Grain Temperature Prediction And Early Warning System Based On Deep Learning

Food security storage is the premise for the state to vigorously develop the economy,maintain basic social stability and protect national strategic security.In recent years,with the great success of China’s economy,China’s grain output has also increased significantly.However,the issue of food storage has also become a concern.Establishing an accurate grain temperature prediction model is the basic work to ensure food quality and an important means to solve food security.After a large number of theoretical and practical analysis,deep learning algorithms perform well under the current situation of data accumulation and computer performance improvement in all walks of life.Therefore,in-depth study of systematic grain temperature prediction technology has important practical significance for grain storage.Based on the above discussion,the main work of this paper is as follows:(1)Considering the temporal nature of grain temperature data and the influence of external environmental factors on grain temperature,this study uses long short-term memory neural network as the preferred model for grain temperature prediction.This model solves the problem of gradient in recurrent neural networks and overcomes the influence of long-term dependence on grain temperature data.Since the initial parameters of the model are often obtained empirically,debugging of a large number of parameter combinations requires manual debugging.In order to optimize the influence of artificial parameters,the sparrow search algorithm was used to update the learning rate,training period and number of hidden layers of the long short-term memory neural network model to find the optimal parameter combination.The preprocessed grain temperature data was used to make predictions and compared with the long short-term memory neural network to verify the performance of the optimized model.Experimental results show that the long short-term memory neural network based on sparrow search algorithm successfully solves the problem of manual parameter update,and realizes the automatic optimization of hyperparameters of grain temperature prediction model.With the help of the optimized combined parameters,the model performs better in the prediction of grain temperature datasets.(2)In order to improve the prediction accuracy of the above model and explore its practical application value in grain temperature prediction,a gated cyclic unit model based on variational modal decomposition and sparrow search algorithm is proposed.Firstly,the variational mode decomposition algorithm is used to decompose the grain temperature data into several components,so as to deeply mine the hidden details in the grain temperature data and remove the noise in the data,so that the change trend of each component is more stable.Then,the hyperparameters of the model are updated using the sparrow search algorithm to obtain the optimal hyperparameter model.The model is constructed based on each component,the prediction results of each component are superimposed,and different models are used for training and comparison to verify the effectiveness of the model.Experiments show that the gated cyclic unit model based on variational mode decomposition and sparrow search algorithm overcomes the influence of manual parameter combination,excavates the internal details of grain temperature data,and further improves the accuracy of grain temperature prediction.(3)In order to enable the grain silo supervisors to intuitively understand the change trend of grain temperature and environmental factors in the grain silo,predict the abnormal change of grain temperature in advance and formulate countermeasures,this paper establishes a grain temperature system integrating prediction and early warning.The system includes five modules: data storage,data processing,model integration,grain temperature prediction and grain temperature early warning,which simplifies user operation,enhances interactive experience,and has certain practical use value.