Dynamic measurement error detection method of high voltage current transformer based on reinforcement learning

The current measurement error detection method of transformer is to get the measurement error by comparing the measured value with the standard value under the condition of the same standard source. The detection process is cumbersome, and it needs repeated detection for many times to ensure the detection accuracy. In order to improve the measurement accuracy of current transformer, aiming at the limitations of current error detection methods, a dynamic measurement method of high voltage current transformer measurement error based on reinforcement learning is studied. Based on the analysis of measurement error of high voltage current transformer, the transfer coefficient is calculated. Taking the transfer coefficient as the feature, the power value of power grid is predicted by reinforcement learning, and the measurement error is obtained by comparing the measured value of transformer with the predicted value. Through the comparative experiment, it is verified that the designed dynamic detection method of transformer metering error has higher detection accuracy, response speed is improved by about 17.1%, and the performance is better.