基于文本智能解析变换器的值守机器人控制方法  

Research on Semantic Perception and Intelligent Parsing Method for Distribution Operation and Maintenance Task

作　　者：韩天星 王宝 张伟 冯磊 张潇 梁庆华[1] HAN Tianxing;WANG Bao;ZHANG Wei;FENG Lei;ZHANG Xiao;LIANG Qinghua(School of Mechanical Engineering,Shanghai Jiao Tong University,Shanghai 200240,China;State Grid Jiangsu Electric Power Co.,Xuzhou Branch,Xuzhou Jiangsu 22100,China)

机构地区：[1]上海交通大学机械与动力工程学院,上海200240 [2]国网江苏省电力有限公司徐州供电公司,江苏徐州221000

出　　处：《机械设计与研究》2022年第2期1-6,17,共7页Machine Design And Research

基　　金：国家电网“配电站智能云机器人研究及其应用验证”项目(5700-202018266A)。

摘　　要：利用智能机器人开展配电站无人化值守作业需要对云平台系统下发的传统的运维作业操作票进行翻译,将文档内容转换为机器人提前约定的可识别的操作指令。针对这一问题,设计将运维文本转换为机器语言的变换器。变换器对语料进行分词预处理,基于doc2vec进行词句向量化,并结合K-means++算法进行语料聚类,将聚类的结果通过LCS(Longest Common Subsequence)算法以及seq2seq模型处理,最后基于历史语料库构建得到操作规则库及翻译模型。以江苏省某市的操作票运维文本进行实验,结论表明变换器对文本聚类的结果满足实际情况且通过LCS算法可以得到一个通用的匹配规则,使用变换器构建的翻译框架可以将作业指令准确变换为机器人可以识别的语言,进而控制机器人完成配电运维任务。Using intelligent robots to carry out unmanned duty operations in distribution stations requires identification of traditional operation and maintenance operation tickets issued by the cloud platform system,semantic recognition and translation,and conversion of document content into operation instructions that can be recognized by the robot.To solve this problem,a set of intelligent semantic parsing algorithm and model for the operation and maintenance text is proposed.Firstly,word segmentation pretreatment was carried out on the corpus.Word and sentence vectorization was carried out based on DOC2VEC,and corpus clustering was carried out with K-means++algorithm.The results of clustering were processed by Longest Common Subsequence(LCS)algorithm and Seq2Seq model.Finally,the operation rule base and translation model were constructed based on the historical corpus.The experiment was carried out with the operating ticket operation maintenance text of a city in Jiangsu Province.The experimental results show that the clustering results meet the actual situation and the LCS algorithm can obtain a general matching rule,and the translation framework constructed by the SEQ2SEQ model can identify and parse the operation instructions accurately.

关 键 词：操作票运维文本 doc2vec 聚类算法 LCS算法 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]