关键词: 电动无人机, 电动力系统, 最小功率控制, 自适应扩展卡尔曼滤波, 神经网络, 改进天牛须算法, 模糊PID

Abstract: The application of unmanned electric propeller aerial vehicles is gaining increasing popularity. However, the endurance time of these aerial vehicles is short, which partly results from the low efficiency and high power consumption of the electric propulsion system. Improving the efficiency and reducing the power consumption are therefore the main measures to extend the endurance time. The motor-variable pitch propeller propulsion system (variable pitch electric propulsion system) can simultaneously change the rotational speed and the pitch to reach an optimal pitch, enabling the minimum power consumption of the system. Compared with motor-fixed pitch propeller system, this system has particular advantages in energy consumption, yet how to achieve the minimum power point is less studied at present. To solve this problem, the back propagation neural network training method combined with the improved beetle antennae search algorithm is used to establish the neural network agent model of the variable-pitch propulsion system. Then a power optimization and control strategy of the system is proposed. With certain inflow speed and force demand, the Newton’s method based on an improved extended Kalman filter algorithm is adopted to optimize the pitch. At a certain pitch point, the fuzzy PID control is applied to adjust the rotational speed in order to produce the target force, finally achieving the minimum power as well as the target force. Simulation results show that this strategy has stronger robustness, faster optimization speed and better convergence effect.

Key words: electric unmanned aerial vehicles, electric propulsion system, minimum power control, improved adaptive extended Kalman filter, neural network, improved beetle antennae search algorithm, fuzzy PID