模型和数据混合驱动的双边协商电力市场合约价格预测方法  

Hybrid Model-driven and Data-driven Approach to Price Forecasting in Bilateral Contract Electricity Markets

作　　者：李亚鹏 韩旭 于旭光 程春田[2] 刘本希[2] 蔡华祥 LI Yapeng;HAN Xu;YU Xuguang;CHENG Chuntian;LIU Benxi;CAI Huaxiang(Faculty of Electronic Information&Electrical Engineering,Dalian University of Technology,Dalian 116024,China;Institute of Hydropower&Hydroinformatics,Dalian University of Technology,Dalian 116024,China;Kunming Power Exchange Center Company Limited,Kunming 650200,China)

机构地区：[1]大连理工大学电子信息与电气工程学部,辽宁省大连市116024 [2]大连理工大学水电与水信息研究所,辽宁省大连市116024 [3]昆明电力交易中心有限责任公司,云南省昆明市650200

出　　处：《电力系统自动化》2022年第18期179-189,共11页Automation of Electric Power Systems

基　　金：国家自然科学基金资助项目(52039002)。

摘　　要：对双边协商电力市场合约价格的合理预测和预判可以给双边市场运营和监管提供重要支撑。传统电价预测方法主要针对可形成稳定时间序列的节点、区域或系统电价,对于具有“一人多单、一单一价”的高度异质性特点的双边协商市场则难以直接适用。中国当前双边协商市场数据类型单一、数据规模小的现状进一步加大了预测难度。为此,提出从微观行为到宏观价格的解决思路,给出了基于电力经济学模型和实证数据混合驱动的预测方案。首先,结合电力市场特征,提出可解释双边市场二元协商机理的公理化经济学模型,将市场主体的电能价值估计值与均衡价格联系起来;然后,基于数据回归的思想建立多主体联合优化模型,逆向求解各主体的价值估计值;最后,结合市场主体的价值最优估计值和所提出的经济学模型,对未知合约成交价格进行预测。算例结果表明,所提方法可适应当前数据环境,不仅预测精度高、速度快、鲁棒性好,还可以揭示双边协商市场的微观经济学规律,具有较强的解释力。Reasonable forecasting and prejudging of contract prices in the bilateral contract electricity market can provide important support for market operation and regulation.Traditional electricity price forecasting methods are mainly used for nodal,regional,or system electricity prices that can form a stable time series.But these methods are not directly applicable to bilateral contract markets as they have highly heterogeneous characteristics,i.e.,each participant has many contracts,and each contract includes a specific price.The current status of China’s bilateral contract market with single data type and small data size further increases the difficulty of forecasting.A solution idea that maps micro-behavior to macro-price is proposed.And a price forecasting approach that is driven by the hybrid of an electricity economics model and empirical data is subsequently given.Firstly,considering the characteristics of electricity markets,an axiomatic economics model is proposed which can interpret the two-player negotiation mechanism and link estimation value of electricity valuations of market participants to the equilibrium price.Secondly,a multiparticipant joint optimization model is established based on data regression ideas to estimate the valuation of each participant by inverse optimization.Finally,the contract prices are forecasted by combining the optimal estimation value of the valuations of market participants and the proposed economics model.Case studies indicate that the proposed approach can adapt to the data limitations in the current context,and it is highly accurate,fast,and robust.Meanwhile,it can also reveal the microeconomic laws of bilateral contract markets,i.e.,it owns a strong interpretation.

关 键 词：电力市场 双边协商 电价预测 模型驱动 数据驱动 电能价值 

分 类 号：F426.61[经济管理—产业经济]