Data-Driven Modeling and Robust Optimization of the Ammunition Ramming Mechanism Based on Artificial Neural Network

Aiming to address the challenges of strong nonlinearity and lengthy experimental and simulation prediction time in the ammunition ramming process of ammunition ramming mechanisms, this study proposes a data-driven modeling method based on artificial neural networks. To begin with, the composition of the ammunition ramming mechanism is analyzed, and a dynamic model of the ammunition ramming process is established. The model’s accuracy is verified through experiments. Subsequently, a data-driven model for the ammunition ramming process is developed using artificial neural networks and simulation data. The effectiveness and accuracy of this model are confirmed through numerical examples. Finally, based on the data-driven model, a robust optimization design is conducted for the ammunition ramming process. The obtained results demonstrate a significant improvement in efficiency and accuracy of both the data-driven model prediction and the optimization design without compromising accuracy. Consequently, this method offers valuable theoretical support for the rapid design of ammunition ramming mechanisms and other equipment.