Scholarlink

<p>Photovoltaic (PV) generation systems are influenced by numerous factors such as weather conditions, geographical location, and meteorological elements, resulting in fluctuating and intermittent power output sequences. Accurate PV prediction can effectively mitigate the impact of uncertainty in photovoltaic power generation. To guarantee renewable energy utilization and maintain system stability, this paper proposes an ultra-short-term distributed PV prediction and online correction strategy considering multiple environmental factors and seasonal differences. First, a correlation analysis is performed to identify key environmental factors affecting PV power. Then, variational mode decomposition (VMD) is employed to derive signal characteristics from PV time series data, resulting in stable components at varying frequencies. A CNN-BiGRU-Attention model is used for ultra-short-term power prediction of these components. Finally, VMD-SSA-LSSVM is employed for secondary prediction of the error sequence, and the secondary error prediction results are used to correct the primary prediction. Simulation comparisons with other models demonstrate that the proposed prediction model significantly reduces the error in PV forecasting, achieving MAE, RMSE, and MAPE values of 0.013, 0.020, and 0.136, respectively, and performs well across different seasons.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Ultra-Short-Term Power Prediction and Online Correction Strategy for Distributed Photovoltaic Systems Considering Multiple Environmental Factors and Seasonal Differences

  • 期刊论文

  • 发表日期：

    2025年8月6日

查看全文
  • Ming Yang,
  • Qingze Pan,
  • Haipeng Chen

摘要

Photovoltaic (PV) generation systems are influenced by numerous factors such as weather conditions, geographical location, and meteorological elements, resulting in fluctuating and intermittent power output sequences. Accurate PV prediction can effectively mitigate the impact of uncertainty in photovoltaic power generation. To guarantee renewable energy utilization and maintain system stability, this paper proposes an ultra-short-term distributed PV prediction and online correction strategy considering multiple environmental factors and seasonal differences. First, a correlation analysis is performed to identify key environmental factors affecting PV power. Then, variational mode decomposition (VMD) is employed to derive signal characteristics from PV time series data, resulting in stable components at varying frequencies. A CNN-BiGRU-Attention model is used for ultra-short-term power prediction of these components. Finally, VMD-SSA-LSSVM is employed for secondary prediction of the error sequence, and the secondary error prediction results are used to correct the primary prediction. Simulation comparisons with other models demonstrate that the proposed prediction model significantly reduces the error in PV forecasting, achieving MAE, RMSE, and MAPE values of 0.013, 0.020, and 0.136, respectively, and performs well across different seasons.