Research on adaptive state prediction method for the metering error of capacitor voltage transformer

Abstract

A capacitor voltage transformer (CVT) is widely used in high voltage power systems because of its good insulation performance. However, the structure of CVT is more complex and the stability of its metering error is poor, which easily causes the loss of power metering. The conventional evaluation method for the metering error of CVT is to compare and calibrate with a standard transformer under off-line condition in a fixed period. Because of the long evaluation period, it is impossible to accurately predict the state change of CVT metering error, which is of more significance practically. To solve this problem, this paper proposes an adaptive state prediction method: Analyze the measurement data of CVT using principal component analysis method under the constraint of electrical physical relationship, the metering of CVT is mapped to residual and score (CRS) statistic. For this way, the self-evaluation of CVT metering error in real-time is realized without a standard transformer to get the high frequency time series of error data. According to the measurement data of CVT in process, the CRS statistics are batch processed adaptively, and the prediction model of CRS statistics is established based on the time series analysis. Experiments show that the method can accurately predict the state change of CVT metering error, and the prediction error is better than 15%. It is helpful to promote the development of CVT metering error detection into on-demand detection.