Prediction of Battlefield Resource Performance Trend Based on Deep Learning

This research is dedicated to establishing an accurate trend prediction model to achieve a global grasp of battlefield resources and to better respond to battlefield contingencies. This paper addresses the problem of insufficient spatial correlation for time-series battlefield resource performance trend prediction, and proposes a method that integrates temporal and spatial correlation. The method uses Pearson correlation coefficient (PCC) to construct a correlation matrix, and combines graph convolution (GCN) layers with long short-term memory (LSTM) networks to incorporate spatial correlation between nodes into the time series. Experiments demonstrate that the combined model can better incorporate spatial correlation compared to using only LSTM networks, and is more accurate and applicable in predicting trends in battlefield resource performance.