Hou Shengren
Research
Research
Methods and themes behind the work
This work is driven by a long-term ambition: improving power-market efficiency and helping build a more sustainable energy system through AI, market design, and trading practice.
I pay particular attention to short-term power markets and market-facing energy assets, where rule changes, renewable uncertainty, price formation, bidding behavior, and operational constraints meet most directly.
Research Themes
Core domains
Electricity Markets
Market design, price formation, short-term market behavior, and cross-border trading mechanisms.
Energy Decision Agents
AI systems that connect rules, forecasts, strategies, execution, risk control, and post-event learning into continuous decision loops.
Storage and Flexibility
Battery dispatch, flexibility assets, and market participation under operational and regulatory constraints.
Energy Digitalization
Turning models, workflows, knowledge bases, and risk boundaries into decision systems that real energy organizations can deploy and use.
Methods
What I work with
Modeling Stack
Time-series forecasting, feature engineering, probabilistic modeling, optimization, and scenario analysis.
Decision Stack
Constraint-aware decision models, reinforcement learning, agent workflows, and safe AI for energy systems and market operations.
Selected Public Projects
Open research repositories
Public GitHub repositories connected to my research on reinforcement learning, energy storage dispatch, and energy-system scheduling.
RL-ADN
A high-performance deep reinforcement learning environment for optimal energy storage systems dispatch in active distribution networks.
Open on GitHubOptimal Energy System Scheduling
Source code for the paper on combining mixed-integer programming and deep reinforcement learning for energy management and safe scheduling.
Open on GitHubDRL for Energy Systems Optimal Scheduling
Code for comparing deep reinforcement learning algorithms on energy-system optimal scheduling problems.
Open on GitHubRelated research line
This code layer connects to broader work on storage flexibility, local energy systems, campus-scale energy digitalization, and decision-support methods.
研究
支撑这些工作的研究方法与主题
这部分工作的长期目标，是通过 AI、市场设计与交易实践提升电力市场效率，并帮助构建更可持续的能源系统。
我尤其关注短期电力市场与面向市场运营的能源资产，因为规则变化、可再生能源不确定性、价格形成、报价行为和运行约束，往往在这里最直接地相遇。
研究主题
核心方向
电力市场
关注市场设计、价格形成、短期市场行为以及跨区交易机制。
能源决策 Agent
关注把规则、预测、策略、执行、风控和复盘学习连接成连续闭环的 AI 决策系统。
储能与灵活性
聚焦储能调度、灵活性资源以及受运行与规则约束影响的市场参与逻辑。
能源数字化
把模型、工作流、知识库与风险边界做成能源组织真正能部署、能使用的决策系统。
方法
我常用的方法栈
建模栈
时间序列预测、特征工程、概率建模、优化与情景分析。
决策栈
约束感知决策模型、强化学习、Agent 工作流与面向能源系统和市场运营的安全 AI。
公开项目
公开科研代码
这些公开 GitHub repo 连接了我的强化学习、储能调度和能源系统优化调度研究。
RL-ADN
面向主动配电网中储能优化调度的高性能深度强化学习环境。
打开 GitHubOptimal Energy System Scheduling
混合整数规划与深度强化学习结合的能源系统调度研究代码，连接能源管理和安全调度问题。
打开 GitHubDRL for Energy Systems Optimal Scheduling
用于比较深度强化学习算法在能源系统优化调度问题中表现的公开代码。
打开 GitHub相关研究主线
这些代码与储能灵活性、本地能源系统、校园级能源数字化和决策支持方法等研究主线相互连接。