With the rapid progress of nanotechnology, application of nano-scale materials has been extensively increased. Due to increase of surface effects at small sizes, the classical theories are not capable of modeling nano-scale systems. On the other hand, more accurate methods at small sizes are based on the quantum and atomistic models which are too time consuming, and hence using these methods is limited to very small sizes for a short period of time. In this research coarse-graining models for accelerating molecular dynamics simulations for the analysis of silicon structures are proposed. In this technique, after assigning a proper map between beads of the coarse-grained model and atoms of the main structure, the system parameters are modified in a scheme that the original model and the coarse-grained models have the same physical properties. By using various static and dynamic simulations and evaluating the size effect, the accuracy and speed of the proposed model is examined. The error of this CG model for investigating the Young modulus, longitudinal and transversal vibration is less than 5 percent, while it is about 8 times faster than AA model.