Predictive Approach to Creep Life of Ni-based Single Crystal Superalloy Using Optimized Machine Learning Regression Algorithms

In this study, the continuation of the recent approach Divide-and-conquer Self-adaptive (DCSA) for predicting the creep life of different generations of Ni-based superalloys was investigated. The creep life of superalloys is the property that describes the sustainability of application and is most crucial to be determined to understand the creep behaviour. However, the factors that affect the creep life are significant in number which in return complicates the estimation techniques to determine the appropriate creep life. With the aid of Machine Learning (ML) inbuilt algorithms, the prediction results are highly efficient and precise with several impacting factors as input features. The dataset used in this study have been extracted from valid database resource and scaled using the Standard Scaler algorithm onto a reliable scale before impinging different regression algorithms onto it. Due to altering creep mechanisms in different generations, the art of using different regression will give maximum appropriate outcomes. Further validation for each optimized selected regression algorithm was done using validation techniques like R2 values.