Research on Temperature and Humidity Prediction Model of Granary Based on RNN-LSTM

This paper studies the temperature forecasting model of grain bins with short-term or small data set temperature data and time series characteristics in the case of relatively short-term temperature changes or relatively small datasets of stored grain and food conditions. Input three temperature and two humidity, use SPSS for principal component analysis, process the collected data, and use the pre-processed group data to build the model. Take the former group of data as the training data set, and the latter group as the test data set. On the basis of RNN, modify each layer of neural network to set three gates, improve the RNN change learning rate to change the possibility of local optimal solution. Based on the analysis of the time series characteristics of the temperature of the granary, and comparing the LSTM prediction model, WU model and the three-pass theory model, it can be seen that the LSTM prediction model has good results on the temperature of the stored grain, and the degree of fitting is high. In the end, a basically satisfactory prediction effect can be achieved.