To improve and optimize the description of high-temperature and high-pressure behaviors of complex nanostructured carbon cluster products in explosives such as triaminotrinitrobenzene （TATB） and benzotrifuroxan （BTF）， and effectively enhance the prediction accuracy of the detonation thermodynamic calculation program VPL for detonation parameters of the above explosives. Based on the physical model established by molecular dynamics research on carbon condensation， this study introduces a carbon phase fraction algorithm. Combined with the single-phase equations of state （EOS） of graphite and diamond， a novel equation of state named NDGP （Nano-Diamond-Graphite-Peng） is established to characterize diamond-graphite core-shell composite nanostructured carbon clusters. Meanwhile， by modifying the single-phase EOS of graphite， a new EOS denoted as NOCP （Nano-Onion-Carbon-Peng） is developed for describing onion-like nanocarbon clusters formed under high detonation temperature conditions. NDGP and NOCP are adopted to calculate the detonation velocity， detonation pressure， overdriven detonation Hugoniot relations and work capacity of detonation products for TATB （including TATB-based composite explosives） and BTF， respectively. By comparing the calculated results with the data obtained from two classic condensed carbon EOS （Fried and Cowan-Fickett models）， it is verified that the two newly developed EOS for condensed carbon products can achieve more accurate prediction of C-J detonation parameters of TATB and BTF. Specifically， the prediction accuracy of TATB detonation velocity is generally increased by 1.5%-2.5%， and more accurate evaluation results can also be obtained for product work capacity and overdriven detonation Hugoniot relations. In addition， an EOS for characterizing the thermodynamic properties of low-density disordered carbon products is introduced. The detonation velocity-density correlation of the typical primary explosive lead trinitroresorcinate （LTNR） is systematically calculated. Compared with the Fried and Cowan-Fickett EOS， the overall prediction accuracy is improved by 3%-7%.