Research On Anti-disturbance Attitude Determination Method For Bionic Polarization Integrated Navigation Systems

Attitude determination is to calculate the matrix between the body frame coordinate system and the reference frame coordinate system based on the kinematic model of the carrier and the measured values from multiple sensors.As the core technology of navigation,attitude determination has attracted extensive attention and has been applied in aerospace and other fields.At present,the development trend is to study the attitude determination method with high precision and anti-interference.For the problems of low precision,high complexity and weak anti-disturbance ability of traditional attitude determination methods,a bionic polarization sensor with strong autonomy,no error accumulation and absolute heading is introduced to complement and fuse with inertial measurement unit,magnetometer and other sensors to establish a bionic polarization integrated navigation system.In this paper,a complementary filter based on reinforcement learning strategy and an Error State Kalman filter are designed respectively.The effect of improving the attitude determination accuracy of the bionic polarization sensor system is compared and analyzed,and the effectiveness of the proposed method is verified.The main research contents are as follows::1.For the problem that geomagnetism is easy to be disturbed by the outside world,which leads to the decline of attitude estimation accuracy,the bionic polarization sensor is introduced to provide heading information correction.The inertial measurement model,bionic polarization sensor measurement model and magnetic intensity measurement model are established respectively.The complementary filter with low computational complexity is designed,and the attitude determination method of complementary filter based on Inertia/Geomagnetic/Polarization is proposed.Experiments such as geomagnetic disturbed and polarization sensor blocked are carried out.The results show that the proposed complementary filter can effectively improve the accuracy of attitude estimation and has the performance of anti geomagnetic interference and occlusion.2.For the problems of multi-mode fast switching demand when multi-sensor fusion attitude determination is faced with interference,the switching mechanism under reinforcement learning strategy is introduced,the neural network of deep Q-learning is built,and an adaptive switching bionic polarization attitude determination algorithm suitable for different environments is proposed.In order to verify the effectiveness of the proposed method under the multi-mode switching mechanism,the geomagnetic interference and polarization angle interference under multiple types and different intensities are simulated,and the influence of interference under different intensities on the accuracy of attitude estimation is analyzed.The effective fusion and switching of multiple sensors such as inertial sensor,magnetometer and polarization sensor are realized,which improves the accuracy of attitude estimation and meets the requirements of multi-mode fast switching.The results show that the proposed method can effectively reduce the heading angle error and realize the effective switching of combined modes;For complex and changing navigation environment,the bionic polarization adaptive switching attitude determination algorithm can effectively realize the adaptive switching of multi-mode integrated navigation according to different interference conditions.3.For the problems of over parameterization and singularity in the process of calculating attitude using rotation matrix,the attitude is described in the form of quaternion.An attitude determination method of Inertial/Polarization integrated navigation based on ESKF is designed,and the nominal state equation and error state equation are established.The nominal state is established by the angular velocity and the initial estimated gyro bias.The error state equation is composed of residual equation constructed by specific force vector and polarization vector.Finally,the nominal state is modified by the estimated error state to improve the accuracy of attitude estimation.In clear and cloudless weather,static experiments based on accelerometer and polarization sensor are carried out to verify the effectiveness of the proposed method and improve the accuracy of attitude estimation.