With the increasing complexity of the working environment and the diversification of mission requirements of UAVs, traditional UAVs have a fixed structure and single function. It is difficult to be applied in occasions with complex environments and changing load demands. The modular reconfigurable flight array (MRFA) is composed of no less than four isomorphic unit modules that are freely spliced together. By adding or removing flight unit modules and adjusting the arrangement of flight unit modules, the configuration of the MRFA can be changed, so that it can adapt to complex environments and then complete different flight missions. In the process of MRFA research and development, online configuration identification has become one of important problem to be solved. In this paper, a configuration recognition algorithm is designed based on the breadth-first searching method to identify the online structure of MRFA firstly. Then, the dynamic model of the aircraft is carried out according to the obtained configuration information. At the same time, the mathematical model and configuration information of the MRFA are combined to establish an optimization control allocation mechanism to allocate suitable rotational speed for multiple rotors. Finally, several examples are carried out to show that the proposed technology can recognize the configuration with 100% recognition rate and average online recognition time 569.4ms. Then a cascade PID controller is designed to realize the flight control of the irregular MRFA with faster tracking performance of no more than 40 seconds.

The modular reconfigurable flight array (MRFA) is composed of multiple identical flight unit modules, which has several advantages such as structural variability, strong versatility, and low cost. Due to the redundant properties of MRFA, it keeps stable by adopting a suitable control law when it suffers actuator fault or actively stops some actuators. To address the attitude stability issue of the modular flight array when actuators actively stop or encounter failures during the flight process, a modeling method based on a switched system is proposed at first, and an arbitrary switched controller design method based on the segmented Lyapunov functions and the average dwell time is also given. By introducing the actuator efficiency matrix, the dynamic switched model of the modular flight array is described. Then, a group of arbitrary switched linear feedback gains is designed to ensure the exponential stability of the flight array if the switched process satisfies the constraint of the average dwell time. Simulation and experiment results indicate that when there is an accident in the actuator states, the switched controllers can achieve precise tracking of the desired trajectory, thus confirming the effectiveness of the proposed modeling method and controller.