基于改进EMD和数据分箱的轴承内圈故障特征提取方法

为解决直驱风力发电机主轴后轴承内圈轻微损伤故障诊断问题,针对实际工程中振动信号的复杂特性,提出一种基于改进经验模态分解(empirical mode decomposition, EMD)和数据分箱的特征提取算法。将信号进行改进经验模态分解,得到一系列平稳的本征模函数(intrinsic mode function, IMF)。对分解后的信号提取均值、方差等幅域参数特征,并根据参数有效性选择部分参数组成特征矩阵。选用等宽分箱方法,用箱内数据均值代替箱体数据,将特征矩阵进行平滑处理。经验证,该方法能准确提取实际工程信号中的有效特征,并从特征选择的角度较好解决了分类器代价敏感问题,减少了机器学习模型的过拟合现象。

Feature extraction method of rolling bearing inner ring in wind turbine based on improved EMD and feature box

According to the characteristics of vibration signal of rolling bearing inner ring in direct-driven wind turbine, a new method of fault diagnosis by improved empirical mode decomposition(EMD)and feature box was put forward. The original signal was decomposed by improved EMD to get a finite number of stationary intrinsic mode functions(IMFs). The characteristics of amplitude domain parameters such as mean and variance were extracted, which were turned into feature matrix chose by effectiveness. To perform data smoothing processing, The feature matrix was divided into boxes and replaced by means of data in each box. Examples showed that the feature matrix, which was divided into boxes finally, could effectively extract the fault feature of rolling bearing, and reduce the over fitting of the machine learning model.