基于智能加权混合模型的新型电力系统电量预测方法

赵阳; 范文奕; 安佳坤; 赵子珩

赵阳,范文奕,安佳坤,赵子珩.基于智能加权混合模型的新型电力系统电量预测方法[J].电测与仪表,2022,59(12):56-63.
Zhao Yang,Fan Wenyi,An Jiakun,Zhao Ziheng.Monthly electric quantity prediction based on Prophet and KELMweighted hybrid model[J].Electrical Measurement & Instrumentation,2022,59(12):56-63.

基于智能加权混合模型的新型电力系统电量预测方法

Monthly electric quantity prediction based on Prophet and KELMweighted hybrid model

DOI：10.19753/j.issn1001-1390.2022.12.007

中文关键词: 月度电量预测,Prophet算法,核极限学习机,组合预测,相关性分析

英文关键词:Monthly electric quantity prediction, Prophet algorithm, Kernel extreme Learning machine, combination prediction, correlation analysis

基金项目:国网河北省电力有限公司科技项目（5204JY200001）

作者中文名	作者英文名	单位
赵阳	Zhao Yang	国网河北省电力有限公司经济技术研究院
范文奕	Fan Wenyi	国网河北省电力有限公司经济技术研究院
安佳坤	An Jiakun	国网河北省电力有限公司经济技术研究院
赵子珩	Zhao Ziheng	国网河北省电力有限公司经济技术研究院

摘要点击次数: 2222

中文摘要:

随着电力系统的转型升级,新型电力系统的能源供应和消费发生了巨大的转变,因此对电量预测提出了更高的要求。月度电量的准确预测为新型电力系统的优化调度和电力市场的营销计划提供可靠的依据。在深入挖掘历史电量数据、综合分析月度电量特征及相关因素影响的基础上,结合Prophet算法和KELM神经网络算法各自的优势,提出了一种考虑气温、经济水平和节假日的月度电量组合预测方法。首先基于月度电量数据建立了Prophet预测模型,并进行了参数调优过程；其次利用KELM神经网络建立了基于历史电量、气温、GDP、节假日信息的预测模型,并通过参数调优确定最佳预测模型；最后,以加权组合的方式,建立月度电量组合预测模型。通过算例分析,比较了组合算法和其他算法的预测误差和预测效果,表明了本文所提组合模型在预测精度方面有所提升,验证了预测算法的有效性。

英文摘要:

With the transformation and upgrading of the power system, the energy supply and consumption of the new power system have undergone a huge change, so the higher demand for the electricity forecast is put forward. The accurate forecast of monthly electric quantity can provide reliable basis for optimal dispatching of new electric power system and marketing plan of electric power market. On the basis of in-depth mining of historical electric quantity data, comprehensive analysis of monthly electric quantity characteristics and the influence of related factors, combined with the advantages of Prophet algorithm and KELM neural network algorithm, a monthly electric quantity combination prediction method considering temperature, economic level and holidays was proposed. Firstly, the Prophet prediction model was established based on monthly electric quantity data, and the parameters were optimized. Secondly, the prediction model based on historical electric quantity, temperature, GDP and holiday information is established by using KELM neural network, and the optimal prediction model is determined by parameter tuning. Finally, the monthly electric quantity combination prediction model is established by means of weighted combination. By example analysis, the prediction error and prediction effect of the combined algorithm are compared with those of other algorithms. It is shown that the combined model proposed in this paper has improved the prediction accuracy, and the validity of the prediction algorithm is verified.

查看全文查看/发表评论下载PDF阅读器

关闭