With AI and machine learning (ML), the coordination of maintenance for machines and systems can be automated to a large extent. This is helped by precise information on the actual usage details, such as operating hours and starts of individual machine units – e.g. individual running times of the drive elements, pumps, counting processes for valve operations, etc. The AI sensor MLS/210I is designed for such digitization solutions. It is subsequently attached non-invasively to any machine and supplied with power via an external battery or power supply. The MLS/210I is configured with Node-RED via USB. The counter and status values are transmitted via LTE-M.
The MLS/210I uses an inertial sensor element to record the vibrations of a machine by measuring acceleration and angular velocity in three axes each. This cyclical data image then goes through data preprocessing, is then classified using AI and finally assigned to the individual counter values. With the MLS/210I measuring method plus additional AI functions, anomalies such as imbalance can also be detected automatically. The MLS/210I transmits the meter readings and other messages to the desired destination address via LTE-M mobile communications.
In order to record, prepare and evaluate the training data required to adapt the ML model directly on site at a machine, SSV offers a decentralized testbed service as an accessory. When installing an MLS/210I, this serves as a tool to determine the desired meter configuration or anomaly detection and to adapt the ML model. An SSV expert is available to provide support via remote service on request. Thanks to this data-centric adaptation concept, the MLS/210I achieves significantly more precise results in practice than the current generation of predictive maintenance sensors available on the market.