Scholarlink

In the era of Industry 4.0, the integration of intelligent systems intSo agriculture and food processing is driving advancements in efficiency and sustainability. A key challenge in this domain is moisture prediction in continuous corn drying systems, where precise monitoring and control are essential for ensuring product quality, reducing energy consumption, and maintaining economic viability. This study proposes a data-driven approach utilizing gated recurrent unit (GRU) time-series models to predict outlet corn moisture content based on an integrated dataset comprising temperature sensors, process parameters, and weather conditions, such as air humidity, temperature, precipitation, and solar radiation. The GRU model was trained on a dataset of 3826 samples collected from an industrial drying system to capture the intricate relationships influencing moisture dynamics. The inclusion of weather data enhanced the dataset’s ability to represent external environmental factors affecting the drying process. A grid search was conducted to optimize the model’s hyperparameters, ensuring peak performance. The GRU architecture was selected for its proven capability to model temporal dependencies in complex time-series data. Model performance was evaluated using key metrics: Mean Squared Errors (MSE), Root Mean Squared Errors, Mean Absolute Errors (MAE) and Mean Absolute Percentage Errors (MAPE). The results demonstrated the GRU model’s strong predictive accuracy, achieving MSEof0.17, RMSE of 0.41, MAE of 0.28and MAPEof 2.08on the test dataset. This study underscores the potential of GRU-based models as intelligent tools for optimizing moisture prediction in continuous drying systems. By integrating machine learning with real-time process data, weather conditions and hyperparameter tuning, this approach exemplifies Industry 4.0 principles, offering a pathway to enhanced efficiency, sustainability, and economic performance in industrial food processing.

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

Intelligent Corn Moisture Prediction for Continuous Drying Systems: A GRU Time-Series Approach in the Era of Industry 4.0

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Marko Simonič,
  • Jernej Hernavs,
  • Lucijano Berus,
  • David Potočnik,
  • Tadej Peršak,
  • Simon Klančnik

摘要

In the era of Industry 4.0, the integration of intelligent systems intSo agriculture and food processing is driving advancements in efficiency and sustainability. A key challenge in this domain is moisture prediction in continuous corn drying systems, where precise monitoring and control are essential for ensuring product quality, reducing energy consumption, and maintaining economic viability. This study proposes a data-driven approach utilizing gated recurrent unit (GRU) time-series models to predict outlet corn moisture content based on an integrated dataset comprising temperature sensors, process parameters, and weather conditions, such as air humidity, temperature, precipitation, and solar radiation. The GRU model was trained on a dataset of 3826 samples collected from an industrial drying system to capture the intricate relationships influencing moisture dynamics. The inclusion of weather data enhanced the dataset’s ability to represent external environmental factors affecting the drying process. A grid search was conducted to optimize the model’s hyperparameters, ensuring peak performance. The GRU architecture was selected for its proven capability to model temporal dependencies in complex time-series data. Model performance was evaluated using key metrics: Mean Squared Errors (MSE), Root Mean Squared Errors, Mean Absolute Errors (MAE) and Mean Absolute Percentage Errors (MAPE). The results demonstrated the GRU model’s strong predictive accuracy, achieving MSEof0.17, RMSE of 0.41, MAE of 0.28and MAPEof 2.08on the test dataset. This study underscores the potential of GRU-based models as intelligent tools for optimizing moisture prediction in continuous drying systems. By integrating machine learning with real-time process data, weather conditions and hyperparameter tuning, this approach exemplifies Industry 4.0 principles, offering a pathway to enhanced efficiency, sustainability, and economic performance in industrial food processing.