摘要
为准确预测电力市场中的短期电价,提出了基于LSTM和XGBoost的组合预测模型。为了验证LSTM-XGBoost模型的有效性,该文先选用法国电力市场2019年1月1日至2020年12月31日的电价数据为训练集训练模型,对2021年1月1日不同模型预测的结果与实际电价值进行对比,得到LSTM-XGBoost以RMSE为0.74的误差率低于BP、LSTM、XGBoost的3.80、1.25、0.88,然后将算法应用到美国PJM电力市场,结果表明本文提出的LSTM-XGBoost组合预测模型MAPE平均值为1.83%,明显低于单一预测模型,也显著低于GRU-XGBoost组合模型,表明并非所有模型单一组合都能有效提高预测精度,该文提出的LSTM-XGBoost组合模型有效提升了短期电价的预测精度,且具有很强的普适性,可应用于电力市场短期电价预测,为市场参与者和监管机构提供有力决策依据。
In order to accurately predict the short-term electricity price in the electricity market,a combined forecasting model based on LSTM and XGBoost is proposed.In order to verify the effectiveness of the LSTM-XGBoost model,this article first selects the electricity price data of the French electricity market from January 1,2019 to December 31,2020 as the training set training model,and predicts the results of different models on January 1st,2021.Compared with the actual electricity value,the error rate of LSTM-XGBoost with an RMSE of 0.74 is lower than 3.80,1.25,0.88 of BP,LSTM,and XGBoost.Then the algorithm is applied to the US PJM power market.The results show that the MAPE average of the LSTM-XGBoost combined prediction model is 1.83%,which is significantly lower than the single prediction model,and also significantly lower than the GRU-XGBoost combined model.It shows that not all models can effectively improve the prediction accuracy by a single combination.The LSTM-XGBoost combined model effectively improves the accuracy of short-term electricity price forecasting,and its strong universality can be applied to short-term electricity price forecasting in the electricity market,thus providing a strong decision-making basis for market participants and regulators.
作者
郑宏
刘立群
ZHENG Hong;LIU Li-qun(School of Electronic and Information Engineering,Taiyuan University of Science and Technology,Taiyuan 030024,China)
出处
《太原科技大学学报》
2023年第2期131-136,共6页
Journal of Taiyuan University of Science and Technology
基金
山西省重点研发计划(201803D121106)
太原科技大学教学改革创新项目(201913)
太原科技大学大学生创新创业训练计划项目(XJ2019020)。