基于深度强化学习的智能体在智慧消防中的应用研究  

Application of Agent based on Deep Reinforcement Learning in Intelligent Fire Fighting

作　　者：牛浩玉 汤文兵[1] 田锦 NIU Hao-yu;TANG Wen-bing;TIAN Jin(School of Computer Science and Engineering, Anhui University of Science and Technology, Huainan Anhui 232001, China;Engineering School of Networks and Telecommunications, Jinling Institute of Technology, Nanjing Jiangsu 211169, China)

机构地区：[1]安徽理工大学计算机科学与工程学院,安徽淮南232001 [2]金陵科技学院网络与通信工程学院,江苏南京211169

出　　处：《通信技术》2019年第10期2567-2572,共6页Communications Technology

基　　金：江苏省高等学校自然科学研究面上项目(No.17KJB510020)~~

摘　　要：消防直接关系到人民的生命财产安全。针对在火灾发生时因救援环境复杂导致被困人员搜救困难的问题,提出了一种基于深度强化学习的智能体在消防场景中的目标识别和路径规划算法。通过将强化学习算法与卷积神经网络相融合,赋予智能体一定的自主判断、规划分析和目标识别能力。以公共环境中的移动智能体开发应用为背景,针对消防中存在的问题,对智能体在火灾发生前期实现可靠的路径规划和目标识别问题进行深入的应用研究。Fire protection is directly related to the safety of people's lives and property. Aiming at the difficulty of searching and rescuing trapped people when a fire occurs due to the complex rescue environment, an agent target recognition and path planning algorithm based on deep reinforcement learning are proposed in the fire scene. By combining the reinforcement learning algorithm with the convolutional neural network, the agent is endowed with certain ability of autonomous judgment, planning analysis and target recognition. Based on the development and application of mobile agents in the public environment and aiming at the problem existing in fire fighting, this paper makes an in-depth application study on how to realize reliable path planning and target recognition by agents in the early stage of fire occurrence.

关 键 词：智慧消防 深度强化学习 智能体 路径规划 目标识别 

分 类 号：TP399[自动化与计算机技术—计算机应用技术]