摘要
输电线路绝缘子由于架设在野外,受天气因素、环境污染等影响,使得绝缘子发生破损,导致输电线路供电中断。为避免输电线路发生故障,最大程度保障电网安全运行,提出基于深度学习的直流输电线路绝缘子破损识别方法。分析航拍图像噪声来源,利用中值滤波算法确定像素中值,经过非线性平滑处理去除噪声。分析常见的绝缘子破损类型、破损表现及原因,建立随机森林决策树。通过深度学习,选出明显的破损特征作为识别依据;构建Alex Net卷积神经网络模型,计算损失函数,确定最佳学习速率;通过学习训练,输出识别结果。实验结果显示,所提方法能够增强图像细节信息,且绝缘子破损识别正确率在0.8以上、收敛速度快。
Due to being installed in the wild,insulators on transmission lines are affected by weather factors,environmental pollution,and other factors,leading to damage and interruption of power supply.To avoid failure and ensure safe operation of the power grid,a method for identifying damage on insulators of DC transmission line was proposed based on deep learning.Firstly,the noise source in aerial images was analyzed,and the median filtering algorithm was used to determine the pixel median values.Then,non-linear smoothing was applied to remove noise.Secondly,common types of damage,damage characteristics,and causes of insulator were analyzed.Meanwhile,a random forest decision tree was established.Using deep learning algorithm,obvious damage features were selected as the basis for recognition.Next,a convolutional neural network model based on Alex Net was constructed,and the loss function was calculated to determine the optimal learning rate.Finally,the recognition results were outputted based on the learning and training.Te experimental results show that the proposed method can enhance image details and achieve an accuracy rate of over O.8 in identifying damage on insulators of DC transmission lines with fast convergence speed.
作者
叶萧然
杜玉红
刘群坡
YE Xiao-ran;DU Yu-hong;LIU Qun-po(Hebi Institute of Engineering and Technology,Henan Polytechnic University,Henan Hebi 458030,China;School of Electrical Engineering and Automation,Henan Polytechnic University,Henan Jiaozuo 454003,China)
出处
《计算机仿真》
2024年第1期251-255,共5页
Computer Simulation
基金
2021年度河南省高等学校重点科研项目(22B520018)
2021年度河南省高等学校重点科研项目(22B470007)。
关键词
深度学习
直流输电线路
绝缘子
破损识别
卷积神经网络
Deep learning
DC transmission line
Insulator
Damage identification
Convolutional neural network
