人工智能驱动的无人值守变电站巡视系统优化设计  

Artificial Intelligence-Driven Unmanned Substation Inspection System Optimization Design

作　　者：王文浩 WANG Wenhao(State Grid Shanxi Ultra-High Voltage Substation Branch Company,Taiyuan 030031,China)

机构地区：[1]国网山西省电力公司超高压变电分公司,山西太原030031

出　　处：《通信电源技术》2025年第6期227-229,共3页Telecom Power Technology

摘　　要：随着人工智能技术的发展,传统变电站人工巡视模式已难以满足智能电网建设需求。针对无人值守变电站巡视系统存在的识别准确率低、路径规划欠优等问题,提出一种基于深度学习与改进蚁群算法的智能巡视方案。该方案构建多模态数据融合模型提升设备缺陷识别能力,采用改进蚁群算法优化巡视路径规划。实验验证表明,设备缺陷识别准确率提升至95.8%,巡视路径长度较传统方法缩短23.6%。优化后系统显著提高了变电站无人值守巡视效率,为变电站智能化运维提供了新思路。With the development of artificial intelligence technology,traditional manual substation inspection modes can no longer meet the demands of smart grid construction.Addressing the issues of low recognition accuracy and suboptimal path planning in unmanned substation inspection systems,this paper proposes an intelligent inspection solution based on deep learning and improved ant colony algorithm.This solution constructs a multimodal data fusion model to enhance equipment defect recognition capability and employs an improved ant colony algorithm to optimize inspection path planning.Experimental validation shows that equipment defect recognition accuracy increased to 95.8%,and inspection path length shortened by 23.6%compared to traditional methods.The optimized system significantly improved unmanned substation inspection efficiency,providing new insights for intelligent substation operation and maintenance.

关 键 词：人工智能 无人值守 变电巡视 深度学习 路径优化 

分 类 号：TM6[电气工程—电力系统及自动化]