Panoramic Convolutional Long Short-Term Memory Networks for Combat Intension Recognition of Aerial Targets

Aiming at the limitation that traditional methods for combat intention recognition of aerial targets are difficult to effectively capture the essential characteristics of intelligence information, we design a novel deep learning method, Panoramic Convolutional Long Short-Term Memory networks (PCLSTM), to improve the recognition ability. First, based on the characteristics of aerial target intelligence information, a panoramic convolutional layer is designed to extract the loosely coupled characteristics of intelligence information, and a time series pooling layer is designed to reduce the scale of neural network parameters on a large scale. Then, the temporal feature extraction capability of the LSTM layer and the depth feature mining capability of the traditional deep learning layer are combined to construct the PCLSTM neural network. Subsequently, the recognition performance of PCLSTM is analyzed by simulation experiments compared with standard deep net, convolutional neural network and LSTM network as benchmark models. Finally, PCLSTM was used to carry out simulation tests on different truncated data sets of original intelligence information, to analyze the optimal length of truncated data for different combat intention recognition. And then a reasonable aerial target combat intention recognition method is designed. The simulation results show that the method presented in this paper has theoretical significance and reference value for command decision-making.