Hybrid-system-based multiple-model approach for transfer alignment with dynamic flexure in IIN

Transfer alignment is the main technique to satisfy the local state estimation function requirement of an integrated inertial network (IIN), and its performance is severely affected by the vibration and flexure of local inertial navigation system (INS). Transfer alignment usually uses linearized differential equations to describe the error propagation process, and it takes the differences between the navigation (attitude, velocity, and position) parameters of the master INS and a slave INS as measurements. For the problem of dynamic flexure, traditional approaches use n^th-order Gaussian Markov (GM) models to describe the dynamic characteristics of flexure. After analyzing the three main limitations of the traditional approaches, this paper then proposes to use a hybrid system to describe the transfer alignment process with dynamic flexure. Then the multiple-model (MM) approach is used to estimate navigation (attitude, velocity, and position) errors as well as flexure angles. Simulation results demonstrate the effectiveness of the proposed model and approach compared with the traditional transfer alignment with the second-order GM model of flexure angles.