| Grain storage temperature is an important indicator of grain security.Obtaining high-precision grain temperature distribution data is of great significance for grain condition analysis decision-making.In this paper,a neural network-based three-dimensional visualization research method of grain storage temperature is proposed combined with the grain temperature data measured on site in the grain warehouse.A three-dimensional grain temperature distribution map is constructed by spatial interpolation algorithm to determine the safety condition of grain storage,which is used to assist grain analysis.The main contents of this paper include:1.Statistical analysis of grain temperature dataFrom the perspective of different cross-sections of sensor layout and time changes,statistical analysis of the obtained grain temperature data set is carried out to initially grasp the grain temperature distribution of the research object in each month.The statistical analysis results show that the spatial distribution of grain temperature presents a half-year cycle from differentiation to homogenization,and the grain temperature distribution gradually disperses from the upper layer to the lower layer of the grain pile in the vertical direction.2.Comparative research on the interpolation algorithms of grain temperatureBP neural network,ordinary Kriging interpolation and inverse distance weighted average interpolation are used to establish the spatial interpolation algorithm of grain temperature on the measured grain temperature data.Experiments are carried out from two perspectives of three-dimensional space and two-dimensional layer section,and the accuracy of these three interpolations are compared.The experimental results show that in the three-dimensional spatial interpolation experiment,the BP neural network has the highest accuracy,with an average absolute error of 1.405℃;in the two-dimensional layer section interpolation experiment,the accuracy of the three interpolation algorithms is similar,and the ordinary Kriging interpolation method has the highest accuracy BP neural network comes next.Therefore,the BP neural network interpolation method has high accuracy and can be applied to the grain condition cloud map drawing in actual engineering projects.3.Realization of three-dimensional visualization software for grain temperature spaceUsing visualization tools such as the matplotlib and the three.js,the grain temperature 3D visualization software was developed.The software uses the method of slicing reorganization to realize the function of three-dimensional visualization,which can more intuitively display the spatial distribution and change trend of the temperature field of the grain pile,and better judge the state of the grain condition and its change trend,and is used to guide the on-site grain condition management and decision-making. |
