It is widely acknowledged that PV-fuel Cell-battery System (PV-FC-BS) hybrid energy storage system can effectively solve the intermittent and volatility of PV output power. And it is significant for the application and promotion of PV-FC-BS to overcome the limitations of each component utilization and realize efficient power dispatching. Traditional energy management strategy often creates separated prediction models in different switch modes. It is difficult to guarantee the stability in mode switching process. New BS charging-discharging constraints are proposed and full-state multiple-input-multiple-output dynamic power exchange model is built. Therefore such energy management problem is transformed into an optimization problem, reducing the complexity of algorithm and avoiding the impact of mode switching. PV-FC-BS hybrid system energy optimal allocation is realized by designing dynamic prediction model and quick rolling optimization methods. Taking charging or discharging efficiency coefficient of batteries into account, an adaptive estimation algorithm of battery constant charging-discharging parameter is proposed to achieve an accurate scheduling of energy. In the Matlab environment, PV-FC-BS hybrid system simulation results show that the proposed strategy has small optional calculation as well as accurate energy assignment and it is adaptable to different application conditions. This work is supported by National Natural Science Foundation of China (No. 51407114), Natural Science Foundation of Shanghai (No. 15ZR1418200 and No. 15ZR1418000), Shanghai Engineering Research Center of Green Energy Grid-connected Technology (No. 13DZ2251900).