Battery capacity plummeting severely threatens the endurance mileage and security of electric vehicles. A profound understanding of the mechanisms of accelerated capacity deterioration is a prerequisite for recognizing capacity plummeting online. In this paper, battery cycle life tests are performed under various cycle conditions until capacity plummeting. Based on the mathematical model of cathode and anode's open circuit voltage, an improved non-destructive diagnostic algorithm is proposed to quantify battery aging modes including loss of lithium inventory, loss of cathode active material and loss of anode active material. By analyzing the evolution characteristics of battery aging modes along with cycling, capacity plummeting mechanisms are concluded, including lithium plating on anode and collapse of cathode active material. In the first case, lithium inventory and anode active material decay sharply after the plummeting. In the second case that is more likely to occur at high temperature and high discharging current, the three aging modes all experience dramatic growth. By analyzing the commonalities of state-of-health indicator parameters evolution under the two capacity plummeting mechanisms, we propose the characterization of accelerated capacity deterioration and explore the identification method of capacity plummeting based on the cycling data.