Attitude resolving system and method based on extended Kalman filtering

Abstract

According to an attitude resolving system and method based on extended Kalman filtering, a triaxial attitude angle with relatively high precision is obtained by resolving and calibrating data of a gyroscope and an accelerometer; and a data processing module performs smooth filtering and normalization processing on noise-containing data of an IMU sensor to obtain measurement data after noise reduction processing and unit unification, a resolving module compensates and corrects gyroscope data by combining linear complementary filtering with PI integral, the processed sensor data is imported into extended Kalman filtering to perform data fusion and attitude resolving, the data fusion and attitude resolving are carried out, a calibration module carries out online training by utilizing an Elman neural network, takes information of two IMU sensors and attitude information output by a calculation module before calibration as input, takes an attitude angle measurement value as a neural network prediction value, carries out online training by adopting three parallel Elman neural network structures, updates a weight and a threshold value of the neural network in real time, and outputs the calibrated three-axis attitude angle, so that the resolving precision and the redundancy of the system are improved.

Description

technical field [0001] The invention belongs to the technical field of attitude calculation of fixed-wing unmanned aerial vehicles, and in particular relates to an attitude calculation system and a calculation method based on extended Kalman filtering. Background technique [0002] The flight control system of the fixed-wing UAV includes many modules such as attitude control, integrated navigation, and fault diagnosis. It is the core component of the UAV, which determines the reliability and stability of the UAV. Processing and calculating the attitude to obtain a more accurate attitude angle is the first step in the entire flight control process. The attitude calculation methods currently used on fixed-wing UAVs generally use the Mahony Complementary Filter and the Extended Kalman Filter. [0003] The complementary filter (complementary filter, CF) obtains low-frequency data from the accelerometer and magnetic compass and fuses the high-frequency data obtained from the gyr...

AI Technical Summary

Problems solved by technology

[0003] The complementary filter (complementary filter, CF) obtains low-frequency data from the accelerometer and magnetic compass and fuses the high-frequency data obtained from the gyroscope for attitude calculation. The dynamic performance is lower than that of the Kalman filter, resulting in hysteresis in the flight control system, slow attenuation near the cutoff frequency, errors in the processed data, and low accuracy

[0004] The Extended Kalman Filter (EKF) uses Taylor series expansion to linearize the nonlinear system. Its dynamic performance is good, and it only relies on the parameter matrix for calculation without affecting the estimation accuracy, but it also relies heavily on prior There is a certain linearization error in the linearization process

Images