Polymer materials have garnered significant attention due to their exceptional mechanical properties and diverse industrial applications. Understanding the glass transition temperature (Tg) of polymers is critical to prevent operational failures at specific temperatures. Traditional methods for measuring Tg, such as differential scanning calorimetry (DSC) and dynamic mechanical analysis, while accurate, are often time-consuming, costly, and susceptible to inaccuracies due to random and uncertain factors. To address these limitations, the aim of the present study is to investigate the potential of Simplified Molecular Input Line Entry System (SMILES) as descriptors in simple machine learning models to predict Tg efficiently and reliably. Five models were utilized: k-nearest neighbors (KNNs), support vector regression (SVR), extreme gradient boosting (XGBoost), artificial neural network (ANN), and recurrent neural network (RNN). SMILES descriptors were converted into numerical data using either One Hot Encoding (OHE) or Natural Language Processing (NLP). The study found that SMILES inputs with fewer than 200 characters were inadequate for accurately describing compound structures, while inputs exceeding 200 characters diminished model performance due to the curse of dimensionality. The ANN model achieved the highest R2 value of 0.79; however, the XGB model, with an R2 value of 0.774, exhibited the highest stability and shorter training times compared to other models, making it the preferred choice for Tg prediction. The efficiency of the OHE method over NLP was demonstrated by faster training times across the KNN, SVR, XGB, and ANN models. Validation of new polymer data showed the XGB model’s robustness, with an average prediction deviation of 9.76 from actual Tg values. These findings underscore the importance of optimizing SMILES conversion methods and model parameters to enhance prediction reliability. Future research should focus on improving model accuracy and generalizability by incorporating additional features and advanced techniques. This study contributes to the development of efficient and reliable predictive models for polymer properties, facilitating the design and application of new polymer materials.
Superalloys are a class of materials renowned for their exceptional ability to retain mechanical properties at elevated temperatures. Nickel superalloys, with a nickel content ranging from 38% to 76%, and iron–nickel superalloys (15–60% iron, 25–45% nickel) are extensively employed within the aviation industry due to their resilience in harsh operating environments. These components encounter extreme temperatures during operation, significantly impacting their tensile strength and melting point. Furthermore, high-speed rotation and abrasive conditions necessitate materials with superior hardness. Consequently, material modifications are crucial to ensure that gas turbine components meet their required properties. Machine learning (ML) and deep learning (DL) offer promising solutions for the design of materials with tailored tensile strength, hardness, and melting point properties. This study investigates the efficacy of direct and inverse machine learning models in predicting crucial material properties and composition, respectively. The model with the most favorable prediction accuracy is identified through the systematic variation of key parameters. The findings show that a fully connected feed-forward Artificial Neural Network (ANN) with three hidden layers using ReLU activation functions performs better than the other models. This capability is leveraged to modify the composition of INCONEL-718, successfully achieving significant enhancements in tensile strength (1592 MPa), hardness (152 HRB), and melting point (1665 °C).
The oil and gas industry is one of the world's largest and most influential energy contributors. All aspects involved in the operation of this industry are fundamental to be reviewed and managed correctly, especially by preventing or minimizing the failures that could occur. Uniform corrosion is the most common component failure mechanism that can cause failure in the oil and gas industry. The company's actions in managing and preventing the risk of this type of failure have a major role in the sustainability of the company due to the possibility of more significant impacts if the risk cannot be handled well, such as high inspection and handling costs, environmental impacts, and threats to work safety. In this study, the Dynamic Risk-Based Inspection (DRBI) method, which is a development of Risk-Based Inspection (RBI), is implemented to handle and analyze risks that are managed in real-time at each inspection period. Risk level analysis was carried out through data processing related to pipe thickness from the risk profile from the inspection results in 5 months using Igor and Rstudio software and calculating corrosion rates using the forward difference approach. Based on the analysis results, five risk levels of pipeline failure at PT. X due to uniform corrosion using DRBI was obtained, consisting of two medium risks and three medium-high risks. In contrast, only one risk level was obtained from the RBI method, namely medium-high. The risk value fluctuates greatly every month, causing the DRBI method to have a higher level of accuracy and the ability to detect potential risks in more detail than the RBI method.