基于EEMD-Treelet和高斯过程的起重机齿轮故障诊断

Fault Diagnosis of Crane Gear Using EEMD-Treelet and Gaussian Process

摘要:

齿轮故障诊断对于起重机安全运行至关重要。本文提出了一种基于集成经验模态分解(Ensemble empirical mode decomposition, EEMD)-Treelet变换和高斯过程(Gaussian Process, GP)的起重机齿轮故障诊断新方法。设计一种细菌觅食算法(Bacterial foraging optimization, BFO)优化高斯过程模型超参数。建立基于集成经验模态分解-希尔伯特变换的多模态时频分布特征提取方法,利用Treelet变换实现齿轮振动信号特征降维。建立基于细菌觅食算法优化高斯过程的齿轮故障模型。实验结果表明, EEMD-Treelet- GP诊断方法不仅可以识别最佳特征向量,而且可以识别故障位置。

Abstract:

Gear fault diagnosis is essential to the safe operation of cranes. This work proposes a new fault diagnosis method of crane gear based on Ensemble empirical mode decomposition (EEMD), Treelet transform, and Gaussian Process (GP). A Bacterial foraging optimization (BFO) algorithm is proposed to optimize the hyperparameters of the GP model. A multi-modal time-frequency distribution feature extraction method is developed based on ensemble empirical mode decomposition and Hilbert transform. Treelet transform is used to reduce the feature dimensionality of gear vibration signal. Finally, the fault location of the gear is identified by the BFO-GP model. Experimental results show that the EEMD-Treelet- GP diagnosis method can not only identify the best feature vector, but also identify the fault location.