【作者】 闫国凤；

【导师】 武建军；

【作者基本信息】 河北工业大学 ， 材料学， 2007， 硕士

【摘要】 由于化学成分、加工工艺和材料性能之间的内在规律目前尚不很清楚,往往需要通过大量的实验才能确定新材料的组成及加工工艺。计算机建模大大缩短了新材料、新工艺和新设计从实验室转移到生产现场所需的时间。本文吸取了人工神经网络和专家系统的优点,基于人工神经网络技术建立了铁基耐蚀合金专家系统:设计了预测程序和优化程序,完善了BP神经网络学习程序,并设计了基于人工神经网络专家系统的人机界面。利用该系统软件,可以完成数据的预处理、样本库的扩充和修改、网络的学习、预测和优化。利用所建立的系统,预测了铁基耐蚀合金中Si,C,Mn各元素含量对腐蚀速率和抗弯强度的影响,得到了与实验数据相吻合的结果。利用所建立的系统,优化了铁基耐蚀合金的成分。实验表明,人工神经网络专家系统具有很好的优化能力,该项技术为合金成分设计与性能预测提供了有效方法。更多还原

【Abstract】 The composition and processing technology of new materials is often determined by a lot of experiments because the relation among the chemical composition, processing and properties of materials is not very clear. Computer modeling greatly shortens the development period of the new materials, new processes and new designs.An expert system based on artificial neural network is built. The prediction program and the optimization program of the artificial neural network expert system are designed. The Back Propagation neural network learning programming is improved. The artificial neural network expert system man-machine interface is designed. This system software can be used to the pretreatment of data, the expansion and revision of sample, the prediction and the optimization of properties of ferrous alloy.Effects of Si, C, Mn on the corrosion rate and bending strength of Fe based corrosion-resistance alloy are predicted with the system. The composition of Fe based corrosion-resistance alloy is optimized by the use of the system. Experiments show that the experimental data is closed to results predicted by the expert system. The artificial neural network expert system has a good compatibility in alloy design and performance prediction.更多还原