Abstract: The negative Poisson’s ratio (NPR) honeycomb is a novel type of metamaterial with excellent mechanical properties. Among them, the Double Arrowed Honeycomb (DAH) with a negative Poisson’s ratio stands out for its exceptional performance. Based the DAH, a new NPR honeycomb material named Circular Double Arrowed Honeycomb (CDAH) was proposed. By replacing the straight edges of the DAH with double circular edges, the CDAH improved the energy absorption by 71% and enhanced the structural impact resistance. The plateau stress of the CDAH under different impact velocities was studied by numerical simulation, and the influence of cell geometry parameters and impact velocity on the plateau stress of the CDAH was analyzed. The results show that the relative density significantly affects the plateau stress of the CDAH. Based on this, an artificial neural network (ANN) machine learning model was proposed to reveal the complex nonlinear relationship between the honeycomb cell structure parameters and mechanical performance indicators. Compared with the empirical formula for predicting the plateau stress of honeycomb materials, the proposed ANN model can predict the plateau stress of CDAH more quickly and accurately, with an average relative error of only 3.82%, while the average relative error of the empirical formula is 45.71%. This study provides a new method to predict the plateau stress of honeycomb materials including NPR honeycombs, which is helpful to accelerate the design process of negative Poisson's ratio honeycomb materials.