Gearboxes are an important part of the mechanical drives element that provides the several applications like automotive industry, wind turbine industry and power plant industry, etc. The condition monitoring of the gearbox reduces its operational cost, maintenance cost and avoid hazardous losses. The features selected for the health status of the gearbox has important parameter to calculate classification accuracy. In the current study the intrinsic time-scale decomposition (ITD) and tunable Q-factor wavelet transform (TQWT) are used to diagnose the faults in the gear. The ITD method decomposed the input signal into the baseline signal with instantaneous parameters of signal and sequence of the proper rotation components (PRCs). The PRC of higher kurtosis value is the input signal for TQWT. The TQWT is a discrete wavelet transform and decomposed the vibration signals of the gearbox into sub-bands. The feature vector is calculated for each sub-band of the TQWT. The proposed approach is analyzed by the classification accuracy of the feature vector. The recommended method is evaluated using experimental data of 2009 PHM Data of gearbox under various health conditions. The SVM and KNN methods are investigated that the improved classification accuracy with ITD-TQWT model are 97.9% and 96.9% respectively.

中文翻译：

---

基于ITD可调Q因子小波变换的变速箱故障诊断

变速箱是机械驱动元件的重要组成部分，可提供多种应用，例如汽车行业，风力涡轮机行业和电厂行业等。变速箱的状态监控可降低其运行成本，维护成本并避免危险损失。为变速箱的健康状态选择的功能具有重要的参数，可以计算出分类精度。在当前的研究中，固有时间尺度分解（ITD）和可调谐Q因子小波变换（TQWT）用于诊断齿轮故障。ITD方法利用信号的瞬时参数和适当旋转分量（PRC）的序列将输入信号分解为基线信号。峰度值较高的PRC是TQWT的输入信号。TQWT是离散小波变换，将变速箱的振动信号分解为子带。针对TQWT的每个子带计算特征向量。通过特征向量的分类精度对提出的方法进行了分析。建议的方法是使用2009 PHM变速箱数据在各种健康条件下的实验数据进行评估的。通过SVM和KNN方法，ITD-TQWT模型的分类准确率分别提高了97.9％和96.9％。