01 Jul 2022

Local application of machine learning in industrial applications

ML1000 hardware add-on for the PLCnext control platform

Detecting unusual behaviour patterns in production machines and systems can not only eliminate malfunctions in processes, it also opens many potential opportunities for increasing the overall equipment effectiveness (OEE). Anomalies can occur in various forms during production. They have greatly different causes and consequences. As production becomes larger and more complex, detecting, locating, and correcting such anomalies is becoming progressively more time-consuming and difficult. To help skilled human workers, AI-based machine learning solutions are therefore being used to achieve the goals of anomaly detection and CO2-neutral production while also reducing costs.

Phoenix Contact has developed both a Hardware platform and software framework, to assist with developing Machine Learning Applications.

The ML1000 hardware add-on for the PLCnext control platform provides a local Google TPU accelerator for AI applications. This reduces the bottleneck caused by communications, or data going to cloud. It also allows the ML application to share data directly with PLC, allowing for high-speed AI processing directly where data is collected.

With the MLnext software framework, users can design their own Machine Learning solution, or use the ready-made ML tools that help guide a user through the Machine learning process. The software framework is also platform independent, so it can be implemented on the ML1000 or on the user’s own platform.

Digital transformation offers numerous opportunities to increase the flexibility, productivity, transparency, and availability of your production. To take advantage of these, the factory data provide the basis for achieving sustainable production. Phoenix Contact has scalable and already-tested solutions on hand for Data collection and Evaluation, Data Security, Data Storage and Transportation.

Company info: Phoenix Contact Ltd