| Fault-tolerant integrated navigation system based on MEMS technology shows apparent advantages due to its small size,light weight and low power consumption.Furthermore,it is especially suitable for some specific applications,in which it could overcome some common problems in previous design,such as low fault-tolerance,weak reliability and poor security.Therefore the fault-tolerant integrated navigation system attracted lots of attention recent years.In this research we developed an engineering prototype of MEMS-SINS/GPS/ magnetometer/barometer integrated navigation system(hereinafter MEMS-SINS unified shorthand for MSINS)and arranged relevant investigation including sensor error analysis,identification and calibration,multi-source information fusion,integrated navigation federated filter design,adaptive kalman filter design based on genetic algorithm and integrated navigation faulttolerant scheme design involved in the development process,aiming to improve the accuracy and reliability of MEMS integrated navigation system.Main work:1.The basic principles of MEMS sensors(gyroscope,accelerometer,magnetometer)in integrated navigation system are introduced,then sensor errors are distinguished according to deterministic and random errors.We derived the deterministic error compensation formula of MEMS gyroscope,MEMS acceleration and MEMS magnetic strength.And the corresponding calibration method were established in the meanwhile.Furthermore,we completed the derivation of Allan variance sub-item formula according to the characteristics of the sensor and realized the random error identification of the MEMS gyroscope and accelerometer via Allan variance method.The corresponding random error model was established in the final.2.We derived the coarse alignment and fine alignment formulas in static base condition.Design of the inertial navigation system and integrated navigation system was completed then.The accuracy of algorithm was well verified in our MSINS coarse alignment,fine alignment,and track simulation experiments.In the following work,where fine alignment was compared to coarse alignment,the pitch angle error was reduced from 20.61 " to 11.97",the roll angle error was reduced from-20.57 " to-13.18",and the azimuth angle was reduced from 8.88° to0.82°.Furthermore,MEMS sensors show great advantages in coal mine environment where the navigation error of our Navigation System(ranging accuracy<2%,direction-finding accuracy<3 °,height accuracy < 4m)was well controlled.Combined with the developed engineering prototype,the pedestrian autonomous navigation algorithm based on MSINS/zerospeed correction algorithm/scene information constraint assistance was tested in a real coal mine.Test results showed great performance,which exactly solved the technical problems of land and energy supervision engineering.3.With the combination of multi-source information fusion technology,we established an error models of strapdown inertial navigation system(MSINS),global positioning system(GPS),magnetometer and barometer based on MEMS technology.In a comparison work between different Federated filters,we analyzed fault tolerance of these filters and designed a reset-free federated filter algorithm based on local feedback correction.Then the correctness of the algorithm was verified by different experiments,including ground sports car test,trajectory simulation and flight test.The results show that the MEMS integrated navigation algorithm designed in this paper has high navigation and positioning accuracy.The integrated navigation system attitude error angle was less than 0.5 The azimuth error was less than 1.5°,the speed error was less than 0.2m/s,and the position error was less than 10 m,10m,and 5m in the three directions of longitude,latitude,and altitude,respectively.4.Aiming at the problem that the measurement noise changes in the system,in which the filtering accuracy could be reduced or even diverged sometimes,we designed an adaptive Kalman filter algorithm based on genetic algorithm.Then the adaptive fast optimization of the measurement noise variance matrix R of the Kalman filter was realized as result.First,we analyzed both advantages and disadvantages of genetic algorithm and adaptive Kalman filter algorithm.A design scheme of adaptive Kalman filter based on matching value(Do M)was proposed latter.Secondly,we built a global real-time optimization based on genetic algorithm and an adaptive real-time and efficient Kalman filter algorithm based on matching value.Finally,the integrated navigation algorithm was simulated by the standard Kalman filter and the adaptive Kalman filter based on genetic algorithm.Our results proved that the method could effectively guarantee the accuracy of filter and reduce the perturbation caused by noise changes,thus ensuring the accuracy of the integrated navigation algorithm.5.In order to improve the fault tolerance and reliability of the integrated navigation system,a sensor-level fault detection method and a residual system-level fault detection method were proposed.First,we designed a state signal serial detection algorithm by wavelet analysis and modulus detection method,which was exactly suitable for fault-tolerant integrated navigation system with many sensors or random failure.Thus we realized the fast and accurate simulation verification of typical sensor failures.Secondly,a residual fault detection algorithm with integrated navigation subsystem and safety period observation was established on the basis of the non-reset federated filter.In our design once an alarm is found,the fault subsystem will be isolated.After the alarm is released,a periodic safety observation will be made to determine the acceptance of original faulty subsystem output.Therefore simulation verification of the typical GPS signal mutation fault is completed.Finally,a GUI interface based on Simulink’s integrated navigation system was built to verify the effectiveness of the sensor-level and system-level fault diagnosis,isolation,and reconstruction simulation algorithms. |
PDF Full Text Request