To solve the problem of high accuracy initial alignment of strap-down inertial navigation system (SINS) for ballistic missile, an on-line identification method of initial alignment error based on adaptive particle swarm optimization (PSO) is proposed. Firstly, a complete navigation model of SINS is established to provide the accurate model basis for subsequent numerical optimization calculation. Then setting the initial alignment error as the optimization parameter and regarding the minimum deviation between SINS and GPS output as the objective function, the error parameter optimization model is designed. At the same time, the mutation idea of genetic algorithm (GA) is introduced into the PSO; thus the adaptive PSO is adopted to identify the initial alignment error on-line. The simulation results show that it is feasible to solve the initial alignment error identification problem of SINS by intelligent optimization algorithm. Compared with the standard PSO algorithm and the GA, the adaptive PSO algorithm has the fastest convergence speed and the highest convergence precision, and the initial pitch error and the initial yaw error precision are within 10′′ and the initial azimuth error precision is within 25′′. The navigation accuracy of SINS is improved effectively. Finally, the feasibility of the adaptive PSO algorithm to identify the initial alignment error is further validated based on the test data.

To realize the error parameter estimation of strap-down inertial navigation system (SINS) and improve the navigation accuracy for aircraft, a hybrid improved restricted Boltzmann machine BP neural network (IRBM-BPNN) approach, which combines restricted Boltzmann machine (RBM) and BP neural network (BPNN), is proposed to forecast the inertial measurement unit (IMU) instrument errors and initial alignment errors of SINS. Firstly, the error generation mechanism of SINS is analyzed, and initial alignment error model and IMU instrument error model are established. Secondly, an unsupervised RBM method is introduced to initialize BPNN to improve the forecast performance of the neural network. The RBM-BPNN model is constructed through the information fusion of SINS/GPS/CNS integrated navigation system by using the sum of position deviation, the sum of velocity deviation and the sum of attitude deviation as the inputs and by using the error parameters of SINS as the outputs. The RBM-BPNN structure is improved to enhance its forecast accuracy, and the pulse signal is increased as the input of the neural network. Finally, we conduct simulation experiments to forecast and compensate the error parameters of the proposed IRBM-BPNN method. Simulation results show that the artificial neural network method is feasible and effective in forecasting SINS error parameters, and the forecast accuracy of SINS error parameters can be effectively improved by combining RBM and BPNN methods and improving the neural network structure. The proposed IRBM-BPNN method has the optimal forecast accuracy of SINS error parameters and navigation accuracy of aircraft compared with the radial basis function neural network method and BPNN method.

Measurement Science and Technology Paper A new IMU error parameter estimation method for hypersonic vehicle based on hybrid neural network Weilin Guo1,2 , Yong Xian1, Xinlong Chang1, Leliang Ren1 and Gang Lei1 Published 28 October 2020 • © 2020 IOP Publishing Ltd Measurement Science and Technology, Volume 32, Number 1 Citation Weilin Guo et al 2021 Meas. Sci. Technol. 32 015011 171 Total downloads Turn on MathJax Get permission to re-use this article Share this article Share this content via email Share on Facebook Share on Twitter Share on Google+ Share on Mendeley Article information Abstract To realize the effective estimation of the inertial measurement unit (IMU) error parameters of a hypersonic vehicle and satisfy the high-precision navigation requirements, a hybrid neural network (HYNN) method that consists of a quasi-Newton semi-determined weight neural network (QNSWNN) and a semi-determined weight recurrent neural network (SWRNN) is proposed. In contrast to conventional neural networks, the weights of the HYNN model, which are determined by the IMU error parameters, have physical meanings. First, the IMU error model and strap-down inertial navigation system (SINS) navigation model are established. Second, the QNSWNN model is constructed based on the SINS/celestial navigation system (CNS) integrated navigation system information. The quasi-Newton is used to adjust the weights, which include the gyroscope error. The rest weights of QNSWNN are fixed based on a SINS attitude calculation model. The gyroscope error parameters can be estimated during the QNSWNN training process. Lastly, the SWRNN model is constructed based on the SINS/GPS integrated navigation system information. The BP algorithm is used to adjust the weights, which include the accelerometer error. The rest weights of SWRNN are fixed based on the SINS navigation calculation model. The accelerometer error parameters can be estimated during the SWRNN training process. The simulation results show that the IMU error parameter can be effectively estimated by the HYNN, and the relative errors are below 25%. Moreover, when the signal of the auxiliary navigation system is interrupted, the HYNN method still has high prediction accuracy for the SINS navigation parameters.

Camera pose optimization is the basis of geometric vision works, such as 3D reconstruction, structure from motion, and visual odometry. A pose optimization method based on learned metrics is proposed to improve the optimization convexity. The neural network was designed and trained based on the collected datasets, respectively. The network inputs pairwise patches and outputs the Euclidean distance of its center. This distance is involved in the residual calculation of Gauss-Newton, and the Jacobian corresponding to this distance can be analytically solved. The simulation verified convergence and generalization of the designed network. The accuracy and robustness of the proposed pose optimization compared with intensity- and feature-based optimizations are also verified.