首页|期刊导航|电力科技与环保|基于数据驱动的厂级负荷分配优化策略研究

基于数据驱动的厂级负荷分配优化策略研究OA

Data driven based strategy of plant-level load dispatch for the coal-fired power plant

中文摘要英文摘要

[目的]在双碳目标推动下,煤电机组正由基荷电源向灵活调节电源转型,长期低负荷运行导致供电煤耗显著升高.如何科学分配厂内多台机组负荷以实现全厂能耗最优,是火电行业亟需解决的关键问题.[方法]以某虚拟电厂1 000 MW、600 MW和300 MW 3类6台机组为研究对象,提出一种数据驱动的厂级负荷优化分配策略.采用灰色关联分析筛选影响供电煤耗率的关键参数,结合XGBoost算法构建煤耗特性预测模型;以全厂供电煤耗率最低为目标,引入基于历史成功的自适应参数差分进化算法,动态求解最优厂级负荷分配方案.[结果]研究表明,本文提出的供电煤耗率预测模型的相对误差在±0.05%之间波动,提出的厂级负荷分配优化策略在不同总负荷需求下均可实现全厂能耗的最小化,节煤潜力随着总负荷呈先增加后降低的变化趋势;当厂级负荷达到2 600 MW时,厂级负荷优化分配的供电煤耗率仅为312.29 g/(kW·h),与传统调度模式相比,节煤潜力最高可达1.92 g/(kW·h).[结论]本文提出的数据驱动厂级负荷分配策略为火电厂在灵活运行工况下挖掘节能潜力提供了低成本、高效率的技术路径,为构建清洁低碳、安全高效的新型电力系统提供了支撑.

[Objective]Driven by the carbon peaking and carbon neutrality goals,coal-fired units are transitioning from baseload to flexible regulation power sources,resulting in prolonged low-load operation and a significant increase in net standard coal consumption.Scientifically allocating loads among multiple units to minimize plant-level energy consumption has become a critical issue for the thermal power industry.[Methods]This study takes six units of three types(1 000,600 and 300 MW)from a virtual power plant as the research objects,a data-driven plant-level load optimization strategy is proposed.Grey relational analysis is employed to screen key parameters affecting the net standard coal consumption rate,and an XGBoost-based coal consumption characteristic prediction model is constructed.With the objective of minimizing the plant-level net standard coal consumption rate,an adaptive parameter differential evolution algorithm based on historical success is introduced to dynamically solve the optimal load dispatch scheme satisfying grid load commands.[Results]The results indicate that the relative error of the proposed coal consumption rate prediction model fluctuates within 0.05%.The plant-level load allocation optimization strategy proposed in this paper achieves minimized overall energy consumption under varying total load demands.The coal-saving potential exhibits a trend of first increasing and then decreasing with the total load.When the plant-level load reaches 2 600 MW,the net standard coal consumption rate under the optimized load dispatch is only 312.29 g/(kW·h).Compared to conventional dispatch modes,it demonstrates a maximum coal-saving potential of 1.92 g/(kW·h),effectively reducing both plant-wide fuel consumption and carbon emissions.[Conclusion]The proposed data-driven plant-level load allocation strategy proposed herein offers a low-cost,high-efficiency technical pathway for thermal power plants to unlock energy-saving potential under flexible operating conditions.This provides support for establishing a new power system that is clean,low-carbon,safe,and efficient.

刁冰;毕志颖;徐婧

中国五洲工程设计集团有限公司,北京 100053太原理工大学电气与动力工程学院,山西 太原 030024太原理工大学电气与动力工程学院,山西 太原 030024

能源科技

厂级负荷分配灰色关联极端梯度提升算法SHADE算法

load dispatchinggrey correlation analysisextreme gradient boostingSHADE algorithm

《电力科技与环保》 2026 (4)

574-581,8

山西省重点研发计划项目(202502090302015)

10.19944/j.eptep.1674-8069.2026.04.006

评论