The 4th International Symposium on Project Management (ISPM 2016)

137 / 2016-04-28 10:20:42

Icing forecasting of transmission lines with dynamic naive Bayesian classifier for stage judgment

9946,9947,9948,9949

Transmission line icing is a serious threat to the security of power system. Accurate icing prediction is the guidance to ensure the safety of power system. The influence of meteorological factors on ice thickness is different, so the same input-output model makes prediction precision decreasing for the whole icing process. In this paper, a prediction system based on dynamic naive Bayesian classifier for stage judgment is proposed. Firstly, dynamic naive Bayesian classifier is used to classify the line icing stage. Secondly, according to the correlation characteristics between meteorological factors and ice thickness, partial least squares regression and least square support vector machine forecasting models are respectively established for each icing stage. Finally, the prediction accuracy of proposed prediction system and other prediction methods are compared in the case analysis, experiments showed that, the prediction system based on dynamic naive Bayesian classifier for stage judgment can effectively improve the prediction accuracy of line ice thickness.