智能电网中分布式电源的净负荷预测方法研究  

Research on Net Load Forecasting Method of Distributed Generation in Smart Grid

作　　者：尹兆磊 白明辉[1] 袁绍军[1] YIN Zhao-lei;BAI Ming-hui;YUAN Shao-jun(State Grid Jibei Electric Power Co.,Ltd.,Chengde Power Supply Company,Chengde 067000 China)

机构地区：[1]国网冀北电力有限公司承德供电公司,河北承德067000

出　　处：《自动化技术与应用》2023年第12期50-54,66,共6页Techniques of Automation and Applications

摘　　要：在真实的含风光发电系统的智能电网园区场景中,风电特征、光电特征以及电源风险特征相互影响,共同维持了园区供电系统的运行。针对现有方法预测性能低下的问题,本文首先针对一个含风力与光伏发电系统的智能电网园区的分布式电源的场景,研究其优化配置方法,考虑智能电网故障下风、光等分布式电源接入园区时带来的运行风险,其次考虑智能电网园区运行风险评估的指标,以及风、光等分布式电源的出力特性,提出一种考虑运行风险和风、光发电出力特性的分布式电源特征分析方法;最后提出了一种基于深度学习的针对净负荷实时预测模型,并验证了联合预测模型在智能电网中的有效性。In a real scenario of a smart grid park containing wind and photovoltaic power generation systems,wind power characteristics,photovoltaic characteristics and power supply risk characteristics interact with each other to maintain the operation of the park power supply system.To address the problem of low prediction performance of existing methods,this paper firstly investigates the optimal allocation method for a distributed power supply scenario in a smart grid park containing wind and photovoltaic pow-er generation systems,considers the operational risk brought by wind and light and other distributed power supplies connected to the park under smart grid failure,and secondly considers the indexes for operational risk assessment of the smart grid park and the output characteristics of wind and light and other distributed power supplies.Finally,a deep learning-based real-time predic-tion model for net load is proposed and the effectiveness of the joint prediction model in the smart grid is verified.

关 键 词：分布式电源 神经网络 特征表达 光电预测 风电预测 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]