基于深度强化学习的火电机组制粉系统自启停智能决策
Intelligent decision-making of start-up and shutdown for coal milling system in thermal power plants based on deep reinforcement learning
蔡佳辰 1李军 1高明 2高林 1高耀岿 1昌鹏1
作者信息
- 1. 西安热工研究院有限公司,陕西 西安 710054
- 2. 陕西延长石油富县发电有限公司,陕西 延安 727502
- 折叠
摘要
针对目前人工决策基础上的制粉系统一键启停技术存在决策主观经验性强、操盘劳动强度大、节能优化潜力难以发掘等问题,提出了一种综合考虑制粉系统能耗与机组负荷跟踪性能的制粉系统启停决策评价模型,以安全引入网调负荷计划指令信号作为输入,研究了基于深度强化学习的制粉系统自启停智能决策方法,开发了制粉系统自启停决策闭环控制系统.研究结果通过仿真验证,并已在某超超临界 1 000 MW机组常用磨煤机上成功应用,节能降耗效果显著.研究结果可为火电机组少人、无人化运行技术提供有效借鉴.
Abstract
A comprehensive evaluation model for the start-up and shutdown decision-making of the milling system,taking into account the energy consumption and tracking performance of the unit load,has been proposed to address issues such as subjective decision-making based on manual experience,high labor intensity in operation,and difficulty in exploring energy-saving optimization potential.This model safely incorporates the grid load scheduling command signal as input.Furthermore,a milling system start-stop intelligent decision-making method based on deep reinforcement learning has been studied,and a closed-loop control system for the automatic start-stop of the milling system has been developed.The research results have been verified through simulation and successfully applied to a commonly used coal milling system in a certain ultra-supercritical 1 000 MW unit,achieving energy savings.The findings of this study can provide effective reference for the development of unmanned or minimally manned operation techniques for thermal power units.
关键词
深度强化学习/制粉系统/自启停控制/智能决策Key words
deep reinforcement learning/milling system/autonomous start-up and shutdown control/intelligent decision-making引用本文复制引用
基金项目
国家重点研发计划项目(2022YFB4100700)
陕西省重点研发计划项目(2023-YBGY-274)
出版年
2024
国家科技期刊平台
NSTL
万方数据