Voltage Transformer Error State Prediction Method Considering Multi-Dimensional Attributes

To address the issue of insufficient efficiency in feature extraction and fusion for high-dimensional heterogeneous data, this paper proposes a voltage transformer error state prediction method that considers multi-dimensional attributes. The method first decomposes historical secondary-side output data of the capacitive voltage transformer using the STL algorithm, reconstructing multi-dimensional attribute features such as trend, seasonality, stationarity, drift, and correlation. Subsequently, an RBF model based on an adaptive weighted fusion strategy is employed to train the multi-dimensional attribute features, thereby achieving the prediction of the CVT’s operational error state. Experimental validation demonstrates that the proposed method achieves an evaluation accuracy exceeding 97.81%.