Scholarlink

This paper presents a Model Predictive Control (MPC) strategy enhanced by a Long Short-Term Memory (LSTM) forecasting model for optimizing indoor environmental conditions. The LSTM model is trained on historical data to predict key variables, including temperature, humidity, CO₂ concentration, and solar irradiance, over a 24-h horizon. These predictions serve as inputs for the MPC, which dynamically adjusts HVAC systems, CO₂ injection, lighting, and humidity control to maintain optimal conditions while minimizing energy and resource consumption. Simulation results demonstrate the effectiveness of the proposed approach in reducing deviations from optimal setpoints and enhancing indoor climate regulation. The integration of LSTM-based forecasting into MPC improves control efficiency, enabling a more adaptive and energy-efficient decision-making process. This study highlights the potential of data-driven predictive control for sustainable and intelligent indoor climate management. Future work will focus on real-world implementation and adaptive learning mechanisms to further enhance the robustness and performance of the system.

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

Predictive Control for Optimal Greenhouse Micro-Climate Management: An LSTM-Based Approach

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Hamza Benzzine,
  • Khalid Bouali,
  • Abderrahim Bajit,
  • Hicham Labrim,
  • Driss Zejli,
  • Rachid El Bouayadi

摘要

This paper presents a Model Predictive Control (MPC) strategy enhanced by a Long Short-Term Memory (LSTM) forecasting model for optimizing indoor environmental conditions. The LSTM model is trained on historical data to predict key variables, including temperature, humidity, CO₂ concentration, and solar irradiance, over a 24-h horizon. These predictions serve as inputs for the MPC, which dynamically adjusts HVAC systems, CO₂ injection, lighting, and humidity control to maintain optimal conditions while minimizing energy and resource consumption. Simulation results demonstrate the effectiveness of the proposed approach in reducing deviations from optimal setpoints and enhancing indoor climate regulation. The integration of LSTM-based forecasting into MPC improves control efficiency, enabling a more adaptive and energy-efficient decision-making process. This study highlights the potential of data-driven predictive control for sustainable and intelligent indoor climate management. Future work will focus on real-world implementation and adaptive learning mechanisms to further enhance the robustness and performance of the system.