基于改进Q学习的虚拟电厂参与调峰辅助服务策略

Auxiliary Service Strategy of Virtual Power Plant Participating in Peak Shaving Based on Improved Q-Learning

新型电力系统推进建设中,大规模光伏并网及优先消纳可能会导致电网阻塞而加剧火电机组调峰压力,使其运行工况恶化和成本增加。为此提出虚拟电厂参与系统日前深度调峰辅助服务策略。首先,分析计算电动汽车群组成的虚拟电厂调峰特性和运行费用,并根据火电机组的煤耗和排放数据,计算供电煤耗和建立环保指标;其次,基于运行费用和指标,以火电机组调峰裕度为优化目标,运用模拟退火改进的Q学习求解机组深度调峰电量和分摊费用。算例结果表明,虚拟电厂参与系统调峰可提高调峰灵活性和降低运行费用,缓解火电机组调峰压力。

In the construction of the new power system,the grid connection and preferential consumption of large-scale photovoltaic power may lead to grid congestion and aggravate the peak shaving pressure of thermal power units,which will worsen their operating conditions and increase their costs.Aiming at this problem,the virtual power plant participation system in the day-to-day in-depth peak shaving auxiliary service strategy is proposed.First,the peak shaving characteristics and operating costs of the virtual power plant composed of electric vehicle groups are analyzed and calculated.According to the coal consumption and emission data of thermal power units,the coal consumption for power supply is calculated and the environmental protection indicators are established.Secondly,based on the operating costs and indicators,taking the peak shaving margin of thermal power units as the optimization goal,the simulated annealing improved Q-learning is used to solve the deep peak shaving capacity and cost allocation.The results show that the participation of virtual power plants in system peak shaving can improve the flexibility of peak shaving,reduce operating costs,and relieve the peak shaving pressure of thermal power units.