用人工神经网络法预测含Re镍基单晶高温合金中TCP相的含量

根据大量含Re镍基单晶高温合金试验所测TCP相含量的数据,采用一种先进的人工神经网络方法建立运算模型,对合金的TCP相含量进行预测,并与报道所用回归方法进行了比较。结果表明,所建网络模型能够比较准确地预测含Re镍基单晶高温合金中TCP相的含量。将正交分析与网络预测相结合,获得几种主要合金元素对TCP相含量的影响顺序。

Artificial neural network prediction for the amount of TCP phase of Re-containing nickel base single crystal superalloys

Based on an advanced neural network method and a large number of data from Re-containing nickel base single crystal superalloys with a considerable amount of TCP phase, an artificial neural network model was constructed to predict the amount of the TCP phase for Re-containing nickel base single crystal superalloys. The results show that the amount of the TCP phase can be more accurately predicted by the present neural network method than the reported conventional regression method for the Re-containing nickel base single crystal superalloys. Therefore, it can be used as an assessment tool for the determination of the amount of the TCP phase in designing Re-containing nickel base single crystal superalloys. The sequence of influence of alloying elements on the TCP phase amount present in the superalloys investigated is analyzed combined with the use of statistical orthogonal tests and neural network prediction.