长沙理工大学学报（自然科学版）2025，Vol.22Issue(6)：89-100,12.DOI:10.19951/j.cnki.1672-9331.20250728001
面向移动端高并发场景的智能体协同调度与绿色计算优化
Agent collaborative scheduling and green computing optimization for mobile-side high-concurrency scenarios
摘要
Abstract
[Purposes]To address challenges such as sudden regional load surges,sensitive terminal energy consumption,and significant increases in ineffective carbon emissions during high-concurrency scenarios in power mobile applications,this study aims to overcome the limitations of traditional resource scheduling models in dynamic prediction,elastic scaling,and energy efficiency optimization.[Methods]An elastic green optimization framework for agent collaboration was proposed.1)It employed a lightweight spatio-temporal convolutional neural network that integrated power work order priority and disaster factors to achieve 15-minute-level load prediction.2)It established a multi-agent collaborative mechanism featuring"monitoring,scheduling,and energy efficiency",dynamically balancing delay and carbon emission targets through dynamic weighting functions.3)It designed a joint optimization model for end-cloud energy efficiency,integrating a power usage effectiveness(PUE)hierarchical deployment mechanism,a terminal energy consumption-aware scheduling module,and retry suppression middleware,solving the coupling optimization problem of energy consumption sensitivity on mobile sides and energy efficiency of cloud resources.[Findings]The proposed method was validated under million-level concurrency in provincial power grids,reducing the service level agreement collapse rate to 4.7%during disaster periods,lowering the average energy consumption per request for legacy terminals to 0.81 J,and optimizing the data center PUE to 1.22.[Conclusions]This framework offers a novel resource scheduling solution for power mobile services characterized by"elastic responsiveness"and"green operation and low carbon emissions",and it provides effective technical support for achieving the"energy conservation and emission reduction"goals,as well as for realizing the"carbon peaking and carbon neutrality"goals and the digital construction of the new power system.关键词
多智能体协同调度/绿色计算/时空负载预测/数字孪生/端云能效优化/动态权重决策/电力移动应用Key words
multi-agent collaborative scheduling/green computing/spatio-temporal load forecasting/digital twin/end-cloud energy efficiency optimization/dynamic weight decision-making/power mobile application分类
信息技术与安全科学引用本文复制引用
王奕,张金帅,李燕,钟玉滨,张泽国..面向移动端高并发场景的智能体协同调度与绿色计算优化[J].长沙理工大学学报（自然科学版）,2025,22(6):89-100,12.基金项目
国家电网公司科技项目(52680021N00U) Project(52680021N00U)supported by Science and Technology Project of State Grid Corporation of China （52680021N00U）
