Abstract: Energy is the main constraint of embedded system design. In this paper, the General Stochastic Petri Net (GSPN) is used to describe the processes of electrochemical polarization and active substances diffusion in Li-ion batteries. A abstract high-level non-real battery model is presented which accurately characterized the rate-dependent and recovery feature of Li-ion batteries. The model simply adopts four easy-obtained parameters. Constant current discharge and pulse discharge experiment demonstrate that the lifetime of chemical battery could be fast and accurately estimated with this battery model.