Scholarlink

In the face of mounting environmental concerns, the utilization of renewable energy sources such as photovoltaics (PV) is critical for lowering greenhouse gas emissions and reducing the environmental impact of energy production. Traditional analytical modeling methods for predicting the performance of PV systems often fall short in accuracy and efficiency, especially when applied to high-power plants. To address this limitation, this research employs artificial intelligence control-based surrogate modeling utilizing a neural network (ANN), specifically a multilayer perceptron (MLP), to improve system performance prediction. With 8000 experimental samples, the ANN model is evaluated in MATLAB software through various performance indices, including mean squared error (MSE), root mean squared error (RMSE), mean absolute deviation (MAD), and mean absolute percentage error (MAPE). The regression coefficient (R2) values demonstrate the model’s accuracy, with R2 = 0.9685 for the training phase, R2 = 0.9461 for the validation phase, and R2 = 0.9643 for overall model performance. In addition, large language models (LLMs) are emerging as transformative tools for renewable energy research. These models can complement ANN-based surrogate modeling by enabling efficient processing of vast datasets, generating actionable insights from technical reports, and optimizing workflows in urban energy planning. By integrating LLMs into renewable energy systems, this study lays the groundwork for a more comprehensive approach to enhancing the annual energy performance of grid-connected high-power PV plants, paving the way for more sustainable urban energy solutions.

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

Harnessing Artificial Neural Networks and Large Language Models for Enhanced Urban Energy Planning: Improving Annual Performance of Grid-Connected High-Power Photovoltaic Plants

  • 章节

  • 发表日期：

    2025年

查看全文
  • Kabache Sabah,
  • Bouregba Hicham

摘要

In the face of mounting environmental concerns, the utilization of renewable energy sources such as photovoltaics (PV) is critical for lowering greenhouse gas emissions and reducing the environmental impact of energy production. Traditional analytical modeling methods for predicting the performance of PV systems often fall short in accuracy and efficiency, especially when applied to high-power plants. To address this limitation, this research employs artificial intelligence control-based surrogate modeling utilizing a neural network (ANN), specifically a multilayer perceptron (MLP), to improve system performance prediction. With 8000 experimental samples, the ANN model is evaluated in MATLAB software through various performance indices, including mean squared error (MSE), root mean squared error (RMSE), mean absolute deviation (MAD), and mean absolute percentage error (MAPE). The regression coefficient (R2) values demonstrate the model’s accuracy, with R2 = 0.9685 for the training phase, R2 = 0.9461 for the validation phase, and R2 = 0.9643 for overall model performance. In addition, large language models (LLMs) are emerging as transformative tools for renewable energy research. These models can complement ANN-based surrogate modeling by enabling efficient processing of vast datasets, generating actionable insights from technical reports, and optimizing workflows in urban energy planning. By integrating LLMs into renewable energy systems, this study lays the groundwork for a more comprehensive approach to enhancing the annual energy performance of grid-connected high-power PV plants, paving the way for more sustainable urban energy solutions.