基于物理模型的行星轮系特征提取与评估方法

特征提取与评估是损伤检测和故障预测的基础。针对2K-H行星轮系缺齿损伤,建立了行星轮系的损伤模型,通过分析模型的仿真信号,提出了基于主频边带、小波变换和经验模式分解的多种损伤特征,并采用双样本一致性检验方法对所提取损伤特征的分类能力进行了评估,采用含噪声的仿真信号和试验数据对损伤特征进行了验证。验证结果表明,所提取的损伤特征均具有较强的分类能力,其中主频边带特征的分类性能最优。

Feature Extraction and Evaluation Method for Planetary Gear Sets Based on Physical Model

Feature extraction and evaluation is the basis of detection and prognosis. Aiming at tooth breakage in 2K-H planetary gear sets, damage model of planetary gear sets was developed. After the analysis of simulation signals, features based on sidebands around dominant frequency, wavelet transform and EMD were presented and valuated with two samples consistency check. Then the features extracted in the above were verified by simulation signal and test data. Validation results show that the features are promising and the feature based on sideband around dominant frequency is the best one to detect damage for 2K-H planetary gear sets.