;WU Zhan;FU Xue;RUAN Fang(Xi’an Aerospace Precision Mechatronics Institute,Xi’an Shaanxi 710100,China)

机构地区：[1]西安航天精密机电研究所,陕西西安710100

出　　处：《传感技术学报》2023年第9期1417-1425,共9页Chinese Journal of Sensors and Actuators

摘　　要：陀螺电机的轴承一般采用小体积、高精度的角接触球轴承,其健康状况无法通过外观检查、计量等方法检测,故障具有隐蔽性、潜伏性,当故障发生时,将直接影响陀螺的精度及寿命,故需提前对潜在故障进行识别。基于电机定子电流评估的基本原理,选取若干定子电流特征参数用于故障诊断;基于隐马尔可夫模型(HMM)的基本原理,对其进行适应性设计,根据特征参数的特点,提出局部、全局线性归一化方案,并设计适用于多特征参数的模型参数求解算法、轴承健康评估算法。实例验证表明,通过对HMM的适应性设计,健康轴承、故障轴承对应HMM的模型参数正确收敛,健康评估模型成功建立;使用该模型对已知健康、故障的6块电机轴承进行健康评估,可迅速、准确获得评估结果,准确率为100%。The bearings of gyro motor are generally angular contact ball bearings with small volume and high precision,the condition of which can not be detected by appearance inspection and measurement methods,and the faults are hidden and latent.When faults occur,they will direct affect the accuracy and life of gyro,so it is necessary to identify potential faults in advance.Based on the basic principle of stator current evaluation,some characteristic parameters of stator current are selected for fault diagnosis.Based on the basic principle of hidden Markov model(HMM),the adaptive design is carried out.According to the characteristics of the feature parameters,the local linear normalization scheme and the global linear normalization scheme are proposed,and the model parameter solving algorithm and bearing health evaluation algorithm are designed.The example verification shows that through the adaptive design of HMM,the model parameters of HMMs corresponding to healthy bearings and faulty bearings converge correctly,and the health evaluation model is successfully established.When the proposed model is used to evaluate the health of 6 motor bearings with known health and failure,the evaluation results can be obtained quickly and accurately with an accuracy of 100%.

关 键 词：轴承健康评估 隐马尔可夫模型 定子电流评估法 陀螺电机 挠性陀螺 

分 类 号：TH133.3[机械工程—机械制造及自动化]