中国与中亚天然气供应链的动态收益研究——基于单智能体强化学习视角
Research on the Dynamic Benefits of Natural Gas Supply Chain between China and Central AsiaBased on the perspective of single-agent reinforcement learning
张丽 1高佳明2
作者信息
- 1. 新疆财经大学会计学院
- 2. 新疆财经大学信息管理学院
- 折叠
摘要
中国与中亚之间的天然气能源合作问题事关中国能源安全和发展需求.所以本文以动态收益为切入点,通过分析中国与中亚地区天然气能源合作现状,创新性地建立基于强化学习理论的天然气供应链模型.根据前人对天然气产业链研究的经济问题结果设置状态变量,结合现实天然气供应数据设置实验参数.实验结果显示,选择最大探索率、增大学习率、减少折现系数和减少断货概率都可以使模型动态收益和收敛速率增加,并且实际天然气供应链问题验证了实验模型的正确性,进而对中国能源合作提出策略建议:(1)增大中国油气企业向外投资建设力度和增加天然气订单,保障未来中国在中亚地区能源产业收益和天然气定价话语权;(2)"一带一路"建设中需要完善能源安全保障机制,应对化解能源供应不足带来的产业冲击,例如:多元化能源进口战略和前期能源基础建设,可以共同应对中国天然气储备薄弱和近年来的能源行业结构性调整问题.本文强化学习模型的成功建立有效促进了强化学习理论在能源产业链方向的发展,具有较高的实用价值.
Abstract
The issue of natural gas energy cooperation between China and Central Asia is related to China's energy security and development needs.This article innovatively focuses on the current situation of natural gas energy cooperation between China and Central Asia and the research model of natural gas supply chain,using dynamic benefits as the entry point to establish a reinforcement learning model.Set state variables based on the results of previous research on economic issues in the natural gas industry chain,and set experimental parameters in combination with real natural gas supply data.The experimental results show that choosing the Epsilon=0.1 strategy,increasing the learning rate,reducing the discount coefficient,and reducing the probability of stockouts can all increase the dynamic benefits and convergence rate of the model,and the correctness of the experimental model has been verified by the real natural gas supply chain problem.Furthermore,strategic suggestions are proposed for China's energy cooperation:1)Increase the investment and construction efforts of Chinese oil and gas enterprises in external investment and natural gas orders,to ensure the future benefits of China's energy industry in Central Asia and its voice in natural gas pricing;2)In the construction of the"the Belt and Road",it is necessary to strengthen the energy security mechanism to address the industrial impact caused by insufficient energy supply.The use of diversified energy import strategy and early energy infrastructure can jointly address China's weak natural gas reserves and the structural adjustment of the energy industry in recent years.The successful establishment of the reinforcement learning model in this article has effectively promoted the development of reinforcement learning theory in the direction of the energy industry chain,and has high practical value.
关键词
中亚/能源合作/供应链/动态收益/强化学习Key words
Central Asia/Energy Cooperation/Supply Chain/Dynamic Earnings/Reinforcement Learning引用本文复制引用
出版年
2024
NETL
NSTL
万方数据