State Estimation for GPS Outage Based on Improved Nonlinear Autoregressive Model

The accurate and immediate state estimation is essential in control of navigation system. The traditional Global Position System/ Inertial Navigation System (GPSIINS) integration may be invalid without location information provided by GPS during outage periods. A state estimation framework is proposed in this paper to obtain GPS location information during satellite outages. Firstly, a Nonlinear AutoRegressive Moving Average model with eXogenous input (NARMAX) is designed to characterize the outage periods. Secondly, the integration of Least Square Support Vector Machine (LSSVM) with NARMAX is implemented by using LSSVM to identify NARMAX parameters. The model is trained at first based on present planar angular information and historical location increment feedback provided by GPS with estimation error feedback before outage periods. It switches to predictor mode during outage periods without GPS information. Also, time-serial data are analyzed in NARMAX-LSSVM model to excavate the data features in time dimension. An experiment was conducted to verify the proposed model with multi-step prediction. The results were compared with other traditional methods to show its improvement and validation in state estimation.