基于深度强化学习的智能地址库信息分析方法  

Information Analysis Method of Intelligent Address Library Basedon Deep Reinforcement Learning

作　　者：杜文勇 牛逸明 常利建 Du Wenyong;Niu Yiming;Chang Lijian(SGCC Customer Service Center,Tianjin 300309,China)

机构地区：[1]国家电网有限公司客户服务中心,天津300309

出　　处：《电气自动化》2024年第3期97-99,103,共4页Electrical Automation

基　　金：国家电网有限公司客户服务中心电力营销业务2021年95598管理子系统设计开发实施项目(SGKF0000XXXX2100109)。

摘　　要：针对地址库内部的地址无法及时更新、系统无法正确识别等问题,采用耦合隐马尔可夫模型对语言接收部分进行优化,利用自然语言处理算法设计一个智能化地址库。在系统空闲时,通过将概念树与自身数据库进行对比、迭代以满足深度强化学习的检测。试验结果表明,采用耦合隐马尔可夫模型的语言模块检测正确率在90%以上,为地址库信息分析提供了一份可行方案。In response to the problems of delayed updates of addresses within the address library and incorrect recognition by the system,a coupled hidden Markov model was used to optimize the language reception part,and an intelligent address library was designed using natural language processing algorithms.When the system was idle,the detection of deep reinforcement learning was achieved by comparing and iterating the concept tree with its own database.The experimental results show that the language module detection accuracy using a coupled hidden Markov model is above 90%,thus providing a feasible solution for information analysis of address libraries.

关 键 词：智能地址库 耦合隐马尔可夫模型 自然语言处理 语言识别 深度强化学习 

分 类 号：TN914[电子电信—通信与信息系统]