基于深度强化学习的风光耦合热电联产机组优化调度

王志霄; 田雪沁; 王鹏朝; 陈海盟; 孙立

doi:10.19805/j.cnki.jcspe.2025.240308

欢迎来到能源电力期刊网！

English Version 期刊登录

您当前的位置：

首页 >

文章列表页 >

基于深度强化学习的风光耦合热电联产机组优化调度

综合能源系统 | 更新时间：2026-10-10

- 基于深度强化学习的风光耦合热电联产机组优化调度
- Dispatch Optimization of Combined Heat and Power Generation Units Coupled with Wind and Solar Power Based on Deep Reinforcement Learning
- 动力工程学报 2025年45卷第7期页码：1134-1143
- 作者机构：
  
  1. 东南大学能源与环境学院, 大型发电装备安全运行与智能测控国家工程研究中心,江苏,南京,210018
  2. 国网经济技术研究院有限公司,北京,102209
  3. 国网宁夏电力调控中心,宁夏,银川,750001
- 作者简介：
- 基金信息：
  
  国家电网有限公司总部科技项目(5108-202335059A-1-1-ZN)
- DOI：10.19805/j.cnki.jcspe.2025.240308
  中图分类号：
- 网络首发：2025-07-14，
  
  纸质出版：2025
- 稿件说明：
移动端阅览
王志霄,田雪沁,王鹏朝,陈海盟,孙立. 基于深度强化学习的风光耦合热电联产机组优化调度动力工程学报, 2025, 45(7): 1134-1143 https://doi.

org/10.19805/j.cnki.jcspe.2025.240308
王志霄,田雪沁,王鹏朝,陈海盟,孙立. 基于深度强化学习的风光耦合热电联产机组优化调度动力工程学报, 2025, 45(7): 1134-1143 https://doi. DOI： 10.19805/j.cnki.jcspe.2025.240308.

org/10.19805/j.cnki.jcspe.2025.240308 DOI：

摘要

集中供热管网蓄热在改善热电联产机组供能成本及可再生能源消纳方面具有较大潜力。但由于热电耦合和源荷不确定性

系统的有效运行和调度优化面临挑战。以风光发电耦合热电联产机组系统为研究对象

构建了考虑热电半解耦的能量调度优化模型。针对系统在运行中遇到的热电耦合和源荷不确定性问题

设计了基于马尔可夫决策过程的深度确定性策略梯度(deep deterministic policy gradient

DDPG)调度优化方法

通过训练神经网络以适应随机变化的源荷场景

实现了机组的实时能流调度优化

并在高负荷、常规负荷和低负荷3种典型场景下进行调度仿真。结果表明:基于DDPG的热电半解耦调度方法能够有效调整机组出力

可同时满足电、热负荷

并保证供热管网的安全性与可持续性;相较于基于规则的算法

DDPG算法在调度综合成本和风光能源的有效利用方面效果优越

在3种负荷场景下

DDPG算法有效降低了调度综合成本和弃风弃光率

前者分别降低了2.46%、7.49%和10.55%

后者分别减少了1.98百分比、3.40百分比和4.86百分比。

Abstract

Heat storage in central heating networks has significant potential in reducing the energy supply costs of combined heat and power (CHP) units and improving renewable energy consumption. However

the system faces challenges in effective operation and dispatch optimization due to the coupling of heat and electricity and uncertainties in energy supply and demand. An energy dispatch optimization model considering heat-electricity semi-decoupling was developed on a system coupling CHP units with wind and solar power. A deep deterministic policy gradient (DDPG)-based dispatch optimization method based on Markov decision processes was designed to address the issues of heat-electricity coupling and uncertainties in energy supply and demand during system operation. The real-time energy flow dispatch of the units was optimized by training neural networks to adapt to randomly changing energy supply and demand scenarios. Dispatching simulations were carried out for three typical load scenarios: high

regular

and low load. Results show that the DDPG-based heat-electricity semi-decoupling dispatch method can effectively adjust unit output

meeting both electricity and heat demands

while ensuring the safety and sustainability of the heating network. Compared with rule-based algorithms

the DDPG algorithm shows superiority in reducing overall dispatch costs and effectively utilizing wind and solar energy. In the three load scenarios

the DDPG algorithm reduces overall dispatch costs by 2.46%

7.49%

and 10.55%

respectively

and decreases curtailment rates of wind and solar power by 1.98

3.40

and 4.86 percentage points

respectively.

关键词

Keywords

references

HASSAN Q, VIKTOR P, AL-MUSAWI T J, et al. The renewable energy role in the global energy transformations[J]. Renewable Energy Focus, 2024, 48: 100545.

SAYED E T, OLABI A G, ALAMI A H, et al. Renewable energy and energy storage systems[J]. Energies, 2023, 16(3): 1415.

黄大为, 史博铭, 于娜, 等. 热电联产机组利用热网动态特性提升实时灵活性的自调度策略[J]. 中国电机工程学报, 2024, 44(20): 7983-7994. HUANG Dawei, SHI Boming, YU Na, et al. Self-scheduling strategy to improve the real-time flexibility of CHP unit by utilizing heat network dynamic characteristics[J]. Proceedings of the CSEE, 2024, 44(20): 7983-7994.

刘华军, 吴倩敏. 新时代十年中国绿色发展之路[J]. 中国人口资源与环境, 2024, 34(3): 102-111. LIU Huajun, WU Qianmin. China's path to green development in the decade of the new era[J]. China Population, Resources and Environment, 2024, 34(3): 102-111.

吴茂乾, 乔磊, 宋明曙, 等. 热网蓄热特性及其对机组调峰能力的影响研究[J]. 动力工程学报, 2024, 44(3): 475-485. WU Maoqian, QIAO Lei, SONG Mingshu, et al. Research on heat storage characteristics of heat supply networks and its influence on peak regulation capacity of units[J]. Journal of Chinese Society of Power Engineering, 2024, 44(3): 475-485.

张红昌, 薛小军, 徐钢, 等. 热电联产机组热电解耦改造方案的调峰特性及能耗分析[J]. 动力工程学报, 2023, 43 (10): 1382-1390. ZHANG Hongchang, XUE Xiaojun, XU Gang, et al. Peak shaving characteristics and energy consumption analysis of the thermoelectric-decoupling retrofit scheme for cogeneration units[J]. Journal of Chinese Society of Power Engineering, 2023, 43(10): 1382-1390.

王晋达, 周志刚, 赵加宁, 等. 集中供热管网的蓄热模式与蓄热能力评估[J]. 暖通空调, 2019, 49(1): 46-51, 88. WANG Jinda, ZHOU Zhigang, ZHAO Jianing, et al. Heat storage mode and capacity evaluation of district heating networks[J]. Heating Ventilating & Air Conditioning, 2019, 49(1): 46-51, 88.

赵稚辉. 利用热网及建筑物储热的热电联产机组深度调峰能力分析[J]. 应用能源技术, 2018(10): 34-38. ZHAO Zhihui. Research on capacity of in-depth peak regulation of heat supply unit based on thermal storage of heat supply network and buildings[J]. Applied Energy Technology, 2018(10): 34-38.

曾艾东, 王佳伟, 邹宇航, 等. 考虑供热管网储热的综合能源系统多时间尺度优化调度[J]. 高电压技术, 2023, 49(10): 4192-4202. ZENG Aidong, WANG Jiawei, ZOU Yuhang, et al. Multi-time-scale optimal scheduling of integrated energy system considering heat storage characteristics of heating network[J]. High Voltage Engineering, 2023, 49(10): 4192-4202.

金玲素. 基于模型的城市级蒸汽供热系统运行调度实时优化研究与应用[D]. 杭州: 浙江大学, 2020.

PRATO A P, STROBINO F, BROCCARDO M, et al. Integrated management of cogeneration plants and district heating networks[J]. Applied Energy, 2012, 97: 590-600.

DING Z H, HUANG Y H, YUAN H, et al. Introduction to reinforcement learning[M]//DONG H, DING Z H, ZHANG S H. Deep Reinforcement Learning. Singapore: Springer, 2020: 47-123.

ZHANG G M, ZHANG C, WANG W, et al. Offline reinforcement learning control for electricity and heat coordination in a supercritical CHP unit[J]. Energy, 2023, 266: 126485.

CHEN X, FANG S, LI K J. Reinforcement-learning-based multi-objective differential evolution algorithm for large-scale combined heat and power economic emission dispatch[J]. Energies, 2023, 16(9): 3753.

王耀臣, 王锡, 刘琦, 等. 基于改进白鲸算法的太阳能与生物质互补CCHP系统多目标优化研究[J]. 动力工程学报, 2024, 44(5): 735-744. WANG Yaochen, WANG Xi, LIU Qi, et al. Research on multi-objective optimization of a CCHP system with solar and biomass complementary based on improved beluga whale algorithm[J]. Journal of Chinese Society of Power Engineering, 2024, 44(5): 735-744.

GAO B, ZHOU Y K. In-field experimental study and multivariable analysis on a PEMFC combined heat and power cogeneration for climate-adaptive buildings with Taguchi method[J]. Energy Conversion and Management, 2024, 301: 118003.

KANG Z Z, DUAN R N, ZHENG Z M, et al. Grid aided combined heat and power generation system for rural village in north China plain using improved PSO algorithm[J]. Journal of Cleaner Production, 2024, 435: 140461.

CHAIBAKHSH A, GHAFFARI A. Steam turbine model[J]. Simulation Modelling Practice and Theory, 2008, 16(9): 1145-1162.

曹浩. 蒸汽管网仿真建模及运行特性分析[D]. 北京: 华北电力大学, 2021.

OLABI A G, ABDELKAREEM M A. Renewable energy and climate change[J]. Renewable and Sustainable Energy Reviews, 2022, 158: 112111.

VEERENDRA KUMAR D J, DEVILLE L, RITTER III K A, et al. Performance evaluation of 1.1 MW grid-connected solar photovoltaic power plant in louisiana[J]. Energies, 2022, 15(9): 3420.

MARCHAND J, SHETGAONKAR A, RUEDA TORRES J L, et al. EMT real-time simulation model of a 2 GW offshore renewable energy hub integrating electrolysers[J]. Energies, 2021, 14(24): 8547.

KEYHANI A, GHASEMI-VARNAMKHASTI M, KHANALI M, et al. An assessment of wind energy potential as a power generation source in the capital of Iran, Tehran[J]. Energy, 2010, 35(1): 188-201.

XUE Y R, WU R, LIU J F, et al. Crowd evacuation guidance based on combined action reinforcement learning[J]. Algorithms, 2021, 14(1): 26.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

基于深度策略梯度的综合能源系统优化调度研究

基于强化学习的风电场自适应调频控制方法

基于Leckner模型的氨掺烧高水蒸气浓度烟气对燃煤锅炉传热特性的影响分析

分形特征对漂浮式风力机仿生平台动态响应的影响

基于多通道特征提取与深度交互融合网络的IES多元负荷短期预测