Navigation System Fault Detection And Diagnosis, And Theoretical Issues

With the development of modern control theory, computer and aerospace technology, the requirements for reliability and accuracy of navigation systems is higher and higher. The error compensation and fault-tolerant technique has become the important measures to improve them. In this dissertation, the Fault Detection and Diagnosis(FDD) methods of navigation systems and the error compensation technique of noncommutativity errors, i.e., coning and sculling errors for Strapdown Inetial Navigation System (SINS) under the high frequency movement are researched. The statistic models of the error coefficients of Strapdown Inetial Measurement Unite(SIMU) with few test data, such as the calibration factor, configuration error of gyroes and accelerometers, and so on, are also be investigated under the entrust of the second artillery engineering college, Xi'an.Generalized Likelihood ratio Test (GLT) and Optimal Parity-vector Test (OPT) methods to detect and diagnose soft faults from sensor redundant measurement system in SINS are researched very carefully from a point of view of their effectiveness to detect and isolate soft faults. Firstly, the relationship between the fault isolation function of OPT and that of GLT is investigated with respect to the redundant system configured by the same precision, the same type single axis sensors. And according to their simulation results, the problem of using OPT method to detect faults is pointed whose reason is analyzed as well. It is very important that we corrected the fault detection tactic of OPT method. Besides, with respect to a general sensor redundant system, we proved their detection and isolation effectiveness are all the same providing that the fault input matrix is some times of the noise input matrix. In order to compare the effectiveness of GLT method with OPT method applied to any sensor redundant measurement system, the FDD problem for a two-axes gyroes redundant system is discussed and more simulations are made. Many satisfied valuable conclusions are summarized.For the dynamic system with the measurement equation and state equation, the robust FDD approach is discussed. At first, the robust design principle and its optimal performance criterions are introduced, and that the OPT method is extended to detect and isolate faults of integrated navigation systems. The optimal parity residual vectors that are maximally sensitive to special navigation system fault and minimally sensitive to other navigation system fault and measurement noise are designed. Finally, the robust FDD method and general FDD method are all applied to integrate navigation systems. By means of the Monte-Cairo simulations, the FDD theory and approaches above are verified.In terms of the rotation vector differential equation, the source of thenoncommutativity errors including coning and sculling errors is analyzed and coning algorithms are given for SINS in high dynamic environment. A generic equivalency between strapdown inertial navigation coning and sculling integrals is introduced based which the general form of sculling compensation algorithms are obtained according to coning compensation algorithms. Furthermore, coefficients in them get optimized under sculling base motion and can be solved in a matrix equation. Simultaneously, we get the drift formula of the optimal algorithms. It is both convenient and precise to use this method to compensation sculling error instead of repeat deduce work .Based on the modern statistic theory, comparing Bootstrap method or random weight method with data fusion theory, the probability statistical characteristics of error coefficients of SIMU including gyroes and accelerometers are analyzed. The quality weights of prior information that shows the consistency or difference between prior information and current information are defined. Then, the Bootstrap and random weight fusion estimation method for small sample of multiple sources with few test data is presented. The prior p.d.f.s of error coefficients for some type SIMU are obtained . At last, some conclusions and advices are given.