Condition Evaluation Method of Overhead Transmission Line Based on Random Forest Algorithm

Carrying on accurate condition evaluation for overhead transmission lines in power grids can decrease failure rate and enhance power supply performance of power grid effectively. A random forest algorithm based condition evaluation method for overhead transmission lines was proposed. Firstly, the factors influencing the condition of overhead transmission lines were analyzed, and the state variables as well as those under special conditions selected from eight kinds of components in transmission lines were selected to construct a state variable system in which 79 state variables for overhead transmission lines were contained. Secondly, based on the random forest algorithm a multi sub-decision tree combinatorial classification model for condition evaluation of overhead transmission lines was proposed. And then, both numbers of sub-decision trees and random characteristic variables were optimized to improve the classification performance and the efficiency of random forest algorithm. Finally, an example analysis based on actual data from the 110 kV overhead transmission line in a certain city was carried out. Simulation results show that using the proposed method, an accurate condition evaluation for overhead transmission line can be obtained, thus it might provide decision support for overhead transmission lines in regional power grid.