Analysis And Prediction Of Spatiotemporal Data Of Grain Temperature And Situation Research On Visualization Systems

In the field of grain warehousing,with the use of a large number of sensors,a large amount of environmental data will be generated.These data describe the internal environment information of the grain warehouse and have a crucial impact on the storage of food.But now,the direct use of in-depth learning method to predict the grain situation data cannot effectively process the sparse and sparse features in the data,nor can it remove the noise data in the data.It does not improve the prediction accuracy effectively.At the same time,the grain situation data is huge,using a single chart to describe the data will be very straightforward and dull,and for the abnormal temperature area of the grain warehouse,because the location of the sensor in the warehouse is sparse,the temperature between points can not be detected,especially for some areas may be in a relatively normal temperature range,but the temperature change between the point and the surrounding temperature is very different.Existing algorithms cannot capture this valid information and therefore cannot adjust the interior environment of the barn in advance to reduce losses.At the same time,there is no systematic integration of the grain situation data and visualization results for the large amount of data.Most of these data results are scattered in different folders,which can not achieve efficient data management and visualization results display.Therefore,this topic selects the traditional two-dimensional plane design method,combines the color theory with the analysis of visual variables,and visualizes the food data on the basis of the GRU prediction model based on sparse feature fusion in time series data.By using visual factors such as color,static visual variables and dynamic visual variables,combined with spatial prediction of grain temperature and spatial temperature transformation rate method,an excellent visualization scheme was reasonably built.Then,based on the micro-service technology,the integrated control of the dataset and visualization scheme is carried out to realize the efficient operation of the grain monitoring system.The main research work of this project is as follows:(1)The GRU(Gate Recurrent Unit)model based on sparse feature fusion and wavelet filtering denoising technology are used to study the changes and predictions of temporal data in grain storage environments.Feature generation is performed on sparse and sparse environmental data features,and wavelet filtering is used to denoise the environmental data.Finally,the data is predicted based on the GRU model with good training speed and accuracy,providing data support for future visualization.(2)Based on the results of time series data prediction,the existing color theory is analyzed,combined with the analysis of visual variables,Hermite interpolation algorithm and TCR(Temperature Change Rate)algorithm for grain temperature change are used to predict the spatial temperature data inside the barn and visualize the overall information of the grain situation.(3)Based on the prediction results and visualization schemes of the previous time series data,integrated management of the dataset and visualization scheme is carried out using microservice technology,and the efficiency of microservice technology is utilized to achieve an integrated management framework for grain storage information.