Modeless Constant Current Control For Wireless Drone-In-Flight Charging

Along with the demand of people for intelligent and modern society,the drones are profoundly known and extensively utilized by various applications,such as agriculture,photography,package delivery,mapping and inspection of power line or oil pipeline etc.In the large-scale and long-time working condition,the flight duration of drones is limited by the capacity of battery.Wireless in-flight charging is an effective way to extend the flight duration.However,the wireless in-flight charging need to deal with the impact of continuous variation of coupling effect,the switching of charging current,the fluctuation of load,and the parameter shifting.In addition,by considering the characteristics of the commonly-used lithium-ion batteries in drones,the constant current(CC)control is an essential technology for wireless in-flight charging systems,since it can provide the prolonged lifecycle and security of batteries.To address the issues mentioned above,this thesis proposes a modeless constant current control for wireless in-flight charging systems,which aims to improve the rapidity and robustness of systems and maintain a constant value of output current.The main contents of this thesis are listed as below:Firstly,based on the circuit theory,the equations of input current and output current in the low-order topology and high-order topology are given.Both the impact of mutual inductance and system parameters on the output current are carried out.To address the disturbance of load,the condition of realizing a load-independent output current is deduced.Secondly,to address the main disturbance in the wireless in-flight charging systems,namely the disturbance of coupling effect,this thesis proposes a constant current control scheme based on the estimation of mutual inductance.The proposed scheme estimated the mutual inductance in a communication-free,calculation-simple,and measurement-convenient way based on the offline-trained neural network.Both the simulated and experimental results are given to verify the effectiveness of the estimation effect.In addition,the parameter dependence is analyzed.Then,the simulation is carried out to verify the feasibility of the constant current control scheme which is based on the estimated mutual inductance.Finally,this thesis proposes an online-trained neural-network-based modeless constant current control.LCC-P compensation network are utilized to realize a load-independent characteristic of output current.Then,the optimized RBF neural network is used instead of BP neural network to deal with the continuous variation of coupling effect.In addition,the proposed scheme improves the robustness of the system when facing the disturbances of system parameters and charging current.Accordingly,by adopting the proposed scheme,the constant current can be effectively maintained,which is meaningful for the practical applications.