Eun-Jung Song

Design of an Initial Fine Alignment Algorithm for Satellite Launch Vehicles

In this paper, an initial fine alignment algorithm, which is developed for the strap-down inertial navigation systems of satellite launch vehicles, is considered. For fast and accurate alignment, a simple closed-loop estimation algorithm using a proportionalintegral controller is introduced. Through computer simulation for the sway condition in the launch pad, it is shown that a simple filter structure can guarantee fast computational speed that is adequate for real-time implementation as well as the required alignment accuracy and robustness. In addition, its implementation results are presented for the Naro-1 flight test.

Error analysis of the ground based navigation test using extended kalman filtering

This paper considers the error analysis of the ground based navigation test of the strapdown type inertial navigation system. Using an extended Kalman filter, the biases in the inertial sensor are estimated in the static condition of the laboratory environment and the dynamic condition by the test car where another highly accurate instrument is carried to provide target states. The results show that the gyro bias estimates of the filter are within 0.2 deg/hr and the accelerometer biases are within 0.1 mg.

Performance Analysis of a Precise Explicit Guidance Algorithm for Space Launch Vehicles

This paper considers one of the explicit guidance algorithms, which has been proposed by Jaggers, to determine the closed-loop guidance algorithm for upper stages of a 3-staged space launch vehicle. Its commanded thrust vector is closer to the optimal solution when compared with that obtained by using the well-known Powered Explicit Guidance (PEG), which has been developed through the Space Shuttle program. Its performance is evaluated here by applying for guidance of the launcher during the second and third stages. Furthermore, to generate more precise guidance commands, it is attempted not to use the approximate formulas for the derivation of the original guidance law, and it is shown that performance is improved in comparison with the original.

Neural-Network Midcourse Guidance with Consideration of the Head-on Attack Condition

The objective of this study is to enhance neural-network guidance to consider the impact condition. Missile impact angle error, a measure of the degree to which the missile is not steering for a head-on attack, can have a significant influence on the final miss distance. Midcourse guidance using neural networks is employed to reduce the deviation angle from head-on effectively in the three-dimensional space. In addition, a coordinate transformation is introduced to simplify the three-dimensional guidance law and reduce training data for the neural network. Computational results show that the current neural-network guidance law with the coordinate transformation can be used to reduce the impact angle errors.

Real-time midcourse guidance with consideration of the impact condition

The objective of this study is to enhance neural-network guidance to consider the impact condition. The optimal impact condition in this study is defined as an head-on attack. Missile impact-angle error, which is a measure of the degree to which the missile is not steering for a head-on attack, can also have an influence on the final miss distance. Therefore midcourse guidance is used to navigate the missile, reducing the deviation angle from head on, given some constraints on the missile g performance. A coordinate transformation is introduced to simplify the three-dimensional guidance law and, consequently, to reduce training data. Computer simulation results show that the neural-network guidance law with the coordinate transformation reduces impact-angle errors effectively.

Decentralized Filters for the Formation Flight

Decentralized filtering for a formation flight instrumentation system by INS/GPS integration is considered in this paper. An elaborate tuning method of the measurement noise covariance is suggested to compensate modeling errors caused by decentralizing the extended Kalman filter. It does not require large data transfer between formation vehicles. Covariance analysis exhibits the superior performance of the proposed approach when compared with the existent decentralized filter and the global filter, which has the target-filter performance.