A Data-Driven Approach to Fatigue Life Characterization in A205 Aluminum Alloy: Additive Manufactured and Cast

In this study, we address the longstanding challenge of time-consuming and labor-intensive fatigue characterization experiments in laboratories. While scientists have developed a series of fatigue models to convert experimental data into analytical or empirical equations fitting S-N curves, the ongoing quest for a universal fatigue model persists. The intricate nature of the fatigue process and the inadequacy of the experimental data contribute to the complexity of this pursuit. Recognizing these challenges, our study proposes a novel machine-learning strategy to expedite the characterization of A205 Aluminum alloys fabricated by Additive manufacturing and casting processes, followed by different post-heat treatments. We adhere to standard operating conditions for stress-controlled uniaxial fatigue tests performed at room temperature. To guarantee dependability, the fatigue studies are conducted three times at each stress amplitude. The collected data undergoes a pre-processing step before feeding into machine learning models. This step is crucial for resolving discrepancies in fatigue data, removing outliers, and ensuring a consistent format. Subsequently, we select the most relevant mechanical features based on their effect on the target variable (fatigue life) using methods such as backward feature selection and then create input and output data matrices. Finally, we develop an efficient machine learning model to estimate the fatigue life of target grades concerning the input variables. Validation of the developed model is performed using the initially separated validation data. Our developed models demonstrate the capability to estimate fatigue life for these alloys under various stress levels, offering acceptable accuracy and reliability. This approach effectively minimizes the necessity for costly and laborious experiments, holding the potential to revolutionize industry practices by automating the fatigue design and characterization of Aluminum grades. It aligns with the principles of sustainable design and manufacturing and the objectives of Industry 4.0.