Research On Fault Detection And Fault Tolerant Technologies For Integrated Navigation Systems

With the development of the Solar Unmanned aerial vehicles(SUAVs)technology,the UAVs need to fly for the longer time.Ensuring unmanned aerial vehicles to obtain their own accurate motion information is one of the key technologies to ensure their long-term autonomous flight.However,in the course of a long flight,changes in the external environment may lead to accuracy deterioration of the UAVs navigation system components.This may also leads to system failure,etc.If the failures can not be timely detected and processed,it is difficult to ensure the UAVs integrated navigation system to work for a long time reliably.In view of the above demand background,this paper studies the fault detection and fault tolerant technology of the high reliability integrated navigation system of the solar UAVs.A multi-information fusion fault-tolerant integrated navigation simulation system for solar UAVs is constructed.And the fault detection and fault-tolerant algorithms of the integrated navigation system are designed.The main work is as follows:(1)In order to improve the fault-tolerant performance and accuracy of multi-sensors integrated navigation system when failure occurs,the Strap-down inertial navigation system/Beidou navigation satellite system/Global positioning system/Celestial navigation system(SINS/BDS/GPS/CNS)integrated navigation system is designed.A multi-sensors fusion structure based on federal filter is proposed.And the federal filter weights are distributed through fault detection information.The reference state information output by the SINS model and the measurement information of each sensor constitute the local filtering subsystem.The state information output by the local filter subsystem and the reference state information are both input to the fault detection module to perform fault detection.The output fault detection value is used as an input variable for the preset membership function.Through this function,the distribution weights of the state estimation for each local filter subsystem can be calculated.The global state estimation of the system is obtained by integrating the state estimation of the local filter subsystem by the weights.The global state estimate is fed back to SINS for compensation to reduce the effect of SINS error on the solution accuracy.Through this structure,the failure is easier to be detected and processed.And the impact of fault sensors on the system can be reduced.(2)It is difficult to detect and deal with dual-fault problems for multi-sensors fusion system.To solve these problems,a fault-tolerant integrated navigation algorithm combining double-state propagation chi-square test and fuzzy adaptive filtering is proposed.Two identical system models are run in parallel as state propagators of the double state propagators chi-square test.The global estimation information fused by the multi-sensors information is regularly used to correct the state propagators.Through this method,the accuracy of the propagators can be improved.The faliure detection value is input to the preset fuzzy membership function of the fault detection.And the probability of failure for each measurement subsystem can be calculated.The federal filter weights of each local subsystem can be calculated by the fault-free probability of each measurement subsystem.The global state estimation of the system is obtained by changing the federated filter weights adaptively to process fault information.The theoretical analysis and simulation results show that the algorithm combines the advantages of both.And the algorithm can effectively detect and deal with various types of faults in the measurement subsystem when the accuracy of SINS is high.(3)To solve the problem that the accuracy of the integrated navigation system is degraded as the change of statistical characteristics of SINS modeling noise in different maneuver modes,an improved interactive multi-models(IMM)multi-sensors data fusion fault-tolerant integrated navigation algorithm is proposed.The system modeling noise statistical characteristics under the smooth maneuver and rapid maneuvering can be obtained according to the prior knowledge respectively.Equivalent system model modeling noise statistical characteristics are expressed as the sum of the two modeling noise statistical properties.And the weights can be calculated by the likelihood function of the corresponding sensor information.The calculated equivalent system model runs in parallel as the state propagators.And the two state propagators chi-square test algorithm based on the improved interative multi-models is designed.The sensor information with the minimum fault detection value is used to update the weights of the modeling noise statistic characteristic.A multi-sensors fault-tolerant algorithm based on improved interactive multi-models is given in detail combined with the fuzzy adaptive filtering algorithm.The simulation results show that the algorithm can detect and deal with different kinds of faults in the measurement subsystem when the SINS accuracy is not very high.(4)Due to the strong nonlinearity as the use of the micro electro mechanical system(MEMS)-SINS in the integrated navigaiton system,it is difficult to deal with the multi-sensors fault-tolerant fusion with the extended Kalman filter(EKF)algorithm.To solve this problem,a multi-sensors fault-tolerant fusion algorithm combining Unscented information filter is proposed.And the multi-sensors federated filter formula based on Unscented information filter is derived.A fault-tolerant integrated navigation algorithm combined with the two state propagators chi-square test algorithm and the Unscented information filter algorithm.The reference state information output by the SINS model and the measurement information of each sensor constitute the local Unscented information filter subsystem.The developed fusion algorithm doesn't need linearization,so the linearization error is avoided.Then the fuzzy logic fault detection algorithm is adapted into the information fusion algorithm.The subsystem validation probability is calculated through the Chi-square test.And the fault subsystem can be isolated by the subsystem validation probability.The simulation results show that the proposed algorithm can effectively detect and deal with different types of faults in the MEMS-SINS integrated navigation system and improve the system tolerance of fault.(5)In order to further verify the effectiveness of the multi-sensors fault-tolerant integrated navigation algorithm proposed in this paper,a SINS/BDS/GPS/CNS integrated navigation applied system in vehicle tests is designed.The SINS/BDS/GPS/CNS integrated navigation system simulation platform is constructed.The satellite navigation data is obtained by the high precision integrated navigation system SPAN-CPT.The high precision attitude data of SPAN-CPT is used to simulate the measurement data obtained by the CNS.Then the vehicle tests were carried out and the experimental data of integrated navigation system is collected.Finally,we run the algorithm proposed in this paper with the collected data to verify the validity of the algorithm.