BP神经网络在材料领域中的应用(综述)

The Applications of BP Neural Networks in the Material Field

成分-结构-性能之间的关系始终是材料科学研究的主题,传统的试错法等经验或半经验的材料研究方法造成了资源、人力和时间上的极大浪费,因此需要从理论上解决材料设计、评价、预报等方面问题.人工神经网络是具有在线学习、记忆和分析推理功能能力的数学方法,它能够获得输入与输出之间的相互关系.其中BP神经网络结构简单、理论研究比较成熟.在材料研究领域中,BP神经网络已用于材料性能的研究与预测,复合材料工艺参数优化和预报,以及对金属的腐蚀研究等方面.

The research of the relations among composition, structure, and properties is a subject throughout the material science. The traditional trial - and - error empirical or semi - empirical materials research methods waste a lot in the resources, the manpower and the time. So the problem of design, evolution, and forecast of materials should be solved theoretically. Artificial neural networks can study online, recall, analyze and infer the mathematic model, and obtain the relationship between input and output. BP neural network has simple structure and is a mature method in the fundamental research. In the material research field, the BP neural network has been used in the research and forecast of material properties, optimization and forecast of composite materials processing parameter, and the research of the metal corrosion and so on.