It is important to understand the amount of tar and soot generated in the thermal decomposition process of coal, because it is a factor inhibiting operation in the downsteram process. A reaction analysis method considering the chemical structure of coal based on the Chemical Percolation Devolatilization (CPD) theory has been used to predict the generation amount of pyrolysis gas, but the calculation error was large in the conventional chemical structure estimation method from ¹³C-NMR spectrum analysis. Therefore, this study aimed to construct a method for selecting chemical structural parameters. First, the CPD-α model was proposed by comparing the proximate volatiles of eight kinds of high-volatile coals with the CPD model calculation results, focusing on the reaction of α carbon connected to the aromatic cluster. Next, a chemical structure estimation method was found in which the chemical shift peak position of the bridgehead carbon was set as a function of aromatic carbon concentration and methylene carbon concentration measured by ¹³C-NMR spectral analysis. In addition, it was found that the CPD-α model combined with the structure estimation method can accurately evaluate the amount of pyrolysis gas generated in the thermogravimetric analysis and high-pressure flow tube reaction test equipment.