With the rapid application of electric vehicles (EVs) in recent years, there are an increasing number of novel electrical appliances and devices incorporated in the existing power grid, e.g., charging piles and energy storage devices [3]. These new components inevitably bring new features and challenges to the conventional power grid. Accordingly, the mechanical features and fault types of OLTCs are also varied. Therefore, it is necessary to design an effective and effective method ofmechanical feature extraction of OLTCs. The designed method aims to identify the hidden troubles in the process of OLTC operation in smart grid with EVs as early as possible. Such an early-warning mechanism is of great significance to ensure the reliable operation of transformers [4].