AkzoNobel is a Dutch multinational company that produces performance coatings for industries and consumers worldwide. The modern-day AkzoNobel came about in 1994 through the merge of the Dutch Akzo and the Swedish Nobel Industries, creating one of the leading coatings companies in the world.

To produce coatings on a global scale, AkzoNobel has an armada of machines and a team of engineers that work to deploy, maintain, and upgrade them. To meet their production needs, AkzoNobel needs to ensure these machines run without unexpected breakdowns. Due to the complexity of the machines, detecting these errors early can be a highly difficult task.

Our Smart Manufacturing Equipment Monitoring is a web application that assists AkzoNobel engineers in discovering machine failures before they happen. Our software alerts engineers about potential failures in real time by using machine learning algorithms to analyze data from sensors installed on machines in AkzoNobel’s plant in Warsaw, Indiana.

Our web application provides an easy breakdown of each machine in the plant, displaying a graph of historical sensor data and detailed prediction information, including estimated time until breakdown and prediction confidence level. If an incoming breakdown is predicted or an abnormal sensor reading is observed, an alert is sent to the user through phone and email.

Machine status, alerts and predictions are consolidated into a dashboard that provides a live overview of each machine’s health. This enables proactive maintenance rather than reactive repairs.

Our product’s front end and back end are built using Next.js. We ingest machine sensor data using VictoriaMetrics. Data for our application is stored in a PostgreSQL database. Our breakdown predictions are made using a Robust Continuous Learning model powered by TensorFlow.