In order to improve the operating stability， output power and energy utilization of multi-rotor hybrid drones， GT-Power and Simulink are used to jointly build models， and rules-based consumption management strategy and equivalent consumption minimization strategy （ECMS） are compared. Adaptive-ECMS （A-ECMS）based on back propagation （BP） neural networks optimization is designed and developed. Simulation shows that in terms of operational stability of A-ECMS， the overall operating speed fluctuation rate is 7.74%， which is significantly lower than those of the rule-based strategy and ECMS. The speed fluctuation rate of A-ECMS under compound disturbance and random turbulence is 8.32%， 7.18%， respectively. Compared with rule-based strategy and ECMS， it operates more stably under emergency conditions. A-ECMS can effectively improve the power performance of the hybrid system， allowing the engine to operate at an economical operating condition of 10 kW， and the battery power can be adjusted in real time according to changes in the state of charge （SOC）. The average fuel consumption rate of A-ECMS is 297.585 g/（kW?h）， and the overall fuel consumption is 3 755.31 g， which is significantly lower than those of the rule-based strategy and ECMS. The engine operating points are concentrated in the fuel economy zone when running under various operating conditions. The proposed method can effectively improve system economy.