Scholarlink

The increased dependence on PV energy systems makes forecasting rather critical because of their intermittent nature and dependency on weather conditions. This study conducts a comparative study of three deep learning models—Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and their combination CNN-LSTM—on forecasts of PV energy production and energy consumption of buildings. Using a dataset from the presidential building of Ibn Tofail University, the work evaluates model performance by the standard metrics: Mean Squared Error, Root Mean Squared Error, and Mean Absolute Error. The hybrid CNN-LSTM model outperforms the standalone CNN and LSTM models in the accuracy concerning spatial and temporal patterns with the lowest error rates. These results demonstrate that the hybrid model effectively captures complex time series data, and thus it is the most effective choice for energy forecasting. The result of this study develops energy optimization techniques by proposing CNN-LSTM as an effective solution for good and accurate forecasting in renewable energy systems.

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

Comparison of Deep Learning Models for Forecasting PV Energy Production and Consumption: A Case Study Using CNN, LSTM, and CNN-LSTM Hybrid

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Kaoutar Ait Chaoui,
  • Oumaima Choukai,
  • Hassan El Fadil,
  • Chakib El Mokhi,
  • Oumaima Ait Omar

摘要

The increased dependence on PV energy systems makes forecasting rather critical because of their intermittent nature and dependency on weather conditions. This study conducts a comparative study of three deep learning models—Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and their combination CNN-LSTM—on forecasts of PV energy production and energy consumption of buildings. Using a dataset from the presidential building of Ibn Tofail University, the work evaluates model performance by the standard metrics: Mean Squared Error, Root Mean Squared Error, and Mean Absolute Error. The hybrid CNN-LSTM model outperforms the standalone CNN and LSTM models in the accuracy concerning spatial and temporal patterns with the lowest error rates. These results demonstrate that the hybrid model effectively captures complex time series data, and thus it is the most effective choice for energy forecasting. The result of this study develops energy optimization techniques by proposing CNN-LSTM as an effective solution for good and accurate forecasting in renewable energy systems.