Transformer Top-Oil Temperature Combination Modeling Based on Thermal Circuit and Support Vector Machine

Increasing accuracy of predicting the transformer top-oil temperature (TOT) and winding hot-spot temperature (HST) is essential to improving the utilization of transformer. This paper presents a combination model to improve TOT prediction accuracy. The main feature of this model is its combination of both model-driven model's and data-driven model's advantages. First, an available thermal circuit is utilized to predict the TOT roughly; second, a data-driven model based on support vector machine (SVM) is established to approximate the thermal circuit prediction error; and finally, the SVM is utilized to correct prediction results of the thermal circuit. The proposed model is tested on a 200-kV A distribution transformer and the obtained results are compared with existing thermal circuit model and data-driven model. The analysis result demonstrates the validity and accuracy of the combination model.