Can artificial intelligence detect impending machine failures even earlier? How can an industrial-scale AI-assisted real-time condition analysis of components be used? A Schaeffler expert provides answers.
When ball bearings are exposed to extreme loads damage is typically presaged by early warning signs. Condition monitoring by means of sensors plays a key role in that regard. However, the data sets gained are enormously complex. Those types of early warnings are hardly identifiable by the human mind because the data volumes are too large and the indicators too concealed. In addition, irregularities may initially appear only seldom.
Where humans would have to compare mammoth amounts of data artificial intelligence can detect damage based on real-time data much faster. AI becomes a game changer for modern condition monitoring. “AI has a unique ability to recognize patterns in complex data aggregations,” says Giulio Cottone, product manager for artificial intelligence at Schaeffler. “AI can merge information from many data sources and recognize patterns.” That not only helps detect anomalies compared to a machine’s normal condition early. Qualitative assessments can be derived from the detection as well. What’s the right strategy for maintenance? Will a bearing exchange be due soon? How long can the machine continue to run regularly? When can maintenance be performed to minimize downtimes? The utilization of AI leads to fewer “false-positive” or “false-negative” reports.
“Artificial intelligence has a unique ability to recognize patterns in complex data aggregations.”
Giulio Cottone, product manager for artificial intelligence at Schaeffler
AI enhances accuracy
Industrial suppliers design their components for loads that are oriented to measurements based on decades of experience. But what if a user encounters unusual loads on a machine that are higher than industrial averages? AI can create forecasts for those scenarios as well. “That’s how maintenance schedules can be optimized,” says Cottone. The next rebuilding jobs can be planned individually, which extends lifecycles. Lubricants can be optimally dosed and individual components be assessed better.
AI helps evaluate data
AI can prioritize data so that experts can evaluate them faster. But its use can also make sense for creating documentations and job orders. The computer intelligence gathers key performance indicators (KPIs) and merges them with other data. To some extent, AI can already pre-formulate entire documents.lower maintenance costs are incurred when operations workers act at an early stage preceding an impending machine failure. Permanent real-time monitoring by means of AI enables prioritization of maintenance or repairs as soon as an anomaly is detected.
In front for a long time thanks to AI
For some time, Schaeffler’s AI competence center has been providing products and services for industry such as the OPTIME system whose various sensor versions, which are integrated with the Schaeffler cloud, monitor vibrations and energy consumption levels, and control smart lubricators. They enable complete condition monitoring with millions of data points. This is where AI-based models and human expertise meet because it’s experts that establish decision rules. The status of a machine can be judged in that way. If thresholds are exceeded, the system warns users and alerts them to perform an inspection, maintenance job, or repair. That ensures maximum transparency without having to go to the machine and inspect it. The OPTIME system has detected defects in thousands of cases, so having reduced costs and downtimes.
By Björn Carstens
