Partial discharge (PD) detection of cable terminals is crucial for the safe operation of the traction power system in trains. However, similar PD signals in complex train‐operating environments cause difficulty to recognise the insulation defects. Therefore, a PD signal image transformation recognition method is proposed for PD detection of cable terminal defects to identify defects in cable terminals with similar PD characteristics accurately. In the proposed method, the raw PD signals are firstly transformed to images via the Gramian angular field (GAF) representation. This can reveal the discriminative characteristics embedded in the original PD signals and subsequently facilitate differentiating the PD sources, which exhibit similar characteristic in the time domain. The obtained GAF representation of PD signals (named as PD GAF images) is extracted from local and global features to train an efficient MobileVIT model, which is then utilised to identify similar types of PD sources in cable terminals. The results show that the proposed method achieves 97.5% recognition accuracy in the field experiment, which is superior to other methods.

High-speed electric multiple units (EMUs) play a pivotal role as the vital arteries of the national economy, essential components of major public infrastructure projects, and the fundamental backbone of a nation’s comprehensive transportation network. Current detection methodologies struggle to achieve effective online monitoring of cable terminal insulation due to challenges in extracting dynamic temporal signal features and insufficient annotated datasets for pattern recognition. A self-supervised evaluation method of insulation state for vehicle cable terminals based on the hypergraph neural network with dynamic features is proposed in this article. Initially, a novel method for constructing a graph structure with spectral differentiation is introduced to characterize the dynamic features and deterioration trends of the insulation state in vehicle cable terminals. Subsequently, to address the general problem of limited labels on-site, the defined hypergraph neural network is embedded to get the self-supervised hypergraph neural network. Furthermore, the fine-tuned hypergraph model is initialized with partial parameters derived from the embedded self-supervised pretrained model. Ultimately, the proposed approach is experimentally verified by vehicle cable terminal HFCT experiments. The proposed method demonstrates robust feature representation and generalization capabilities for assessing the insulation of vehicle cable terminals. It achieves an accuracy of up to 99.01%, providing a reliable framework for intelligent fault diagnosis in high-speed EMU traction power supply systems and industrial predictive maintenance.

It is important to accurately assess the thermal defect state of equipment to ensure its safe operation. A novel thermal defect detection method called the square even‐order derivative method is proposed. First, the derivative functions of temperature were constructed to investigate the main‐side‐peak characteristics of the derivative curve, and the mathematical expressions for the position, depth, intensity of the defects, and parameters of the derivative curve were deduced, which can be used to quantitatively detect the defects. Then, numerical and experimental analyses of an example case of defected epoxy resin plates were carried out to verify that thermal defects can be effectively and accurately detected by the proposed method. Finally, the method was applied to a vehicle cable to detect its thermal defects in another example case of defected vehicle cable, which preliminarily validated the feasibility of the proposed method in practical applications.

Partial discharge detection is considered a crucial technique for evaluating insulation performance and identifying defect types in cable terminals of high-speed electric multiple units (EMUs). In this study, terminal samples exhibiting four typical defects were prepared from high-speed EMUs. A cable discharge testing system, utilizing high-frequency current sensing, was developed to collect discharge signals, and datasets corresponding to these defects were established. This study proposes the use of the convolutional neural network (CNN) for the classification of discharge signals associated with specific defects, comparing this method with two existing neural network (NN)-based classification models that employ the back-propagation NN and the radial basis function NN, respectively. The comparative results demonstrate that the CNN-based model excels in accurately identifying signals from various defect types in the cable terminals of high-speed EMUs, surpassing the two existing NN-based classification models.