基于人工智能技术的配网机器人巡检线路自动化识别系统  

Automated Identification System for Distribution Network Robot Inspection Lines Based on Artificial Intelligence Technology

作　　者：郭俊杰[1] GUO Junjie(Zhangjiakou Vocational and Technical College,Zhangjiakou,Hebei 075000,China)

机构地区：[1]张家口职业技术学院,河北张家口075000

出　　处：《自动化与仪器仪表》2025年第1期319-322,共4页Automation & Instrumentation

摘　　要：配网线路暴露在室外,环境复杂多变,为准确识别巡检线路,协助机器人避障巡检,研究基于人工智能技术的0.4 kV配网机器人巡检线路自动化识别系统。在0.4 kV配网机器人身上安装2个用于采集巡检线路图像的双目摄像机,采集机器人巡检线路图像;使用基于自适应同态滤波的巡检线路图像增强方法,抑制巡检线路图像低频成分、放大高频成分,增强图像细节信息;增强后巡检线路图像输入基于改进卷积神经网络的障碍物识别模型,以LeNet-5网络为基础结构,添加批量归一化层,优化网络学习效率后,提取增强后巡检线路图像特征,由Softmax分类器分类输出巡检线路图像特征属于障碍物的概率,完成0.4 kV配网机器人巡检线路识别。实验结果显示:所提算法在0.4 kV配网机器人巡检线路识别问题中,显著提高障碍物识别精度,机器人与障碍物碰撞次数为0次。The distribution network lines are exposed outdoors,and the operating environment is complex and variable.In order to accurately identify the inspection lines and assist robots in obstacle avoidance inspection,a 0.4 kV distribution network robot inspection line automatic recognition system based on artificial intelligence technology is studied.Install two binocular cameras on the 0.4 kV distribution network robot to capture images of inspection lines,and capture images of the robot’s inspection lines;Using an adaptive homomorphic filtering based inspection line image enhancement method to suppress low-frequency components of inspection line images,amplify high-frequency components,and enhance image detail information;The enhanced inspection line image input is based on an improved convolutional neural network obstacle recognition model.The LeNet-5 network is used as the basic structure,and batch normalization layers are added to optimize the network learning efficiency.The enhanced inspection line image features are extracted,and the Softmax classifier is used to classify and output the probability that the inspection line image features belong to obstacles,completing the 0.4 kV distribution network robot inspection line recognition.The experimental results show that the proposed algorithm significantly improves the accuracy of obstacle recognition in the 0.4 kV distribution network robot inspection line recognition problem,with zero collisions between the robot and obstacles.

关 键 词：0.4 kV配网 机器人 巡检线路 自适应同态滤波 人工智能技术 

分 类 号：TN964[电子电信—信号与信息处理]