The Attitude Optimization Of Multi-rotor UAV

Multi-rotor Unmanned Aerial Vehicles(UAVs)have been extensively used in the fields of industry,agriculture,defense and military due to the advantages of simple mechanical structure,strong maneuverability and convenient control.Under the condition of strong maneuver,such as variable acceleration or high-speed motion,the body of a multi-rotor drone has generated motion acceleration,which can result in the significant reduction in the accuracy of attitude estimation.In complex environments where GPS signals are weak or absent(such as indoor,underground mines),the position estimation errors of the multi-rotor UAV increase dramatically,and even the attitude is out of control.In view of the above problems,this paper studies the attitude optimization of multi-rotor UAVs,and on the basis of building an experimental platform of multi-rotor UAVs,an attitude solving algorithm based on improved extreme learning machine and a combined navigation algorithm based on UWB are proposed.The main work includes the following aspects:In the view of above problems,an attitude estimation algorithm based on improved extreme learning machine and a combined navigation algorithm based on UWB are proposed in this paper,which are used to realize the attitude optimization of multi-rotor UAV by building an experimental platform of multi-rotor UAV.The main researches are listed as follows:Firstly,the flight control system and the whole hardware platform of the multi-rotor UAV have been constructed.The schematic diagram and PCB of the system(including the main control module,AHRS module,UWB module,PWM output module,various communication modules,etc.)were designed and drawn,the complete system circuit board was generated through a series of procedures of proofing,welding and debugging,and a complete hardware platform was formed by assembling with the frame,brushless motors,propellers and other components.Secondly,the UAV flight control system software and corresponding host computer software have been designed and developed by using the development tools such as STM32CubeMx,MDK IDE,Qt and so on.UAV system software includes sensor reading tasks,attitude control tasks,upper and lower computer communication tasks,etc.The host computer software has been used to realize the communication and display of data such as attitude information,position information and positioning information.Thirdly,in order to solve the problem of the low attitude estimation accuracy in multi-rotor UAVs under the condition of strong maneuver,an attitude estimation algorithm based on improved extreme learning machine(IELM)has been proposed.The algorithm has two working modes,namely training mode and prediction mode.In the training mode,the multi-rotor UAV is in a hovering or slow-moving state,and the acceleration data is reliable.The sample data were classified according to the flight mode and the network parameters of ELM were trained online with the classified sample data.In the prediction mode,the multi-rotor UAV is in the state of strong maneuver,and the motion acceleration component of the acceleration data increases,which will adversely affect the attitude calculation.The extreme learning machine is used to dynamically adjust the fusion weight in the complementary filter algorithm and the step size of gradient descent to reduce the system's trust in the accelerometer,which optimizes the contribution of accelerometers and gyroscopes to attitude estimation.The experimental results show that the accuracy of the improved attitude estimation algorithm is significantly improved and the attitude optimization effect is remarkableFinally,for the problem that the position estimation accuracy of multi-rotor UAVs decreases in the complex environment of no GPS signals,an integrated navigation algorithm based on UWB has been proposed.The three-sided positioning algorithm was used to solve the position of the multi-rotor UAV in the ground frame by taking advantage of the UWB module with high ranging accuracy and strong real-time performance.UWB positioning data were input measurement equation of Kalman filter,the position and speed estimates of the multi-rotor UAV were calculated through state equation of Kalman filter.In order to verify the feasibility of UWB-based integrated navigation algorithm,simulation experiments were carried out based on Linux system and real aircraft flight experiments based on hardware platform.Experimental results show that UWB-based integrated navigation algorithm has improved the accuracy of position control in complex environment without GPS.Figure[73]table[5]reference[69]...