首页 | 本学科首页   官方微博 | 高级检索  
     

基于人工智能的电网调度操作智能防误系统建设及实践
作者姓名:蔡新雷  齐颖
作者单位:广东电网有限责任公司电力调度控制中心,广东省农村信用社联合社银信中心
摘    要:本文将人工智能算法引入电网调度业务,结合调度规程和指令规范,通过语音识别平台实时转化调度电话为文本信息,对于识别的文字通过语义理解、深度学习提取关键词,识别和探测业务场景。利用提取的关键信息在电网操作平台基于电网实时状态校核、调度业务场景规则进行校核和防误。通过语音平台对于不规范和不正确的调度指令进行告警和提示。通过运行操作历史大数据不断学习发现规律,建立完善的调度业务知识图谱,不断提高语音识别的准确率和场景探测的准确度,进而实现调度电话业务24小时安监的功能。本系统实现了操作全过程状态、潮流等全链条智能防误管控,可解决电话下令时由于监护不到位、下令不规范、调度指令理解错误等情况发生时,调度误下令、误操作问题。

关 键 词:人工智能  语音识别  大数据  防误  校核
收稿时间:2020/4/8 0:00:00
修稿时间:2020/4/27 0:00:00

Construction and practice of intelligent system for power grid dispatching based on artificial intelligence
Authors:CAI Xinlei and Qi Ying
Affiliation:Guangdong Power grid power dispatch and the control center,Guangdong Rural Credit Union Bank Credit Center
Abstract:In this paper, artificial intelligence algorithms are introduced into the power grid dispatch business. Combined with dispatch procedures and instruction specifications, the dispatch phone is converted into text information through a voice recognition platform in real time. Key words are extracted through semantic understanding and deep learning for the recognized text to identify and detect business scenarios. The extracted key information is used to verify and prevent errors on the grid operation platform based on the grid real-time status verification and scheduling business scenario rules. Alarms and prompts for irregular and incorrect scheduling instructions through the voice platform. Through the operation and operation of historical big data, it constantly learns and discovers laws, establishes a perfect knowledge map of dispatching services, and continuously improves the accuracy of voice recognition and the accuracy of scene detection, thereby realizing the function of 24-hour security supervision of dispatching telephone services. The system realizes full chain intelligent anti-misoperation management and control of the entire process status and current flow, which can solve the problems of dispatching wrong orders and misoperations when the phone is ordered due to inadequate monitoring, irregular orders, and incorrect understanding of scheduling instructions.
Keywords:artificial intelligence  speech recognition  big data  error proofing  check
点击此处可从《》浏览原始摘要信息
点击此处可从《》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号