The flow characteristics of coalbed methane (CBM) are influenced by the coal rock fracture network, which serves as the primary gas transport channel. This has a significant effect on the permeability performance of coal reservoirs. In any case, the traditional techniques of coal rock fracture observation are unable to precisely define the flow of CBM. In this study, coal samples were subjected to an in situ loading scanning test in order to create a pore network model (PNM) and determine the pore and fracture dynamic evolution law of the samples in the loading path. On this basis, the structural characteristic parameters of the samples were extracted from the PNM and the impact on the permeability performance of CBM was assessed. The findings demonstrate that the coal samples' internal porosity increases by 2.039% under uniaxial loading, the average throat pore radius increases by 205.5 to 36.1 μm, and the loading has an impact on the distribution and morphology of the pores in the coal rock. The PNM was loaded into the finite element program COMSOL for seepage modeling, and the M3 stage showed isolated pore connectivity to produce microscopic fissures, which could serve as seepage channels. In order to confirm the viability of the PNM and COMSOL docking technology, the streamline distribution law of pressure and velocity fields during the coal sample loading process was examined. The absolute permeability of the coal samples was also obtained in order for comparison with the measured results. The macroscopic CBM flow mechanism in complex low‐permeability coal rocks can be revealed through three‐dimensional reconstruction of the microscopic fracture structure and seepage simulation. This study lays the groundwork for the fine description and evaluation of coal reservoirs as well as the precise prediction of gas production in CBM wells.