面向多时间尺度市场交易的电网企业代理购电决策模型

Decision-making model of electricity purchasing agent service of power grid enter⁃prises for multi-time-scale market transactions

电网企业在多时间尺度电力市场中分配购电量时需要跟踪代理购电负荷和市场价格的变化,评估市场交易价格波动、实际负荷波动和需求响应多重不确定性导致的风险。为此,提出了面向多时间尺度市场中考虑用户需求响应的电网企业代理购电决策模型。首先,分析了电网企业参与的电力市场购电的交易策略,建立电网企业在多时间尺度市场代理购电策略的收益与风险模型。然后,采用场景分析法模拟实际负荷波动对本月购电量的影响,并基于投资组合理论构建电网企业代理购电电量分配优化模型。最后,采用浙江省现货市场试运行实际数据和需求响应演练数据进行仿真分析。结果表明,所提出的电网企业代理购电决策模型能有效降低风险,并能根据电网企业不同风险偏好和负荷预测精度分配其在多个不同时间尺度电力市场的购电比例。

Power grid enterprises need to monitor the purchasing load of the agent and the changes of the market price,and evaluate the risks caused by the multiple uncertainties of the fluctuation of market transaction price,ac⁃tual load fluctuation and demand response when allocating the proportion of power procurement in the multi-time scale electricity market.Based on the portfolio theory,a power procurement surrogate decision-making model of electricity purchasing agent service of decision-making model for power grid enterprises in the multi-time scale elec⁃tricity market considering users’demand response is proposed.Firstly,the transaction strategies of power grid enter⁃prises participating in the electricity market are analyzed,and the profit and risk models of electricity purchasing agent service strategy in the multi-time scale electricity market are established.Then,the scenario analysis method is used to simulate the effect of actual load fluctuation on power procurement surrogate in this month.Based on the portfolio theory,an optimization model for the electricity purchasing agent service of power grid enterprises is con⁃structed.Finally,the actual data of the Zhejiang spot market trial operation and the demand response exercise are used for simulation analysis.The results illustrate that the decision-making model of electricity purchasing agent ser⁃vice of power grid enterprises can effectively reduce the risk,and allocate the proportion of power procurement in the multi-time scale electricity market according to different risk preferences and degrees of load forecasting accuracy.