Abstarct: Due to the large size of modern power systems, dynamic studies are very time consuming. Many elements and variables of large power systems are unnecessary or had no effect on the behavior of dynamic systems. On the other hand accessing the whole dynamic information of a large system is very difficult. Because of this dynamic equivalent of power systems commonly used for implementation power system dynamic studies. In this thesis, Iran national power grid system is studied. At the beginning load buses are eliminated using sparse method. This has led to a decrease in the size of the network. This thesis presents a new method for choosing buses for elimination. In addition, many of the loads are eliminated by using a new algorithm and transport to the equivalent network. The dynamic behavior of the equivalent networks with high accuracy is similar to the real network. In The next step generator buses of the national power grid in diagnosis, electric buses distance is calculated using the new method, the close generator buses were marked. After identifying close generator buses, generators were placed in close groups. Network generators are divided into eight groups. The n each of these generator groups aggregated into one or two generators. Generator aggregation is done using structure preservation technique. Finally, in three different scenarios the behavior of complete and equivalent model has been compared.