The Benefits of Predictive Maintenance in Manufacturing
According to Oracle and Intel, predictive maintenance outranks other maintenance models. For starters, it can reduce downtime by identifying equipment and systems that are not running optimally, flagging potential problems early on. Second, it increases production by keeping equipment operational. Third, it improves worker safety by preventing potentially dangerous equipment failures so that workers know to take advance caution around certain equipment. Fourth, it runs quality control on equipment parts, as poorly running machines are more likely to produce defects. And finally, it extends the lifecycle of equipment by minimizing premature wear and tear.
The BMW Group plant in Regensburg, Germany, saw these benefits in action when its in-house machine-learning models created heat maps to visualize fault patterns that maintenance workers could focus on. BMW estimates that AI-supported systems save teams “more than 500 minutes” (or more than eight hours) of disruption per year at the Regensburg plant alone.
“Optimal predictive maintenance not only saves us money, it also means we can deliver the planned quantity of vehicles on time, which saves a huge amount of stress in production,” says Deniz Ince, a data scientist on the innovation team at the plant.
DISCOVER: Stories about manufacturing's tech revolution.
Unlocking the Power of Predictive Data
For manufacturing companies to maximize predictive maintenance models, they must first understand that AI’s power lies in its data.
Before the first measurements are taken, an AI-driven maintenance system needs a benchmark of success. Machine-learning models can be trained on historical equipment data, so the AI begins to learn what is a normal operation and what is not.
At this stage, Internet of Things sensors monitor equipment and gather real-time data such as temperature, vibration and pressure. These current values are compared against historical markers.
AI and ML can make inferences that indicate a problem — for example, if a motor’s temperature and current draw normally change in relationship to each other, but deviations in either measurement start to appear, Ruparelia explains.
“Outliers and anomalies are indicative of something amiss, and AI/ML is great at spotting that,” he writes.
When edge computing is integrated into equipment sensors, data can be collected and analyzed at the source rather than sent to onsite servers or the cloud for processing.
RELATED: Solutions and services can help your manufacturing team.
A Predictive Maintenance System to Manage It All
Another critical element of AI-driven predictive maintenance is a system to manage it all. This system is where data analytics, IoT sensors and detected anomalies come together to produce performance trends, optimized maintenance schedules and general insights.
Toyota’s Indiana assembly plant uses Maximo Application Suite from IBM, a cloud-based enterprise asset management system, to equip maintenance workers with the data necessary to drive faster, predictive decision-making.
“Maximo allows a skilled team member to see the health of the equipment and its components, monitor for any abnormal activities and use predictive solutions to change our maintenance work from reactive to truly proactive,” says Brandon Haight, a general manager at Toyota North America.
Rafi Ezry, a managing partner at IBM, adds that “shop floor data powered by AI and IoT can come together to reduce downtime by 50%, reduce breakdowns by 70% and reduce overall maintenance cost by 25%.”
