Application of a SVR model to predict lightning impulse flashover voltages of parallel gaps for insulator strings

Abstract

Discharge voltage prediction of engineering gap configurations is a great challenge for external insulation design in high‐voltage engineering. This paper presents an air gap discharge voltage prediction model based on the support vector regression (SVR). A series of features are defined on the shortest interelectrode path of an air gap, and they are extracted from the electric field calculation results using the finite element method. These electric field features are considered as the input parameters of the SVR‐based model, and the output is the air gap discharge voltage. The genetic algorithm is used for parameter optimization of the SVR model. Trained by a few experimental data of the Institute of Electrical and Electronics Engineers standard rod–rod gaps, this model is applied to predict the standard lightning impulse flashover voltages of parallel gaps for insulator strings under different gap configurations. The prediction results coincide well with the experimental data collected from previous publications. The mean absolute percentage error of the eight test samples is within 4.07%, which is acceptable for engineering applications. The SVR‐based model may be helpful to guide the structure design and insulation coordination of parallel gaps and insulator strings, thus reducing the required full‐scale tests. © 2019 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.