基于位置补偿系数距离估计的滚动轴承特征损伤敏感性评估算法研究

Damage sensitivity evaluation algorithm for rolling bearing features based on PCCDET

针对采用距离估计方法对干扰其识别精度的因素考虑不足而影响其评估性能的问题,以及多指标间存在的不相关或冗余干扰滚动轴承损伤程度信息的有效获取问题,提出了一种基于位置补偿系数距离估计的特征评估方法,该方法在传统距离估计技术的基础上引入中值方法以提高算法鲁棒性,提出采用"位置系数"以筛选出能较好抑制不同状态间重合度的特征,并从特征样本对待识别状态的聚合性、可分性以及位置分布关系等多个方面的表现综合考察各特征的敏感性。随后,利用自制滚动轴承疲劳试验台不同损伤阶段的声发射数据开展了算法的有效性验证,通过与其他距离评估方法的对比可知,该方法能更好的指出滚动轴承的损伤敏感特征,筛选出的敏感特征集具有更高且更稳定的损伤识别精度。

Aiming at the problem of not sufficiently considering identification precision’s interference factors affecting distance estimation’s evaluation performance,and the problem of damage level information of rolling bearing not able to be acquired effectively due to irrelevance or redundancy among multiple indicators,a feature evaluation method based on position compensation coefficient distance estimation technique(PCCDET)was proposed.Based on the traditional distance estimation technique,the median approach was introduced to improve the algorithm’s robustness.The position coefficient was used to screen out features to better suppress the overlap ratio among different states.Furthermore,the sensitivity of each feature was comprehensively investigated according to feature samples’polymeriability,divisibility and position distribution relation.Finally,the acoustic emission data in different damage stages of rolling bearings tested on a self-made rolling bearing fatigue test platform were used to verify the effectiveness of the proposed algorithm.Through comparing it with the other distance estimation techniques,it was shown that the proposed method can better indicate rolling bearings’damage sensitive features,and the screened sensitive feature set has higher and more stable identification accuracy.