Temperature Plays Crucial Role in Grain Storage Process, Food Security

Strategic Insights -

Researchers at Henan University of Technology have made significant breakthroughs in understanding the role of temperature in the grain storage process and its impact on food security. Led by Hailong Peng, the team proposed a data augmentation method that operates in both the time and frequency domains, enhancing the accuracy of grain storage temperature prediction. The study utilized a long short-term memory (LSTM) network, enhanced with convolution layers and a Squeeze-and-Excitation Networks (SENet) module, achieving a 74.77% reduction in Mean Absolute Error (MAE) and a 74.02% reduction in Root Mean Square Error (RMSE) compared to the original LSTM network.

Key Takeaways:

  • Researchers at Henan University of Technology proposed a data augmentation method for grain storage data that operates in both the time and frequency domains, addressing the issue of small sample sizes in grain storage temperature data.
  • The proposed method adds Gaussian noise to grain storage temperature data in the time domain and employs the Fast Fourier Transform (FFT) in the frequency domain to highlight periodicity and trends.
  • The enhanced grain storage temperature prediction model utilizes a LSTM network with convolution layers for feature extraction and a SENet module to suppress unimportant features.
  • Experimental results show that the FTA-CNN-SE-LSTM model outperforms the original LSTM network, achieving a 74.77% reduction in MAE and a 74.02% reduction in RMSE.
  • The research concluded that the proposed method solves the problem of data limitation in the actual grain storage process, significantly improving the accuracy of grain storage temperature prediction and preventing problems caused by abnormal grain pile temperature.
  • The study was funded by the National Key Research And Development Program Project and the Natural Science Project of Henan Provincial Department of Education.

Statistics:

  • 74.77% reduction in Mean Absolute Error (MAE) compared to the original LSTM network
  • 74.02% reduction in Root Mean Square Error (RMSE) compared to the original LSTM network
  • 14(10):1671, issue and page number of the journal article in Foods
  • June 2025, month and year of the study publication
  • 450001, zip code of the location of Henan University of Technology
  • 10.3390/foods14101671, DOI of the journal article

Sources:

  • Research on Storage Grain Temperature Prediction Method Based on FTA-CNN-SE-LSTM with Dual-Domain Data Augmentation and Deep Learning. Foods, 2025,14(10):1671.
  • (Foods - http://www.mdpi.com/journal/foods)
  • MDPI AG, publisher of the journal Foods
  • Hailong Peng, lead researcher and contact person at Henan University of Technology
  • Yuhua Zhu and Zhihui Li, additional authors of the research
  • National Key Research And Development Program Project
  • Natural Science Project of Henan Provincial Department of Education