Creep Fracture Life Prediction of 9%Cr Heat‐Resistant Steel Based on the Back‐Propagation Artificial Neural Network Considering Microstructure Characteristics

This study proposes a creep‐life prediction model based on a back‐propagation artificial neural network (BP‐ANN). The model predicted the minimum creep rate by optimizing microstructure characteristics, including martensite lath width, sub‐grain, precipitates (M 23 C 6 and Laves phases), and dislocation density. The BP‐ANN‐based predicted creep life of 9%Cr heat‐resistant steel is compared with the traditional time‐temperature life prediction method and the machine learning support vector regression model. The results show that the BP‐ANN model has better accuracy and performance. The % error between the predicted and the experimental creep life is 5% to 10%, primarily distributed in the two‐factor band line (2% to 5%). The BP‐ANN model gives the lowest mean absolute error (0.9926), root mean square error (29.411), average absolute relative error (4.7%), and multiple coefficient of determination ( R 2 0.96693). The results show that the BP‐ANN model improved the creep fracture life prediction of 9%Cr heat‐resistant steel.