Research on Online Compensation of Electronic Transformer Based on Ensemble Learning Stacking Algorithm

The electronic transformer is affected by factors such as running time and environment, and the error performance in the long-term operation process is not stable enough to cause inaccurate energy measurement. This paper proposes an online error compensation method for electronic transformers based on Stacking integration algorithm. Firstly, the mutual information (MI) coefficient is used to reduce the dimension of the original feature set to obtain the optimal correlation feature set. Then, the high correlation features are used as the input of the model, and the K-fold (KF) cross-validation method is used to train each sub-model. Finally, the real-time error of the transformer is predicted and the error of the transformer is calibrated online. The example analysis shows that the prediction loss of the proposed method is less than 5 %, which is applied to the output compensation of the secondary end of the 0.2 stage transformer, so that the measurement accuracy of the transformer is improved to 0.1.