摘要
为了提高短期负荷预测的精度,利用长短期记忆(Long Short-Term Memory,LSTM)与seq2seq(sequence to sequence)模型预测短期电力负荷。根据电力负荷数据的组成结构和产生原理,收集历史负荷数据,通过缺失补全、归一化等步骤,完成初始收集数据的预处理。构建LSTM与seq2seq模型,利用该模型提取历史电力负荷数据特征,推导出电力负荷数据的变化规律。综合考虑了各因素对电网的影响,得到了电网的短期负荷预测结果。实验结果证明,与传统预测方法相比,在工作日和休息日中,优化设计预测方法的平均误差分别降低了5.64 kW·h和3.53 kW·h,提高了电力负荷预测精度。
In order to improve the accuracy of short⁃term load forecasting,it is proposed to use Long Short⁃Term Memory(LSTM)and seq2seq(sequence to sequence)models to predict short⁃term power loads.According to the composition structure and generation principle of power load data,collected historical load data,and completed the preprocessing of initially collected data through steps such as missing completion and normalization.The LSTM and seq2seq models were constructed,and the features of historical power load data were extracted by using this model,and the variation law of power load data was deduced.The influence of various factors on the power grid is comprehensively considered,and the short⁃term load forecasting results of the power grid are obtained.The experimental results show that compared with the traditional forecasting method,the average error of the optimal design forecasting method is reduced by 5.64 kW·h and 3.53 kW·h respectively on working days and rest days,and the power load forecasting accuracy is improved.
作者
李建芳
纪鑫
张海峰
赵晓龙
陈润东
LI Jianfang;JI Xin;ZHANG Haifeng;ZHAO Xiaolong;CHEN Rundong(Big Data Center of State Grid Corporation of China,Beijing 100052,China;Beijing Sgitg Accenture Information Technology Co.,Ltd.,Beijing 100052,China)
出处
《电子设计工程》
2023年第24期150-153,158,共5页
Electronic Design Engineering
基金
国网大数据中心科技项目(52999019000800K0000000)。