摘要
电力市场中梯级水电站担负着系统调峰、备用等功能,由于梯级水电站受到丰水、枯水季节变化,自身发电效率降低,进而影响其在市场中的交易。针对梯级水电站自身入库径流随机性,汛期需要兼顾防洪、发电、航运等等不确定因素以及市场中的电价波动等等特点,提出把梯级水电站交易方式按照时间先后尺度分为上下两层,每层采用序贯决策中的M arkov决策,把梯级水电站的在电力市场中的交易过程描述为多时间尺度的M arkov动态决策,以构建梯级水电站在市场中交易的序贯决策模型。根据系统负荷需求与梯级自身发电容量限制,采用随机动态规划算法优化各层的交易决策变量、梯级各电站出力过程,以实现梯级水电站期望收益最大目标。实例表明该方法具有很好的实用价值。
The cascade hydropower station bear the tasks of peak regulation and reserve supply in power systems, and its power generation efficiency varies greatly in high and low water periods, which affects its power transactions in the power market. Since the water flow into its reservoir is stochastic, and the uncertainty factors of flood control, power generation, and navigation should be synthetically considered during its flood period, and the electricity price keeps fluctuating in the power market, it is proposed that its power transaction method be classified into 2 levels by time scale, and the Markov of sequential decision-making be adopted for each level. The transaction process can be then described as dynamic multi-time Markov decision-making to build the sequential decision-making model for power transaction of cascade hydropower stations in the power market. Based on the system load demand and the power output of the cascade hydropower stations, the stochastic dynamic programming is adopted to optimize the transaction strategy variables of each level and the power output process of the cascade hydropower stations in order to realize the expected maximum profits. The application case shows that the method is pragmatic.
出处
《华东电力》
北大核心
2006年第7期10-14,共5页
East China Electric Power
基金
国家自然科学基金资助项目(50579022
50539140)
高等学校博士学科点专项科研基金(20050487062)
关键词
电力市场
交易策略
序贯决策
多时间尺度Markov决策过程
梯级水电站
实时电价
power market
transaction strategy
sequential decision-making
multi-time scaled Markov decision-making process
cascade hydropower station
real-time electricity price