Estimation of Guidance Law Parameters Using Recurrent Neural Network

초록

Proportional navigation guidance (PNG) and impact angle control guidance (IACG) are generally used for terminal guidance of missiles. If the guidance law and parameter for a target are known, the future trajectory can be predicted using the position and velocity information obtained from radar. Therefore, estimating the guidance law and its parameter is important, and it should be rapidly determined because missile engagements occur very quickly. This paper proposes a real-time machine learning-based estimation model. The trajectory characteristics, velocity, line-of-sight (LOS) rate, and other variables under various flight conditions are analyzed. Then, critical variables for estimation are identified, and a learning model is designed based on these variables. To utilize the time series variable characteristics, a type of recurrent neural network (RNN) called Bidirectional Long Short Term Memory (BiLSTM) is used. Finally, the trained results are validated through numerical simulation. © ICROS 2023.

키워드