The Research On New Terrain Elevation Matching Approaches And Their Applicability

Terrain-aided navigation (TAN) is the technique of using variations in terrain elevation below aircrafts to produce a position estimation for bounding errors in Inertial Navigation System (INS). TAN is an autonomous, all-weather, and low-altitude navigation technique. Like satellite navigation and INS, TAN has many important applications in navigation of long-range, autonomous, and precision-strike weapons such as modern aircrafts, unmanned aircraft vehicles(UAV), and cruise missiles. Terrain Elevation Matching (TEM) strategy is one of the key techniques in TAN, and many solutions to TEM have been applied to fighter planes and cruise-missiles. However, these solutions only work well for some special types of terrains, because they need excessive time to prepare a flight course and may have false fixing, which limit applications of TEM systems. Therefore, studying high performance TEM algorithms with high applicability has become an important research topic in TAN area. This dissertation studies new TEM algorithms and their applicability in the context of long-range UAV application.Firstly, the formalized model of TEM is investigated, the factors affecting the TEM system performance are analyzed, and the ways to enhance the performance of TEM system are proposed. The TEM problem is essentially a nonlinear state estimation problem. The nonlinear estimation method with probability criterion can directly deal with nonlinear model, and can avoid errors introduced by linear approximation of nonlinear model. Theoretically, this method shows superiority in solving TEM problems.Secondly, to improve the performance of the current TEM algorithms, three TEM algorithms based on probability criterion are studied, which are suitable for different applications.Considering the false-fix problem in Terrain Contour Matching (TERCOM), a new batch approach to TEM problem, Probability Data Associate Filtering(PDAF) based TEM algorithm (PDAFTEM) is proposed. The simulation shows that PDAFTEM has better fault tolerance to Line Error Probability(LEP) of Digital Terrain Elevation Database(DTED) and lower probability of false-fix than TERCOM. In case of no high quality DTED available, PDAFTEM is applied prefer to TERCOM.Considering that Sandia Inertial Terrain Aided Navigation(SITAN) may divergence when there exists large initial position errors in INS, this dissertation proposes and formulizes a recursive Bayes estimate model, which takes INS position error as its state variable and height as the observation measurement. A new recursive approach applicable to short-time continuous TEM cases, Recursive Bayes Estimation based TEM algorithm (RBETEM) is proposed.Compared with SITAN, RBETEM presents two advantages: 1) when there exists large initial position errors in INS, the convergence of RBETEM will not be influenced 2) RBETEM has no restrictions on the distribution of noise.In addition, a new recursive approach applicable to long-time continuous TEM cases, Hidden Markov Model(HMM) based TEM algorithm (HMMTEM) is also proposed. A Markov chain model for INS position errors and a HMM model for TEM is established. Compared with RBETEM, HMMTEM takes into account the drift of INS errors between the current and last sampling time. The simulation indicates that estimation errors of HMMTEM are less than that of RBETEM and SITAN. However, this is obtained at the cost of higher computational burden.Thirdly, to make a better use of last measurement information, a Fading Memory based TEM (FMTEM) algorithm is proposed to deal with the shortcomings of TERCOM method, which is sensitive to direction-error in INS. FMTEM algorithm uses a new correlation-value, which is calculated recursively, to assign different weights to measurement according to its duration. The simulation indicates that FMTEM outperforms TERCOM when INS direction-errors are larger.At last, to analyze and evaluate the performance of TEM algorithms and their applicability, a communication model for TEM and a concept of Terrain Navigation Information Content(TNIC) are investigated from the view of information theory. Some metrics related to TNIC are investigated: terrain elevation standard deviation, terrain elevation entropy, terrain Fisher information content, and terrain code distortion. The influence of TNIC on the TEM performance is analyzed theoretically, and is validated by a set of Monte-Carlo simulations. All simulation results are processed using step regression analysis to investigate the polynomial relations between the terrain characteristics and the errors of previous TEM algorithms, which may be used to guide the applications of previous TEM algorithms. i...