| 引用本文: | 吐松江·卡日,吴 现,马小晶,等.基于地基云图数据多维特征融合的光伏功率预测算法[J].电力系统保护与控制,2025,53(11):84-94.[点击复制] |
| TUSONGJIANG Kari,WU Xian,MA Xiaojing,et al.Photovoltaic power prediction algorithm based on multidimensional features fusion of ground-based cloud images[J].Power System Protection and Control,2025,53(11):84-94[点击复制] |
|
| 摘要: |
| 针对传统光伏功率预测算法无法获取准确云层状态信息和预测精度低等问题，提出一种基于地基云图与双流数据融合的光伏功率预测算法。首先，利用地基云图提供的精确云层状态信息，结合稠密光流法获取相邻帧图像间的时空特征与细节变化特征。其次，结合卷积神经网络(convolutional neural network, CNN)在特征提取上的优势和残差网络在模型学习中抑制信息丢失上的优势，提升预测模型对光伏功率与图像数据间长期映射关系的学习能力。此外，引入注意力机制弥补模型训练过程中关键信息利用不充分的缺陷。实验结果表明，地基云图与光流数据的加入为多云天气提供了更多时空特征。与基准模型相比，其晴天与多云情况下均方根误差(root mean squared error, RMSE)指标和平均绝对误差(mean absolute error, MAE)指标分别降低了15.50%、11.65%、4.05%与5.15%，有助于充分利用云层运动状况来实现准确可靠的光伏电站输出功率预测，提升光伏电站调度工作的及时性与准确性。 |
| 关键词: 深度学习 功率预测 地基云图 注意力机制 稠密光流算法 |
| DOI：10.19783/j.cnki.pspc.241221 |
| 投稿时间：2024-09-09修订日期：2025-01-15 |
| 基金项目:国家自然科学基金项目资助(52067021)；新疆维吾尔自治区自然科学基金面上项目资助(2022D01C35)；新疆维吾尔自治区优秀青年科技人才培养项目资助(2019Q012) |
|
| Photovoltaic power prediction algorithm based on multidimensional features fusion of ground-based cloud images |
| TUSONGJIANG Kari1,WU Xian2,MA Xiaojing1,LEI Kesong3,YU Kaifeng1,SI Weizhuang1 |
| (1. School of Electrical Engineering, Xinjiang University, Urumqi 830049, China; 2. Lishui Power Supply Company,
State Grid Zhejiang Electric Power Co., Ltd., Lishui 323000, China; 3. Changji Power Supply Company,
State Grid Xinjiang Electric Power Co., Ltd., Changji 831100, China) |
| Abstract: |
| To address the limitations of traditional photovoltaic (PV) power prediction algorithms, particularly their inability to accurately capture cloud conditions and their low prediction accuracy, a PV power prediction algorithm based on the fusion of ground-based cloud images and dual-stream data is proposed. First, accurate cloud condition information from ground-based cloud images is utilized, and dense optical flow is employed to extract spatiotemporal and detail change features between adjacent image frames. Then, the advantages of convolutional neural network in feature extraction and residual network in suppressing information loss in model learning are combined to improve the learning ability of the prediction model on the long-term mapping relationship between PV power output and image data. In addition, an attention mechanism is introduced to compensate for the underutilization of critical information during model training. Experimental results indicate that integrating ground-based cloud images and optical flow data offers more spatiotemporal features under cloudy weather conditions. Compared with benchmark models, the proposed method reduces the root mean square error (RMSE) and the mean absolute error (MAE) by 15.50% and 11.65% under sunny conditions, and by 4.05% and 5.15% under cloudy conditions, respectively. This contributes to accurate and reliable forecasting of PV power output by effectively utilizing cloud motion information, thereby improving the timeliness and accuracy of scheduling operations in PV power stations. |
| Key words: deep learning power prediction ground-based cloud mapping attention mechanism dense optical flow algorithm |
