Fusion Attitude Solving Algorithm of Multi Rotor Aircraft Based on Extended Kalman Filter

Aiming at the noise disturbance of accelerometer caused by body vibration of multi rotor aircraft during flight, a structure model of radial basis function (RBF) neural network was established, and an extended Kalman filter fusion attitude solving algorithm was proposed. The attitude angles calculated by the accelerometer were used as the input vectors, and the attitude angles calculated by the quaternion complementary filter were used as reference vectors, the optimal weight of network was adjusted continuously by learning, and the attitude angle calculated by accelerometer after filtering was obtained. The attitude angles calculated by the quaternion complementary filter algorithm and the accelerometer solution angles obtained by the nonlinear filtering of the RBF neural network were used to perform the extended Kalman filter fusion attitude calculation and to improve the accuracy of attitude angle calculation. The experimental results show that the nonlinear filtering algorithm based on RBF neural network can effectively filter the attitude angle of the accelerometer solution, and improve the accuracy of the aircraft attitude angle. The proposed extended Kalman filter fusion attitude algorithm has fast convergence and strong stability, and can more accurately calculate the current attitude of the aircraft in real time, and the feasibility of the algorithm is verified.