Research and design of VR one-handed warhammer balance point based on CNN

To address the engineering challenges of dynamic balance optimization for single-handed warhammers in virtual reality (VR) interactive devices, this study proposes a hybrid design framework that integrates physical modeling with deep neural networks. Initially, a multi-stage motion trajectory model based on modified rigid body dynamics is developed; by incorporating an air resistance coefficient and a nonlinear deformation compensation term, a multi-dimensional balance evaluation system is established, encompassing impact energy loss, impact force distribution, and control stability. Building on this foundation, an improved convolutional neural network (CNN) multi-task learning architecture is designed to extract features from various warhammer parameters, thereby enabling the joint prediction of handle length and hammerhead mass. Experimental results indicate that the proposed model achieves mean absolute errors (MAE) of 3.92% and 3.53% on the simulation test set. By dynamically adjusting impact force and energy consumption thresholds, the framework effectively supports multi-objective optimization for weapon control and immersion. The research outcomes provide an interpretable theoretical model and a practical implementation pathway for the intelligent design of VR interactive devices, offering significant reference value for enhancing the physical realism of virtual combat scenarios.