Predicting Solution Color in Centrifugal Contactor Solvent Extraction
- Program: Applied Mathematics and Statistics
- Course: Other
Project Description:
In solvent extraction processes, accurate prediction of solution characteristics is essential for enhancing process control and operational safety. Traditional monitoring methodologies often fail to capture the complex dynamics of solvent interactions effectively. Our research addresses this gap by employing a novel application of machine learning (ML) algorithms, analyzing multimodal sensor data collected from a testbed equipped with non-traditional sensors for predicting solution color outcomes in solvent extraction processes.
Our methodology involves the collection and analysis of multimodal sensor data, including pH, conductivity, temperature, and direct color measurements, to inform the ML models and focuses on the analysis of data collected from these sensors deployed on the solvent extraction testbed. The research adopts a two-pronged model approach: the first incorporating all available sensor data, and the second excluding conductivity readings. Through extensive preprocessing, our models are trained to discern intricate patterns within the dataset, leading to improved predictive outcomes.
Project Photo:
Centrifugal contactor solvent extraction testbed experiment setup with sensors monitoring a single contactor. Non-traditional sensors have been installed around single and multiple contactors for testing.