基于超高压变电站智能巡检的多模态综合分析算法  

Multimodal Comprehensive Analysis Algorithm Based on Intelligent Inspection of Ultra-high Voltage Substations

作　　者：杨利民 李松 钱程 李飞翔 田莉 YANG Limin;LI Song;QIAN Cheng;LI Feixiang;TIAN Li

机构地区：[1]国网新疆电力有限公司超高压分公司,新疆乌鲁木齐830001

出　　处：《今日自动化》2024年第9期173-175,共3页Automation Today

基　　金：国网新疆电力有限公司科技项目(5230CD240009)。

摘　　要：随着智能化进程的推进,超高压变电站智能化巡检也迎来了新的发展需求。文章结合基于正样本图像判别和可见光与红外热成像异常温度检测算法,提出了一种基于多模态综合分析的专家系统算法。该算法与传统目标检测方法不同,其综合变电系统中的多模态信息,输出变电站的异常概率,通过信息论的方法对算法的输出量化来提高最终识别效果。试验结果表明,该算法在超高压变电站智能巡检中能够提高巡检自动化效果,为变电站安全运行提供了一种新的解决方案。With the advancement of social intelligence,the intelligent inspection of ultra-high voltage substations has also ushered in new development needs.This article proposes an expert system algorithm based on multimodal comprehensive analysis,combining positive sample image discrimination and visible light and infrared thermal imaging anomaly temperature detection algorithms.This algorithm is different from traditional object detection methods.It integrates multimodal information in the substation system,outputs the probability of anomalies in the substation,and quantifies the output of the algorithm through information theory to improve the final recognition effect.The experimental results show that this algorithm can improve the automation effect of intelligent inspection in ultrahigh voltage substations,providing a new solution for the safe operation of substations.

关 键 词：超高压变电站 智能巡检 正样本图像判别 多模态综合分析 

分 类 号：TM76[电气工程—电力系统及自动化]