一种自适应报警的轴承性能退化评估新方法

A new Method for Evaluating of the Bearing Performance Degradation Based on Adaptive Alarm

论文首次将线性预测倒谱系数(LPCC)引入到轴承声信号特征提取,并提出了一种自适应报警的轴承声信号性能退化评估新方法:首先使用正常状态下的声信号进行隐马尔可夫模型(Hidden Markov Model,HMM)训练,建立正常HMM模型;然后计算当前评估数据在正常HMM模型条件下的输出概率,当输出概率超出阈值范围时,即认定轴承性能状态处于退化Ⅰ级。在启动报警的同时采集退化Ⅰ级声信号重新训练HMM,使用新的HMM对下一时刻运行的轴承进行性能退化评估,直到依次出现退化Ⅱ级,退化Ⅲ级或更严重的退化状态,实现了自适应性能退化评估功能。轴承加速疲劳寿命实验表明,论文提出的自适应性能退化评估方法能够描述轴承整个性能退化过程,具有良好地应用前景。

LPCC is introduced to the feature extraction of bearing acoustic signal for the first time in this paper,and a new Method for Evaluating of the Bearing Performance Degradation Based on Adaptive Alarm is proposed.First,the normal HMM model is established through training Hidden Markov Model(HMM)with normal sound signals.Then the output probability of the current evaluation data under the normal HMM model condition is calculated.When the output probability exceeds the threshold range,it is found that the bearing performance state is in the degenerate grade I.HMM is retrained with degraded I sound signal at the same time of starting the alarm,then this new HMM is used to evaluate the performance degradation of the bearing running at the next time,until bearing performance occurs the degradation level II,degradation level III or more serious degradation.The adaptive performance degradation evaluation function is achieved.Experiments on bearing accelerated fatigue life show that the proposed adaptive degradation evaluation method can describe the whole performance degradation process of bearings,and it has good application prospects.