Pre-warning Model Of Stored Grain Safety Risk Based On Temperature And Humidity And Its Application

China is a big country in grain production,a country with a large population,and a big country in storage grain.Since ancient times,grain has been directly related to social stability.The state will maintain a certain amount of grain reserves for this purpose.In the process of grain storage,the inevitable losses are caused by improper storage and unreasonable grain conditions.With the development and application of China's grain storage monitoring technology and grain condition measurement and control system,a large amount of grain data is generated.Through the research and analysis of grain data,combined with the temperature trend,the grain status and change law in the warehouse are predicted in advance.Early warning and timely disposal of dangerous grain conditions to ensure the safety of grain storage has become a new research direction for safe grain storage.The safe state of grain storage is affected by various environmental factors,such as grain temperature,granary temperature,humidity in the warehouse,temperature outside the warehouse,and humidity outside the warehouse.At present,the research on early warning analysis of grain conditions mainly based on the experience of grain storage business,predicting abnormal temperature points and warming up points in the grain pile,but it is impossible to predict the development trend of future grain conditions.This paper uses the advantages of deep learning technology in forecasting and early warning,such as accurate results,simple operation and rapid response.Deep learning technology is used to predict grain temperature.Through the study of historical grain data and the safe storage of grain in different grain storage areas.The study of humidity threshold value establishes a pre-warning model for grain storage safety risk based on temperature and humidity,and provides early warning of the state of grain storage in a certain period of time,and give advice on the disposal of dangerous grain situation.The main research contents are as follows:(1)Historical grain data has typical time series.Long-short Term Memory(LSTM)can deal with timing problems very well.Therefore,LSTM network is selected to predict grain temperature and compare the advantages of different parameter optimization algorithms.The adaptive moment estimation(Adam)optimization algorithm is selected as the parameter optimization algorithm,and the grain data is preprocessed into the form of sparse matrix.The advantages of Adam optimization algorithm are fully exploited.The model is verified by comparative experiments.Applicability and accuracy of grain pile temperature prediction.(2)According to the climatic characteristics of different grain storage ecological zones,a pre-warning model for grain storage safety risk based on temperature and humidity is established.According to the critical temperature and humidity values of safe grain storage in different grain storage areas,the safety status indicators of different grain storage areas are determined.Adjusting the model parameters,combined with the grain pile temperature prediction model,can predict the safety status of the grain pile in advance by comparing the predicted value with the safe storage grain threshold value,and give reasonable disposal opinions for the dangerous grain situation.(3)The early warning module of the grain condition monitoring and control system did not provide early warning analysis of the grain situation change trend,and applied the grain storage safety risk early warning model based on temperature and humidity grain condition to the grain condition measurement and control system.It compensates for the inability of the module to warn of future grain security status and verifies the feasibility of the model.