Sheet Metal Forming in a Virtual Reality Environment using LS-DYNA and Neural Networks

Ashwini S. Gokhale Manufacturing process simulation using finite element techniques has immensely contributed to ensuring the success of concurrent design methodologies. However, Finite Element Methods (FEM) is computationally expensive and consequently unsuitable for design and manufacturing optimization in a production environment. In this research, a coupled Artificial Intelligence (AI) and FEM technique was developed to simulate and predict process response to changes in part design. Generic process models of part families are developed using Artificial Neural Networks (ANNs) and FEM. The generic models are used to predict the response of the manufacturing process to variations in geometric, material and process parameters, in real time. The predicted results are graphically displayed in a Virtual Reality environment. Standalone software VRForm was developed based on this methodology. VRForm can be used to optimize component, tool and process designs.