Growth Height Prediction for the Trees under Overhead Lines based on Deep Learning Algorithm

Abstract

The safe operation of transmission overhead lines is often threatened by the fast growing and high growth trees under their line corridor. When the safe distance between the line and the tree is insufficient due to the limited height of transmission network, it is easy to occur tree related fault and tripping. Therefore, in order to effectively prevent the damage to overhead lines caused by the growth of extra high trees, it is necessary to know the growth rule of extra high trees and predict their growth height. In this paper, the deep learning algorithm is used to study the growth rule of extra high trees under overhead transmission lines. Different deep learning and artificial neural network algorithms such as Deep Belief Network, Auto-Encoder and Long-Short-Term-Memory Algorithm are used to predict the tree height, and the validity of these algorithms is verified. Furthermore, these algorithms are combined to verify that the combined algorithm has higher prediction accuracy than the single multilayer perceptron and mathematical statistical model.