基于CVaR的风电机组日前申报出力优化模型研究

在我国，风电机组需提前一天提交次日24小时的申报出力信息，由于来风具有明显的不确定性和随机性特征，风电机组次日实际出力与日前申报出力会存在一定的偏差。为解决以上问题，首先采用K-means聚类法对风电机组实际出力数据进行聚类，并生成样本集；其次，基于风电不确定性样本集构建风险中性风电机组日前申报出力决策优化模型；再次，构建基于条件风险值（CVaR）的日前申报出力优化模型；最后，基于实际数据进行了算例分析。算例分析结果表明，基于CVaR的日前申报出力决策优化模型能有效减少风电机组日前申报出力与次日实际出力的偏差，且使风电机组获得风险价值更好的申报出力，从而实现申报利润的风险可控性。

Decision-Making Optimization Model of Day-ahead Bidding for Wind Turbine Based on CVaR

Wind power producers need to offer their 24 hours’ bidding one day in advance In China. Due to the uncertainty and randomness of the wind power，the actual power output often deviate from the bidding. Firstly，K-means clustering method is introduced to cluster actual wind power producers’ output data and to form a sample set. Secondly，based on the probabilistic wind power sample set，the decision-making model of risk-neutral wind power enterprises is established. Then，the decision-making optimization model based on CVaR （Conditional Value at Risk） is proposed. Finally，an example is given based on the actual data. The analysis results of calculation examples show that the decision-making optimization model based on CVaR can effectively reduce the deviation between the daily declared output and the actual output the next day，and the model will enable the wind turbine to obtain better declared output at risk value，so as to realize the risk controllability of declared profit.