变电站遥视系统中智能多模式行为识别的应用  

Application of intelligent multi-mode behavior recognition in remote vision system in substation

作　　者：范智翔[1] 蔡乐才[2] 

机构地区：[1]四川理工学院自动化与电子信息学院,四川自贡643000 [2]四川理工学院计算机学院,四川自贡643000

出　　处：《电力系统通信》2009年第5期47-51,共5页Telecommunications for Electric Power System

基　　金：四川省教育厅科研项目(2007ZL048)

摘　　要：针对变电站遥视系统监控不同环境下各种目标的多种形态和行为特征,为优化其监控能力,从系统解决方法的角度,运用模糊和粗糙集的特征提取方法,采用粗集等价类知识编码和数据融合技术,对分类特征和目标行为方法化简,利用智能反馈控制强化目标特征,形成系统的环路识别。同时,确立不同环境目标的机器自学习模式,建立一个具有普遍适用性、时间空间的高鲁棒性、低复杂性的智能多模式行为识别方法。其理论方法在变电站智能遥视、高级感知接口、动态视频内容检索等应用领域具有指导意义。Based on the multifarious modalities and behaviour characteristics of the targets acquired by remote vision system in substations at different conditions, this paper introduces the application of intelligent multi-mode behaviour recognition in remote vision system in substation. To improve the monitoring ability, it simplifies the classification feature and target behaviour by using feature extraction method of fuzzy and rough set as well as equivalence class knowledge encoding and data fusion technology to form system loop recognition. At the same time, machine self-learning mode in different conditions is established and an intelligent muhi-mode behaviour recognition method with universality, higher robustness and lower complexity is formed. This method has guiding significance and huge economic and social value in many fields such as intelligent remote vision in substations, premier perceiving interfaces and retrieving of dynamic video and so on.

关 键 词：变电站 遥视系统 智能多模式 粗集理论 智能反馈控制 监控终端 

分 类 号：TM764.2[电气工程—电力系统及自动化] TP181[自动化与计算机技术—控制理论与控制工程]