A Frequency Prediction Method for Frequency-Agile Radar Based on LSTM

Abstract

The tuning parameters of the frequency-agile radar of rotary-tuned magnetron are time-varying. By detecting self-calibration error, the conventional prediction method determines that the parameters have been changed and then start to accumulate data all over again, during which the data is required, and the error is large. In this paper, the frequency variation of the rotary-tuned magnetron frequency-agile radar with various parameters changing is predicted in real time by using the long short-term memory network. Compared with the existing work, the proposed method not only improves the prediction accuracy, but also has strong adaptability to the time-varying frequency law of multiple segments of data. Therefore, it has better application value and stability. The performance of the proposed algorithm is proved by simulation data experiment.