COMBINED CORNER REFLECTOR ARRAY INTERFERENCE RECOGNITION BASED ON IMPROVED TCN AND DGRU

Combined corner reflector is an important passive interference faced by radar in sea detection. Combined corner reflector recognition is the key for radar to achieve accurate countermeasures. At present, most combined corner reflector recognition methods rely on explicit feature extraction with physical connotations. For complex targets under the condition of low signal-to-noise ratio (SNR), explicit features often cannot cover the full attitude when the number of samples is severely limited, resulting in a decrease in recognition accuracy. This paper proposes a combined corner reflector array interference recognition method based on improved Temporal Convolution Network (iTCN) and double-Gate Recurrent Unit (DGRU). TCN is used to improve the ability to extract deep features of the High Resolution Range Profile (HRRP) sequence. Improving the original activation layer structure and using the Swish activation function to place the activation layer in front of the weight layer, which provides greater nonlinear mapping capabilities for data, thus improving the anti-noise capability of the network. Combined with the DGRU network, deep and implicit features of HRRP sequence are extracted to make full use of ship attitude invariance. Experiments show that this model is capable for identifying combined corner reflector array interference under small samples and low SNR.