A prediction model of ice thickness based on grey support vector machine

Abstract

In order to reduce the icing accidents of transmission lines, the prediction of icing thickness on transmission lines will be able to effectively guide the anti-icing work of power grid. In this paper, a short-term prediction model based on grey support vector machine for icing thickness of transmission lines is proposed, and the elimination of dirty data and the method of data preprocessing are analyzed. The accuracy and applicability of the proposed model are verified by the comparison between the model predictions and the measured data based on the predicted maximum ice thickness, it can provide guidance on monitoring icing condition, the early warning and AC/DC melting ice work. The proposed model is compared with support vector machine (SVM) and particle swarm optimization algorithm (PSO) prediction model, and the average error of the proposed model is 0.28mm, and the average absolute error is 4.33%, which is suitable for short-term prediction of icing thickness of transmission line. In the ice area, the application of the prediction model can guide the transmission line ice-resistant work.