| 引用本文: | 沈仲恺,宗星辰,万灿,郑舒,赵景涛,鞠平.基于光伏功率概率预测的新能源配电系统节点电压不确定性量化[J].电力自动化设备,2026,46(1):31-39 |
| SHEN Zhongkai,ZONG Xingchen,WAN Can,ZHENG Shu,ZHAO Jingtao,JU Ping.Photovoltaic power probability forecasting based uncertainty quantification of node voltage in distribution system with renewable energy[J].Electric Power Automation Equipment,2026,46(1):31-39 |
|
| 摘要: |
| 针对分布式新能源的大规模接入导致的配电系统电压安全挑战，提出一种基于光伏功率概率预测的新能源配电系统节点电压不确定性量化方法。基于Bootstrap方法与双向长短期记忆网络模型，对光伏功率进行概率预测，将总预测误差分为预测模型误差及数据噪声误差。在光伏预测结果的基础上进一步构建配电系统电压-功率灵敏度矩阵，建立从光伏功率不确定性到节点电压不确定性的线性映射关系。基于配电网电压-功率的物理模型耦合分析，得到节点电压的期望值及波动特征，实现节点电压短期预测的不确定性量化。基于IEEE 33配电系统的算例分析验证了所提方法相对于传统方法具有更高的预测精度。 |
| 关键词: 新能源 配电系统 电压灵敏度 概率预测 不确定性 深度学习 |
| DOI：10.16081/j.epae.202509015 |
| 分类号:TM732 |
| 基金项目:国家电网有限公司科技项目(5400-202219417A-2-0-ZN) |
|
| Photovoltaic power probability forecasting based uncertainty quantification of node voltage in distribution system with renewable energy |
|
SHEN Zhongkai1,2, ZONG Xingchen1, WAN Can1, ZHENG Shu3, ZHAO Jingtao3, JU Ping1
|
|
1.College of Electrical Engineering, Zhejiang University, Hangzhou 310027, China;2.Polytechnic Institute, Zhejiang University, Hangzhou 310015, China;3.NARI Technology Co.,Ltd.,Nanjing 211106, China
|
| Abstract: |
| Addressing voltage security challenge in distribution system caused by large-scale integration of distributed renewable energy, a node voltage uncertainty quantification method for the distribution system with renewable energy is proposed based on photovoltaic power probability forecasting. Probabilistic forecasting of photovoltaic power is conducted using the Bootstrap method and bidirectional long short-term memory network model, decomposing total prediction error into model prediction error and data noise error. Based on photovoltaic forecasting results, the voltage-power sensitivity matrix of the distribution system is constructed to establish a linear mapping relationship from photovoltaic power uncertainty to node voltage uncertainty. Through physical model coupling analysis between distribution network voltage and power, the expected value and fluctuation characteristics of node voltage are derived, achieving quantification of short-term node voltage forecasting uncertainty. Case studies of the IEEE 33-node distribution system validate superior forecasting accuracy of the proposed method compared to conventional approaches. |
| Key words: renewable energy distribution system voltage sensitivity probabilistic forecasting uncertainty deep learning |
