基于云管边端协同的抽水蓄能电站视频监控智能体的研究与应用

Research and application of pumped-storage hydroelectricity video monitoring agent based on cloud tube edge collaboration

摘要

随着新一代人工智能技术的发展，机器学习和深度学习框架已成为开发和推进人工智能应用的重要工具。在充分考虑端侧设备算力资源局限性的前提下，提出基于云管边端协同的抽水蓄能电站视频监控智能体，解决传统依赖云数据中心进行密集计算的问题。通过算法模型分层部署和云管边端协同机制，极大程度上解决了抽水蓄能电站视频监控智能体在算力、网络、存储等方面的问题，降低远距离广域网传输大量数据产生的高时延和高耗能，优化智能体边缘计算模式，提高人工智能技术应用计算复杂度高的能力，提升抽水蓄能电站视频监控智能体实时应用水平。

Abstract

With the development of the new generation of artificial intelligence technology, machine learning and deep learning frameworks have become important tools for developing and promoting artificial intelligence applications. On the premise of fully considering the limitations of computing resources of end devices, a video monitoring intelligent agent for pumped storage power stations based on cloud management edge end collaboration was proposed to solve the problem of traditional reliance on cloud data centers for intensive computing. Through the hierarchical deployment of algorithm models and the cloud management side to side collaboration mechanism, the problems of the video surveillance agent in the pumped storage power plant in computing power, network, storage, etc. are solved to a great extent, the high latency and high energy consumption caused by the transmission of large amounts of data over a long distance WAN are reduced, the agent edge computing mode is optimized, and the ability to apply AI technology with high computational complexity is improved, Improve the real-time application level of video monitoring intelligent agents in pumped storage power stations.

关键词

智能物联网 / 边缘计算 / 马尔可夫决策过程 / 任务推理 / 卸载决策 / 智能视频监控

Key words

intelligent internet of things / edge computing / markov decision process / task reasoning / offloading decision / intelligent video surveillance

Author summay

王保根(1969—), 男, 高级工程师, 硕士, 主要从事电力系统及其自动化。E-mail: [e-mail]

基金资助

国网新源控股有限公司科技项目“国网新源公司人工智能平台研究与应用”(SGXYKJ-2021-006)

许继集团科技项目“基于人工智能的水(发)电安全生产应用开发”(2022G059)