Construction of"Hot Skin"Thickness Prediction Model of Grain Pile Based on LSTM
In this paper,a prediction model of"hot skin"thickness of grain pile based on long short-term memory(LSTM)was proposed based on the temperature data extracted from the dense array temperature measurement system of grain silo.The results show that LSTM method can be used to predict the thickness of"hot skin"of grain pile.The three LSTM"hot skin"thickness prediction models proposed in this study,inter row east,inter row west and interlayer,all have good fitting effect,and the R2 of the three models is higher than 0.889,with low prediction error.Among them,the inter-layer"hot skin"prediction model has the best prediction effect,and the training set and validation set R2 are 0.936 and 0.921,respectively.The mean absolute prediction errors were 0.058 m and 0.049 m,respectively.The research shows that the prediction performance of the"hot skin"thickness prediction model based on LSTM has certain advantages,which can realize the accurate prediction of the"hot skin"thickness of grain storage,provide a new idea for the quantitative analysis of"hot skin"of grain storage,help warehouse managers to realize auxiliary decision-making,and provide theoretical reference for ensuring food safety storage.
long short-term memory(LSTM)thickness of"hot skin"prediction model
万方数据
