Measurement error prediction of CVT based on DenseNet-GRU network

Capacitor voltage transformer (CVT) out-of-tolerance is one of the main factors causing power metering losses. In order to accurately predict CVT errors and avoid losses caused by CVT out-of-tolerance in a timely manner, this paper proposes a CVT error prediction algorithm based on the DenseNet-GRU network. This method uses CVT historical error data and environmental parameters as data samples to establish DenseNet-GRU Error prediction model achieves accurate prediction of CVT errors. In order to verify the effectiveness of the model, experimental comparisons were conducted with error prediction models established by similar algorithms. The results showed that the average absolute percentage error, root mean square error, and coefficient of determination of the prediction results of the DenseNet-GRU model were optimal and had higher prediction performance. The accuracy and robustness provide ideas and support for subsequent CVT error prediction research work.