Neural Networks in Material Modeling and Computational Mechanics

本研究将探讨一种基于神经网络的材料建模新方法。这种方法不同于传统的基于相对简单的数学模型的构建和使用经典参数识别技术的方法。相比之下，在神经网络建模中，材料行为的知识直接从观察到的实验行为中获取，并在学习、自组织神经网络的连接中进行编码。

This research will investigate a new method of material modeling based on the use of neural networks. This approach is different from the conventional methods that are based on the construction of relatively simple mathematical models and the use of classical parameter identification techniques. By contrast, in neural network modeling the knowledge of the material behavior is captured directly from observed experimental behavior and is encoded within the connections of a learning, self-organizing neural network.