Predictive machine learning model for improved safety and product quality

Principal investigators: Lutfi Agartan and Dimitrios Fafalis

University: Drexel University

Industry partners: Eastern Controls, Inc. and Endress+Hauser, Inc.

Drexel University (DU), Eastern Controls Inc. (ECI), and Endress+Hauser started a collaboration a year ago to develop teaching lab equipment, offering co-op positions, and sponsoring senior design projects. Now, DU and ECI teamed up to develop a predictive machine learning model (PMLM) capable of continuously tracking the data coming from sensors to detect abnormalities and shut the system before failure which can lead to downtime and expensive maintenance. We will utilize the flow measuring equipment (FME) designed and built by Drexel Co-Ops in ECI for collecting experimental data to be used for training the PMLM. Real-time performance and prediction accuracy of trained PMLM will be tested after its implementation to FME, where some conditions suggesting normal or abnormal operation will be run. A successful PMLM will enhance the product automation capabilities of ECI and will offer a strong tool for DU to offer their students real-time applications of state-of-the-art technologies.