Development of a connected platform for industrial equipment monitoring to enable predictive maintenance using supervised machine learning methods

Abstract

SHAPE Technologies is the world leader in ultra high pressure industrial waterjet systems for cutting and cleaning with applications from metal to food. Although SHAPE is the technological leader in this space, SHAPE must continuously look toward developing new capabilities to differentiate its products. SHAPE has historically outfitted its machines with a suite of sensors, however these systems in the field do not store the data, thereby losing the time series relationships and historical log of machine health. One opportunity is to create a connected platform that leverages this data to help SHAPE's customers move away from a break fix model to a predictive maintenance program. This project seeks to expand on a sensor connectivity proof of concept ("POC"), which the team successfully built on a prototype grade Raspberry Pi, and make the platform ready for customer beta trial. First, this project explores important infrastructure, legal, and supply chain challenges that impact the commercial business when connecting industrial equipment to the internet as well as the technological considerations to make the platform both backwards and forwards compatible. Second, this project helps define the minimum viable product requirements for industrial infrastructure and devices configuration. Third, this project merges the POC captured data and lab data to train and validate supervised machine learning models to predict failures several days in advance and demonstrates how such a system can help customers mitigate unplanned downtime.