Research On A GNSS/SINS Integrated Navigation Algorithm

In the field of integrated navigation technology,the integrated navigation system that combines the satellite navigation system with the strapdown inertial navigation system is one of the technologies that people pay more attention to.The advantages of both the global satellite navigation and positioning system and the inertial navigation system have played a very good complementary role.Combining the respective advantages of the two systems can improve the performance of the navigation system such as output frequency,positioning accuracy,and reliability.In addition,the advantages of various aspects of strapdown inertial navigation and the development of related technologies have attracted more attention.Therefore,it is of great significance to study the integrated navigation system that combines the satell ite navigation system with the strapdown inertial navigation system.Firstly,the basic algorithm of the quaternion-based integrated navigation system that combines a satellite navigation system with a strapdown inertial navigation system is deduced.The information fusion algorithm used in the integrated navigation system is an extended Kalman filter.The system state of the extended Kalman filter is selected using the position and velocity error of the carrier,the quaternion representing the attitude error,and the zero drift of the inertial sensor.After extended Kalman filter fusion to obtain the corresponding error correction parameters,the error correction parameters are fed back to the inertial navigation subsystem for error correction.Since the quaternion algorithm is fast and has no Euler angle singularity problems,most navigation algorithms use quaternions to represent the attitude angle of the carrier.But the existence of the error,when the extended Kalman measurement updates the state quantity,the normative system of the quaternion representing the rotation of the rigid body is destroyed,and therefore the quaternion must be re-normalized.For this reason,a novel method is proposed in this paper to maintain the quaternion's norm,that is,t o eliminate one of the numbers by using the property that the quaternion norm representing the rotation of the rigid body is equal to 1,using onl y three numbers of quaternions.And using this method to re-derived the existing quaternion-based satellite navigation and strapdown inertial navigation integrated navigation algorithm.The algorithm reduces the amount of state of a system and reduces the amount of calculation of the algorithm.In addition,a method for estimating the parameters of an acceleromete r error model using a linear neural network is also proposed.This method can be used to solve the problem that the accelerometer model parameters have no calibration or inaccurate calibration.In this method,the output of the integrated navigation system that combines the satellite navigation system with the strapdown inertial navigation system is retrieved to get a set of acceleration data which is closer to the real value.This set of acceleration data is used as input value of linear neural network,an d then the output data of accelerometer as expected value is trained for linear neural network.The weight matrix of linear neural network is the required model parameter.The weight matrix of the linear neural network is a parameter of the required accelerometer error model.Finally,simulation verification is performed on the reconstructed integrated navigation algorithm and the method of using linear neural network to estimate the parameters of the accelerometer error model.The results of simulation experiment show that the proposed integrated navigation algorithm is indeed effective,because the simulation results of this algorithm are basically the same as the simulation results of the existing quaternion-based integrated navigation algorithm,and the calculation amount is relatively small.The method of using the linear neural network to estimate the parameters of the accelerometer error model is indeed effective because the parameters of the accelerometer error model can be accurately estimated.