Icing thickness prediction model using BP Neural Network

Affected by microtopography and micro-meteorological phenomena condition, accidents such as lines trip out, electric arc burn, hardware fittings and insulator damage, broken wire stocks, break line and downed towers often happen on overhead transmission lines, which has seriously threatened the safety operation of the power system, and brought the huge losses to the national economy. Therefore, it is of great importance to make further studies on the icing thickness prediction model of transmission lines in order to guarantee the safety, stability and reliable operation for the grid. Aiming at the catastrophic damage caused by the icing transmission lines in grid, the principle of BP Neural Network was described and it was introduced into the prediction model, then according to the icing monitoring data collected from eight transmission lines in Guizhou grid, the structure of the Neural Network was confirmed and consequently the BP Neural Network was established, the relevant program was designed by using MATLAB language, simultaneously the simulation and prediction results were obtained. The result shows that the error between the predicted data and the field data of the icing thickness is less than 5mm, which has successfully demonstrated the icing thickness prediction model using BP Neural Network is feasible and effective.