Design And Implementation Of Grain Monitoring And Early Warning System Based On GRU Model

Grain is the most essential and strategic material for any countries,which guarantee the basic livelihood of people.While increasing grain output,it is necessary to maintain reasonable grain reserves.In order to effectively monitor the granary to reduce the grain loss,this thesis designed a grain situation monitoring and early warning system based on GRU model.The system can automatically collects the monitoring grain data and uploads it to the cloud server,where the neural network algorithm was used to analyze the data and predict the temperature of the grain pile.Main research works of this thesis can be summarized as following aspects:(1)Aiming at the layout of temperature sensor nodes in the granary,Finite Element Analysis(FEA)is used to simulate the grain temperature.This thesis adopts Ansys Fluent finite element analysis software to construct the temperature distribution model of granary.Based on the analysis of granary temperature in different seasons,the layout of sensors is optimized.In addition,to improve the granary monitoring and early warning,a carbon dioxide detector is added to obtain carbon dioxide data in the grain silo,where temperature,humidity and carbon dioxide are analyzed to predict the change of grain.(2)Aiming at the early warning of dangerous grain conditions,such as pests and mildew in the granary,a monitoring and early warning model of the grain is designed and implemented.The model adopts the Gated Recurrent Unit(GRU)of Recurrent Neural Network(RNN)and involves genetic algorithm to optimize GRU model parameters to improve the accuracy.Besides,the dropout method is introduced to solve the overfitting phenomenon in the model.After that,different activation functions and optimization algorithms are used to test the model.Experiments results show that,by Adam algorithm,the model whose activation function is the GRU model of Softplus,MAE is 0.048;MSE is 0.003;MAPE is 0.289% and R2 is 0.911 has the best prediction effect.(3)A grain condition monitoring and early warning system based on GRU model is designed and implemented.In the hardware design,in order to achieve more convenient and accurate data collection,4G wireless communication module is used to transmit temperature,humidity and carbon dioxide data to Alibaba's cloud server.The Linux server in the cloud server is used to store and analyze data,and the Windows system is used to display grain situation data and early warning results.Software design mainly includes data acquisition,data query and temperature warning.The system test shows that this system has fast response speed and high prediction accuracy.