In this study, an online fault diagnosis (FD) algorithm is developed to detect stator’s turn-to-turn faults (TTFs) in direct torque control (DTC)-driven induction motors (IMs) for the electric vehicle’s (EV) powertrains. The developed FD algorithm is based on the application of the discrete wavelet transformation (DWT) technique on the stator currents. Being a time–frequency domain-based algorithm, DWT can deal with nonstationary signals. The fault-sensitive decomposition levels were selected using Ostu’s-thresholding formulation to detect the stator’s faults at their embryonic stage. To determine the faulty phase, the concept of differential discrete wavelet energy (DDWE) was introduced. A finite-element (FE) cosimulation platform and an experimental setup were built to examine the proposed FD algorithm. The results depicted the robustness of the FD method. A comparative analysis is presented between the developed DDWE based FD and the conventional motor current signature analysis (MCSA)-based FD. The experimental results proved that developed FD is faster and more accurate than the MCSA-based FD one. Moreover, the presented FD routine was examined under load step-change and intermittent TTFs, and its diagnosing effectiveness was verified.

中文翻译：

---

通过基于Ostu的DDWE检测用于电动汽车的感应电动机矢量驱动器中的TTF

在这项研究中，开发了一种在线故障诊断（FD）算法，以检测电动汽车（EV）动力总成的直接转矩控制（DTC）驱动的感应电动机（IM）中的定子匝间故障（TTF）。所开发的FD算法基于离散小波变换（DWT）技术在定子电流上的应用。作为一种基于时频域的算法，DWT可以处理非平稳信号。使用Ostu阈值公式选择对故障敏感的分解级别，以检测定子萌芽阶段的故障。为了确定故障相位，引入了差分离散小波能量（DDWE）的概念。建立了有限元（FE）协同仿真平台和实验装置，以检验所提出的FD算法。结果描述了FD方法的鲁棒性。在已开发的基于DDWE的FD与基于常规电机电流签名分析（MCSA）的FD之间进行了比较分析。实验结果证明，开发的FD比基于MCSA的FD更快，更准确。此外，在负载阶跃变化和间歇TTF下检查了所提出的FD例程，并验证了其诊断有效性。