A physics-information-enabled self-updating method to monitor steady-state error of capacitor voltage transformers.

Understanding long-term monitoring steady-state errors of capacitor voltage transformers (CVTs) is critical for acquiring accurate and continuous voltage signals from the power grid. However, existing methods cannot maintain good performance over the long term because they cannot update the initial model while primary voltages fluctuate nor maintain stability while steady-state error deteriorates. To remedy this problem, this paper proposes a physics-information-enabled self-updating method. Differences in fluctuations between primary voltages and steady-state error are illustrated based on the physical structure of CVTs. This information is then used to select samples that only contain fluctuations from primary voltages and update the model. At the same time, the mapping relationship between statistics and errors is derived, which makes the evaluation of error possible. Simulations on data from real substations show that its recognition accuracy reaches 97%, and the evaluation deviations of ratio error and phase displacement are within ± 0. 085 % and ± 2. 5 ′ . • Detect CVT with abnormal long-term measurement in real-time without power outages. • Theoretical analysis of error variation characteristics caused by the CVD in the CVT. • Construct an index that describes the degree of change in complex periodic data. • Give a physics-information-enabled self-updating method to update monitoring models. [ABSTRACT FROM AUTHOR]