Scholarlink

This paper proposes a multi-stage model for photovoltaic power forecasting designed for regions where direct measurements of global horizontal irradiance (GHI) and tilted irradiance (GTI) are unavailable. Two independent irradiance prediction models were developed using smoothed GHI and GTI data, along with the solar zenith angle (Z). Subsequently, a prediction chain framework was introduced to integrate the forecast results. Experimental results show that the accuracy of the GTI prediction chain (RMSE = 0.84) is comparable to that of models based on raw data (RMSE = 0.80), demonstrating high reliability and application potential. Although the GHI prediction chain's accuracy is slightly lower (RMSE = 0.95 compared to 0.82), it still provides reliable photovoltaic power predictions. The study demonstrates the effectiveness of this method under data-scarce conditions, offering a new pathway for optimizing photovoltaic systems in complex environments and laying the groundwork for future applied research.

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

Photovoltaic Power Forecasting Framework Based on a Multi-stage Irradiance Prediction Chain: Applications for Areas Lacking Direct Measurement Data

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Chengcheng Jiang,
  • Qunzhi Zhu

摘要

This paper proposes a multi-stage model for photovoltaic power forecasting designed for regions where direct measurements of global horizontal irradiance (GHI) and tilted irradiance (GTI) are unavailable. Two independent irradiance prediction models were developed using smoothed GHI and GTI data, along with the solar zenith angle (Z). Subsequently, a prediction chain framework was introduced to integrate the forecast results. Experimental results show that the accuracy of the GTI prediction chain (RMSE = 0.84) is comparable to that of models based on raw data (RMSE = 0.80), demonstrating high reliability and application potential. Although the GHI prediction chain's accuracy is slightly lower (RMSE = 0.95 compared to 0.82), it still provides reliable photovoltaic power predictions. The study demonstrates the effectiveness of this method under data-scarce conditions, offering a new pathway for optimizing photovoltaic systems in complex environments and laying the groundwork for future applied research.