GPS BASED ATTITUDE ESTIMATION OF AIRCRAFT USING NEURAL NETWORK AIDED KALMAN FILTER

摘要

This paper shows how a neural network can augment the Extended Kalman filter (EKF) in attitude estimation of the aircraft using GPS. The carrier phase of GPS is used to estimate the attitude information of an aircraft. The typical problem of attitude determination using GPS is a non-linear function of the attitude angles. Conventionally Extended Kalman Filter is used for attitude estimation. The adaptive capability of the Kalman Filtering is known to improve by incorporating the Neural Network in the normal Kalman fdter. A Radial Basis Function Network (RBFN) is subsequently employed to aid the EKF in attitude estimation. The study is extended to compare the error reduction and estimation improvement in attitude parameters under the influence of measurement and bias error. The performance of the filter during the transient phase of satellite change over is also analyzed.