论文部分内容阅读
固相生成金属间化合物的液相分层二元系在Ｍｉｅｄｅｍａ的Δ—Δｎｗｓ１／３图中，分布在Δ＞１．３的区域，然而在该区域混杂有固相没有金属间化合物的液相分层二元系。为了消除这种混杂现象，应用扩展的Ｍｉｅｄｅｍａ合金元胞模型研究了金属液相分层二元系固相能否生成金属间化合物的规律。在由原子参数Δ与Δｎｗｓ１／３及ΔＺ张成的多维空间中，上述两类二元系各自分布在特定的区域。据此结果，以Δ，Δｎｗｓ１／３，ΔＺ，ＲＡ／ＲＢ作为人工神经网络的输入特征量，采用误差反向传递算法，利用经已知样本集训练的人工神经网络对上述二元系的会溶温度和偏晶温度进行预报。预报结果与实测结果符合较好
The liquid-phase stratified binary system of solid-phase-forming intermetallic compounds is distributed in the area of Δ> 1.3 in the Δ-Δnws1 / 3 diagram of Miedema, however in this area there is a mixture of a solid phase with no intermetallic Liquid-phase stratified binary system of compounds. In order to eliminate this confounding phenomenon, the extended Miedema alloy cellular model was used to investigate whether the metal-liquid layered binary system solid phase can generate intermetallic compounds. In the multidimensional space spanned by atomic parameters Δ and Δnws1 / 3 and ΔZ, the above two types of binary systems are each distributed in a specific area. Based on the above results, using Δ, Δnws1 / 3, ΔZ and RA / RB as the input features of artificial neural network, the error backtransmission algorithm is adopted, and the artificial neural network Of the solution temperature and the deviation of the temperature forecast. The forecast result is in good agreement with the measured result