The full text of this item is not available at this time because the student has placed this item under an embargo for a period of time. The Libraries are not authorized to provide a copy of this work during the embargo period, even for Texas A&M users with NetID.
Machine Learning Based Prediction Models for Flammability Characteristics in the Chemical Industry
Abstract
The prediction models based on Quantitative Structure-Property Relationship (QSPR) and machine learning have been widely applied in the research. Flammability characteristics are critical information for the chemical industry to prevent, mitigate, and prepare for the potential process safety incidents. In the chemical industry, changes happen often, and sometimes employees are required to make timely decisions to maintain the safety and avoid affecting production. Machine learning based models could serve as accessible references because of high accuracy and reliability on its predictions. Here we show that the models on the basis of QSPR and gradient boosting could have excellent predictions for flammability characteristics of concern. With the application of Xgboost, Mordred, and RDkit libraries in Python 3, k-fold cross validation, and published experimental data, we tune the combination of hyperparameters to obtain the best QSPR models. According to the statistical assessments, all the models’ R^2 are higher than 0.9; thus, the evaluations indicate the good predictions. Moreover, with the use of empirical formulas, the applicability of the predictions on flammability characteristics is broadened to meet the needs in the chemical industry. With accessible, efficient, and reliable models to predict the changes, it makes the chemical industry be able to seek for the safety and production at the same time. Therefore, safety is no longer a choice to make, but a thing to do.
Subject
process safetyCitation
Li, Chi-Yang (2022). Machine Learning Based Prediction Models for Flammability Characteristics in the Chemical Industry. Master's thesis, Texas A&M University. Available electronically from https : / /hdl .handle .net /1969 .1 /198813.