高压输电线路多无人机自主协同巡线设计与测试
发布时间:2018-02-12 14:40
本文关键词: 多旋翼无人机 高压线巡线 多无人机协同 机器学习 一致性 出处:《电力系统自动化》2017年10期 论文类型:期刊论文
【摘要】:针对目前巡检高压输电线路的固定翼无人机不能精细化巡查,而多旋翼无人机运动速度慢且需要人工遥控操作导致安全风险较大的问题,提出多架多旋翼无人机自主协同精细化巡线方案。探讨了利用两架部署在导线两侧的无人机通过机间通信协同巡检高压输电杆塔和导线的方法,并采用机器学习技术和多智能体一致性控制算法设计无人机自主飞行控制器。根据设计结果开发一站四机系统验证巡线方案,实验部署两机为一组,两组并行作业。结果表明该系统按预期完成高压输电线路的巡检作业,未发生误报缺陷,相比单个遥控操作的多旋翼无人机巡检方案,多机协同方案无需人工干预飞行,总巡线速度超过单机最高时速,双机协同对单个杆塔的平均巡检时间小于遥控单机巡检时间的一半。
[Abstract]:In view of the problem that the fixed-wing UAV which is currently inspecting HV transmission lines can not be meticulously inspected and the multi-rotor UAV is moving slowly and needs manual remote control operation, it has a high safety risk. This paper presents a scheme of autonomous cooperative and fine line patrol for multiple multi-rotor UAVs, and discusses the method of using two UAVs deployed on both sides of the wire to patrol and inspect the high voltage transmission towers and conductors by means of inter-aircraft communication. The autonomous flight controller of UAV is designed by means of machine learning technology and multi-agent consistency control algorithm. According to the design results, the scheme of one station and four aircraft system is developed to verify the inspection line, and two aircraft are deployed as a group. The results show that the system completes the patrol inspection of HV transmission line as expected, and there is no false alarm defect. Compared with the multi-rotor UAV patrol scheme operated by single remote control, the multi-aircraft cooperative scheme does not need manual intervention flight. The total patrol speed exceeds the maximum speed of the single machine, and the average inspection time for a single tower with two machines is less than half of that of the remote control.
【作者单位】: 电子科技大学航空航天学院;国网四川省电力公司攀枝花供电公司;国网四川省电力公司电力科学研究院;
【基金】:高等学校博士学科点专项科研基金资助项目(20130185110023)~~
【分类号】:TM75
【相似文献】
相关硕士学位论文 前3条
1 孔祥慧;基于机器学习的变电站数据检测技术研究[D];华北电力大学;2016年
2 陈金楷;基于机器学习的锅炉主辅机状态监测研究[D];华中科技大学;2015年
3 王琦;机器学习研究及在风力预测中的应用[D];复旦大学;2012年
,本文编号:1505867
本文链接:https://www.wllwen.com/kejilunwen/dianlidianqilunwen/1505867.html