The Key Technology Research On Redundant Inertial Navigation System

The Strapdown inertial navigation system is the main development direction of inertial navigation system due to it’s advantages.Since the inertial sensors’ accuracy and reliability has been unable to obtain guarantee,the development of strapdown inertial navigation system is limited.The redundancy technology can significantly improve the accuracy and reliability of the inertial navigation system on the existing inertial sensor level.At present,the redundant inertial navigation system has become a hotspot in the research of inertial navigation technology.In this paper,several key technical issues of redundant inertial navigation system are studied,such as redundant sensor configuration,initial alignment,fault detection and isolation,redundant sensor optimal data fusion and optimal navigation solution etc.The redundant sensor configuration is first issue to be addressed in redundant inertial navigation system.In a redundant inertial navigation system,the more number of redundant sensor,the navigation accuracy and reliability of the system is higher,but the complexity of the system itself,the weight,volume and cost also greatly increased.The relationship between the number of sensors and the system reliability was studied,and the figure of merit for system reliability performance was proposed.According to the figure of merit,the optimum number of sensors for redundant inertial navigation system can be obtained.The optimal sensor configuration for navigation performance and fault detection and isolation performance has been studied when the number of sensors was determined.The simulations show that the dodecahedron symmetric configuration is optimal redundant sensors configuration.Before performing the navigation task,initial alignment of the strapdown inertial navigation system must be carried out.The accuracy and speed of Initial alignment are two important technical indicators.The alignment system observability is seriously affecting the speed of initial alignment.According to the redundant output of sensors in redundant inertial navigation system,a new non-linear alignment model was proposed.And the inertial sensor outputs directly are as system observables,attitude angles are as the initial alignment state variables in this alignment model.The dimension of this non-linear model was reduced to three,the system observability was improved,and the common errors of sensors could be eliminated.The simulation results show that the new non-linear model is not muchimprovement in the steady-state accuracy,but the rapidity is greatly increased.In alignment process,the three misalignment angle convergences in five seconds when using new alignment model.The initial goal about the redundant inertial navigation system is focused on the reliability of navigation system.The advanced fault detection and isolation techniques can ensure the reliability of redundant inertial navigation systems.For the problems of fault detection method based on parity equations,an improved real-time drift compensation method was proposed.It can estimate value of the drift error signal using real-time state estimation techniques and reduce the signal error using state feedback technique.The adaptive extended Kalman particle filter is used to solve that the noise does not satisfy the Gaussian model and the model is not accurate problem.A drift factor is introduced,which can weaken dependence of feedback gain for drift compensation and increase the range of feedback gain.Simulation results show that the improved fault tolerant algorithm can compensate the sensor drifts effectively and improve the robustness and reliability of the redundant inertial navigation system.And the improved fault tolerant control algorithm is insensitive to the feedback gain,and it can enlarge the range of feedback gain due to the drift factor.Once the reliability of redundant inertial navigation system can be met,the accuracy of system should be improved as far as possible.The redundant sensor optimal fusion algorithm can take full use of the redundant observation for data fusion,and the navigation accuracy can be improved.The projection of redundant sensor output to the left null space of the sensor configuration matrix is used as these redundancy observations of fusion algorithm.The optimal fusion algorithm can get the best characteristics of all sensors.The optimal fusion algorithm performance would reduce rapidly when the noise characteristics of the system is unknown.In order to solve the problem,an improved adaptive optimal fusion algorithm was proposed.The strong tracking algorithm with noise estimator was brought to optimal fusion algorithm,which can both ensure that system noise was estimated in real time and ensure that the system can be convergence when uncertainty model.When the sensors occur fault drift,the fault drift compensation algorithm and optimal fusion algorithm are combined as an optimal fusion drift compensation algorithms.Simulation results show that the improved adaptive fusion algorithm can effectively solve the problem of performance degradation when the system noise characteristics is unknown or unreasonable setting.However,the estimationaccuracy of the algorithm was sacrificed when it solved the robustness of filtering.The optimal fusion drift compensation algorithm can effectively solve the problem drift of the sensor,and which can get optimal estimation due to the full integration of the transient characteristics of all sensors.The optimal fusion method of redundant sensors can get accurate inertial information,but it can not obtain the optimal navigation accuracy necessarily.The optimal navigation solution for redundant inertial navigation system is developed.A pseudo-optimal navigation solution is proposed to solve the defects of the optimal navigation solution.In pseudo-optimal navigation solution,the redundant observation is no longer as navigation Kalman input,just for optimal estimation inertial information.When the system noise can not be accurately informed,the improved pseudo-optimal navigation solution with noise estimator is proposed.The autocovariance least squares estimation algorithm is used to estimate the measurement and system noises,performance of the pseudo-optimal navigation solution is ensured.Finally,the simulations prove that the performance of pseudo-optimal navigation solution is not poor than optimal navigation solution.The performance of pseudo-optimal navigation solution is significantly better than the optimal navigation solution when sensors suffer fault drift.The improved pseudo-optimal navigation solution can ensure the optimal performance of navigation when the system noise is unknown or unreasonable setting.