Development of AI Platform Based on Machine Learning for Separation Process

This study developed an AI platform with a machine learning model for optimizing the separation process, which is a representative process in the chemical plant, and optimized the commercial process by applying the AI platform. Although the operating conditions for optimization were presented in the existing process using the theoretical model, it is difficult to apply them to the commercial process because the theoretical operating conditions and the actual operating conditions are not the same. In this study, a machine learning model based on actual process data and the AI platform was developed to predict major operational variables of the process. The AI platform consists of three steps: development, validation, and application. The process data collection, parameter extraction, and selection of learning methods are specified during the development step, the validation step is improvement through model validation and hyper-parameter adjustment, and finally applied in the commercial process by the software program in the application step. The AI platform was applied to the commercial mixed butane separation process using the distillation column. The AI platform was applied to a distributed control system (DCS) that controls the process and operated the process with optimal operating conditions. When the AI platform applies to other similar chemical processes, it will be the foundation for building smart factories in the chemical industry.