In this paper, we combined structure-based and thermodynamic quantities-based coarse-graining (CG) approaches and adopted the bulk density, local conformational distributions and radial distribution function of the underlying atomistic model system as the target properties to derive the CG force field (FF) for the polymer system of trans-1,4-polybutadiene. The resultant CG FF exhibits good observable representability. In addition to the overall chain size, the extracted bulk modulus, isothermal compressibility, storage and loss moduli from CG model systems are of the same level of magnitude as the atomistic results, suggesting that this hybrid CG approach improves the reproduction capability of the CG model on mechanical property, pressure-dependent density fluctuations, and viscoelasticity. Upon the introduction of a standard dissipative particle dynamics (DPD) thermostat to remedy the removed degrees of freedom and reduced friction during CG, the CG model can completely mimic the dynamics of the atomistic model. Furthermore, we found that the standard DPD thermostat approach is applicable to the steady shear flow simulation at low and moderate rates and the relevant dissipative factor derived by matching diffusion coefficients can only quantitatively reproduce the shear viscosity at low shear rates. Overall, these findings demonstrate the ability of combined CG methods in optimizing CG FF to reproduce multi-properties of polymers. (C) 2015 Elsevier Ltd. All rights reserved.