Corn kernel storage temperature prediction based on the LSTM algorithm

Grain storage is an extremely complex process that is easily affected by the external environment. A temperature gradient is formed between the inside and outside of the grain pile, making the pile prone to rot and mildew.

To reduce the loss of grain and the occurrence of pests and mildew, this study uses a self-made test chamber and detection system to detect the grain temperature at different locations of corn grain storage and analyze the temperature change and the heat transfer process across the entire grain pile. Next, the prediction of corn grain storage temperature based on the LSTM algorithm is investigated by analyzing the corn grain storage test data. Based on the attributes and characteristics of the data, linear algebra, probability theory, and algorithm principles are applied to form a comprehensive decision and predict the corn grain storage temperature.

The test results show that the atmospheric temperature outside the bin greatly influences the grain temperature, and the grain temperature in the bin falls behind that outside the bin as well as the ambient temperature. The temperature in the bottom No. 2 position always remain slow during the detection period, while the highest temperature occurs in the top No. 12 position.

Based on the analysis of grain pile temperature change, this study investigates the prediction of grain storage temperature based on the LSTM algorithm. Compared with the experimental values, the accuracy of the test sets of the first, second, third, and fourth layers is 0.62, 0.89, 0.83, and 0.79, respectively. Thus, the prediction results of the second and third layers of the grain pile have higher accuracy. The bottom and top layers of the grain pile in the test bin are easily affected by the ambient temperature, the heat exchange rate of the grain pile is fast, and the temperature changes rapidly, yielding less accurate prediction results for these layers. The research presented in this study provides new insights into the prediction of grain storage temperature.