【摘要】 应用原子参数 人工神经网络方法研究了过渡元素二元合金相图中的有限固溶度问题.选择了原子体积、平均族数、元素熔点等影响固溶度的原子参数作为人工神经网络的输入值,使用已知的过渡元素二元合金相图的固溶度作为输出值,训练人工神经网络,应用"留一法"估计了人工神经网络的预报误差,并用训练好的网络对全部未知的过渡元素二元合金相图的固溶度进行计算机预报,得到了令人满意的结果.更多还原

【Abstract】 Limited solid solubility of transition elements in binary alloy phase diagram is studied by (using) atomic parameters and artificial neural network. The atomic volume, average group number and (element) melting point are selected as the inputs of ANN, and the solid solubility as the outputs. The ANN is trained by known data of binary alloy systems. The unknown solid solubility of the (binary) (alloy) (systems) is predicted with the trained ANN and the results are satisfactory.更多还原