Radar HRRP target recognition method based on spectrogram transformation and attention mechanism recurrent neural network

Abstract

The invention discloses a radar HRRP target recognition method based on spectrogram transformation and an attention mechanism recurrent neural network. The method comprises the steps of: firstly, performing preprocessing to reduce sensitivity in an HRRP sample, and performing spectrogram transformation on the sample; then establishing a dynamic adjustment layer, and adjusting the spectrogram; thenadjusting the importance degree of each time point sequence in the spectrogram through an importance network, modeling time sequence correlation through a bidirectional stacking RNN, extracting high-level characteristics of the time sequence correlation, finally adjusting the importance degree of a network hidden layer state by adopting a multi-level attention mechanism, and carrying out target classification through softmax.

Description

technical field [0001] The invention belongs to the field of radar target recognition, and in particular relates to a radar HRRP target recognition method based on spectrogram transformation and attention mechanism cyclic neural network. Background technique [0002] With the rapid development of science and technology, radar target recognition technology plays an increasingly important role in military defense and future wars. For radar target recognition, the echoes of high-resolution broadband radar contain extremely valuable structural information for classification and recognition, such as the radial size of the target, the distribution of scattering points, etc., and have broad engineering application prospects. The echo of the high-resolution broadband radar is also called the one-dimensional high-resolution range profile (High Resolution Range Profile, HRRP) of the target. Therefore, the radar automatic target recognition method based on HRRP has gradually become a ...

AI Technical Summary

Problems solved by technology

The first two methods directly perform feature extraction and modeling on the overall envelope information of HRRP, ignoring the sequence correlation between HRRP distance units that can reflect the physical structure characteristics of the target.

Although the third method is based on serial correlation modeling, there are still the following problems: (1) The original time-domain segmentation method is used for the local strength information of HRRP, and the obtained features are highly redundant. , which brings difficulties to the follow-up RNN (Recurrent Neural Network) modeling; (2) The distance unit with a small range may contain some highly separable features, but these features are rarely used ; (3) The one-way RNN can only use the structural information at the current moment and before the current moment when predicting, and cannot make good use of the overall structural information a priori contained in HRRP

Images