极端天气下基于时间卷积神经网络的新能源出力评估策略  

An evaluation strategy for renewable energy output under extreme weather conditions using a temporal convolutional neural network

作　　者：杨龙 秦建翔 杨波 高海洋 李金东 YANG Long;QIN Jianxiang;YANG Bo;GAO Haiyang;LI Jindong(Dispatch Control Center,State Grid Ningxia Electric Power Co.,Ltd.,Yinchuan Ningxia 750001,China)

机构地区：[1]国网宁夏电力有限公司调度控制中心,宁夏银川750001

出　　处：《宁夏电力》2024年第3期14-20,共7页Ningxia Electric Power

基　　金：国网宁夏电力有限公司科技项目(5229NX220027)。

摘　　要：新能源高渗透率背景下,极端天气造成的出力骤降对电力系统安全运行构成严重挑战,准确评估极端天气下的新能源出力深度是制定电力系统运行方式的重要依据。本文提出一种基于改进的深度时间卷积神经网络(deep temporal convolutional network,DeepTCN)的新能源出力评估方法。通过设计具有动态输入权重的时间卷积神经网络架构,准确量化估计了极端天气对新能源出力的影响,从而为制定极端天气下的电力系统运行方式提供风险程度信息。基于实际电力系统新能源出力数据集的算例结果表明,与常规的时序预测方法相比,所提方法的标准化均方根误差、对称平均绝对百分比误差和平均绝对标度误差3个指标分别最高可提升1.2,0.1和0.22。所提方法能够实现对极端天气下新能源出力的更精准评估。Against the backdrop of high penetration rates of renewable energy,the abrupt reduction in output caused by extreme weather conditions poses a significant challenge to the safe operation of the power system.Therefore,accurately assessing the depth of renewable energy output under extreme weather conditions is crucial for formulating operational strategies for the power system.This paper proposes a new method to evaluate renewable energy output using an improved deep temporal convolutional network(DeepTCN)approach.By designing a temporal convolutional neural network(CNN)architecture with dynamically weighted inputs,this method precisely quantifies and estimates the impact of extreme weather on renewable energy output,providing risk assessment information for formulating the operation mode of power systems under extreme weather conditions.Calculation results on the actual renewable energy output dataset of the power system show that,compared with conventional time series prediction methods,the proposed method can improve the normalized root mean square error,symmetric mean absolute percentage error,and mean absolute scaled error indicators by 1.2,0.1,and 0.22,respectively.Therefore,this approach enables a more accurate evaluation of renewable energy output under extreme weather conditions.

关 键 词：极端天气 时间卷积神经网络 非参数估计 新能源出力 

分 类 号：TM732[电气工程—电力系统及自动化]