Online Error Evaluation of CVT Based on Memory and Prediction Enhanced Temporal Convolutional Autoencoder

Capacitor voltage transformers (CVTs) are critical components in high-voltage energy metering, and their measurement accuracy directly impacts the fairness of energy billing. However, environmental factors can cause CVTs to degrade over time, leading to measurement errors and potential system instability. This paper proposes an online evaluation method for CVTs using the Memory and Prediction Enhanced Temporal Convolutional Autoencoder (MPE- TCA). The model incorporates memory and prediction modules to enhance anomaly detection and improve data reconstruction accuracy. Experimental results demonstrate that the model outperforms other models in terms of accuracy, precision, and recall, effectively capturing normal data patterns and detecting anomalies. The proposed method provides a reliable solution for real-time CVT monitoring in modern power systems.