主催: 日本エネルギー学会 石炭科学部会, コークス工学研究部会, 重質油部会
共催: 石炭・炭素資源有効利用研究会
会議名: 第60回石炭科学会議発表論文集
開催地: 京都リサーチパーク
開催日: 2023/10/31 - 2023/11/01
p. 28-29
Part of the ash generated during the combustion process of solid fuels such as pulverized coal can adhere to the surface of heat transfer tubes, causing heat transfer inhibition and impairing boiler operation. In this study, the physical properties of three types of coal were compared to evaluate the shape classification of fly ash. In addition, the accuracy of the shape classification was investigated by using deep learning, including the use of training images, various parameters, and data expansion, as well as the relationship between the shape classification and melting properties based on thermodynamic equilibrium calculation. We utilized deep learning to categorize the morphology of fly ash particles and analyzed the formation process of particles, considering the thermophysical characteristics of coal ash.