【摘要】 用人工神经网络和化学键参数相结合，总结算化物系相图的规律。据此可判断这类相图的中间化合物的化学配比是同分熔化还是异分熔化，并能对中间化合物的熔点或分解温度作粗略估计，估算结果与实验符合较好。本文还从结构化学角度探讨了这些经验规律的实质。更多还原

【Abstract】 Artificial neural networks together with chemical bond parameters can be used to find the regular-ities of the characteristics of oxide phase diagrams.It can also be used to predict the chemical stoichiometry of in-termediate compounds,their melting types(congruent or incongruent melting),and the melting point or decom-position temperature of intermediate compounds. The results of the prediction are usually in agreement with ex-perimental data.The physical significance of these empirical relationships are discussed based on the principles of structural chemistry.更多还原