Key Technical Research Of Information Fusion For Multiple Source Integrated Navigation System

High-precision time and space informations is the basis for the orderly flow of people and things,and the key technology for the information intelligence revolution.Based on the satellite navigation system,it integrates a variety of positioning and timing technology and systems to achieve ubiquitous navigation services at any time and place.It is a major development direction for technological advancement and industry leapfrogging between current and future 10-20 years.The formation of a ubiquitous navigation system is a large collection,integration,and integration process.It is actually an integrated solution integrating navigation and communication,space-based and ground-based,autonomous and auxiliary,in indoor and outdoor environments,and highly efficient integration of all effective positioning and timing technologies.Through the phenomenon,multi source integrated navigation system is a complex system of multisensor and multi-source navigation information collection,coordination,fusion and optimization.The most core of the system is the integration and processing of all subsystem navigation information.The purpose of the multi-source integrated navigation system is to achieve ubiquitous positioning navigation and timing services,but a variety of individual navigation systems have their inherent limitations,the performance of each independent subsystem(such as their navigability will still affect the performance of multi-source integrated navigation systems).In addition,in terms of overall architecture,multi-source integrated navigation systems require new information fusion processing methods and architectures to support the reconfiguration of sensors and inertial navigation units in the system.It is also necessary to develop and design new fusion filtering algorithms.It can not only be applied to the real environment for real-time processing,but also can be fully compatible with various types of navigation algorithms,including but not limited to Gaussian,non-Gaussian statistical algorithms,linear,nonlinear measurement models.Further,corresponding fault diagnosis and intelligent active fault tolerance schemes need to be formulated to make them more compatible with various navigation algorithms,and the measured values of heterogeneous sensors can be intelligently adjusted to achieve collaboration of multiple navigation systems and heterogeneous navigation information.Seamless Intelligent Fusion and Fault Tolerance in Flight Mode Further,based on the inaccuracy and time varying of the uncertainty,nonlinearity,and noise characteristics of the system model,the multi-source integrated navigation system can be made aware of the environment and modulate with the changes of environmental parameters.Adaptive adjustment to improve the real-time and stability,thereby further enhancing the system's scalability.The paper starts the following research on the above issues:(1)Research the navigation enhancement method of multi source integrated navigation system.The Uncontinuity of the DGNSS navigation service maybe due to the loss of the GNSS signal,the short time of the DGNSS reference station,or the unintentional interruption caused by the unintentional interruption of the DGNSS correction transmission caused by the hardware or software faults.Aim at the discontinuity of DGNSS,proposed a combination prediction based pseudorange and carrier phase correction prediction method.Both the theoretical analysis and the simulation results show that the combined model is better in conformity with the DGNSS difference correction prediction and is helpful to improve the prediction accuracy.Then,consider the shortcomings of the large noise and error accumulation of the strapdown inertial measurement unit,and the disadvantage of the traditional attitude calculation algorithm,such as the low precision and poor adaptability of the environment,the attitude solution algorithm aided by the Elman neural network is proposed.The simulation results show that this method greatly improves the accuracy and adaptability of attitude calculation.Finally,based on the research of the relative navigation algorithm of X-ray pulsar for formation flying spacecraft,a new multi-satellite formation relative navigation algorithm is proposed.The constrained adaptive Kalman filter is used to estimate the relative position and velocity between the formations.Finally,a simulation is performed to evaluate the proposed navigation algorithm.Simulation results demonstrate the effectiveness of the proposed method.(2)The state estimation of multi source integrated navigation system and the design of intelligent active fault-tolerant scheme.According to the navigation characteristics of inertial,terrain and geomagnetism in the environment which the multi source integrated navigation system needs to face,the Laplacian Eigenmaps feature space transformation(LE)is used to projection the data from the original high dimensional space to the low dimension space to extract more representative features.The LE is integrated with the deep neural network model,and a new method is proposed to estimate the stability of the navigation state when the model is not very accurate and the noise of the system and time-varying observation is changed.The simulation shows the effectiveness of the method under the condition of bad environment or large disturbance.The system fault diagnosis method based on one-class support vector machine is proposed.The simulation results show that the method can detect the abrupt and slow change faults of the navigation subfilter system well.Moreover,through analysis and comparison,it shows that the method has better detection performance in small sample.On this basis,an intelligent active fault-tolerant scheme based on neural network and support vector machine is proposed.Compared to the general fault-tolerant integrated navigation,active fault-tolerant integrated navigation can make every subfilter work uninterruptedly,and regardless of whether the sub navigation system fails,the overall state can be estimated optimally.(3)The information fusion model and filtering method of multi-source integrated navigation systemBased on the general composition of integrated navigation systems,and the analysis of the information fusion structure of traditional integrated navigation systems,when the traditional algorithms in a data-intensive and complex environment,the speed of a single data processing cannot match the amount of data received per second,resulting in more and more data accumulated in the pending queue.As time passes,the waiting time in the queue of the data received later is getting longer and longer and becomes a vicious circle,which seriously affects the efficiency of system operation.Based on analyzing the parallel structure of GPU,a particle filter algorithm,which deals with non-linear problems,is taken as an example for GPU-based information fusion model of MSINS.Finally,the filtering methods of multi-source integrated navigation system based on hybrid filter algorithm is studied and simulated.