Prediction of compressive strength based on visualization of pellet microstructure data

In recent years, with the wide application of image data visual extraction technology in field industrial engineering, development economy has reached a new situation. To explore interaction between pellet microstructure and compressive strength, firstly, needed for experiment was obtained using Leica DM4500P microscope. The area proportions hematite, calcium ferrite, magnetite, silicate pore were extracted by data. Moreover, relationship mineral components strength established backpropagation neural network (BPNN), generalized regression (GRNN) beetle antennae search-generalized (BAS-GRNN) algorithms, which proves that can be used as prediction standard strength. errors BPNN BAS-GRNN are 5.13% 3.37%, respectively, both less than 5.5%. Therefore, through visualization, we able to discuss connection various provide research ideas improving metallurgical performance pellet.