Use of Machine Learning Techniques in the Modeling of an Industrial Reaction System

Abstract

Purpose: The objective of the research is to address the industrial production of hydrogen and develop the modeling and simulation of hydrogen-related industrial reaction systems using Machine Learning techniques. Theoretical Framework: The research explores the innovation and promise of Machine Learning techniques in modeling industrial reaction systems, enabling the creation of flexible and adaptive models to deal with complexities in industrial processes. Method/Design/Approach: The method involves the application of machine learning methods, such as linear regressions and the kriging or Gaussian process method, to develop metamodels that analyze the steps of an industrial reaction involving hydrogen. Results and Conclusion: The results of the training and analysis have achieved satisfactory outcomes, with expected values assessed through constraint parameters for each output variable. Research Implications: The research aims to improve real-time prediction accuracy, process variable control, and early fault detection, resulting in greater sustainability and economic efficiency in the industry. Originality/Value: The values underpinning the research include the promotion of technological innovation, operational efficiency, and environmental sustainability, as well as valuing quality, safety, social responsibility, continuous improvement, global competitiveness, and regulatory compliance in the industry.