Predictive maintenance (PdM) involves the execution of system checks at predetermined intervals to analyze equipment health. These controls are usually in the form of continual data collection (i.e., temperature, light, pressure, and sound/vibration) from equipment through the use of sensors. The results of these checks determine whether maintenance activities are required.
For manufacturers that work on tight margins and even tighter timeframes, unscheduled downtime can be a nightmare. It can cut right to the bottom line—ruining a quarter, a year, or even a company. The calculable costs are clear: ARC Advisory Group reports that the global process industry loses up to $20 billion of its annual production (about $12,500 per hour) due to unscheduled downtime. Hoping for the best and waiting for something to break is costly—it’s somewhere around 50% more expensive to repair an asset that broke in production than if the problem was identified before the failure.
So, with failures potentially harming personnel and the environment, manufacturers need to tame their complex, interdependent operations. Reliability itself can be a significant competitive advantage. But getting there requires a fresh approach.
Predictive maintenance promises to enable the scheduling of corrective maintenance before an issue surfaces. It should also prevent surprise equipment failures. It shows what equipment will need maintenance and when. As a result, companies can allocate the right parts and ensure they can deploy field technicians only when needed. Instead of dealing with an overflowing schedule of unplanned failures that require immediate and time-consuming production stops, predictive maintenance helps companies schedule shorter outages when it makes sense to slow production.
Although it might be confused with preventive maintenance, predictive maintenance is different. Instead of looking at averages or comparable statistics, it looks at the condition of the equipment in real time. As a result, it can make predictions based on the actual conditions, not averages or suppositions.
Just-in-time manufacturing is the goal for most companies. It means that a company doesn’t get stuck with too much inventory and reaps profits faster because it only invests in parts or other components exactly when it needs them. Of course, it requires precise timing—and every element in the value chain needs to be ready when called on. So, a faulty piece of equipment that malfunctions at just the wrong time can cause a company to miss production quotas, lose business or even threaten the safety of a plant.
There are numerous technologies that predictive maintenance employs, including infrared, acoustic, video, and vibration analysis. It can even look at the oil that lubricates a machine to determine if it is functioning to spec.
Traditional methods monitor single machines or scattered pieces of equipment. They don’t see the entire picture. Predictive maintenance using cognitive machine learning techniques can take all the individual views of thousands of assets to build an integrated view of a factory floor, providing complete visibility and highlighting how assets and their workflows work together—so that if one asset is predicted to go down, it’s easy to understand the broader impact.
One reason that predictive maintenance is a rising trend is that it greatly reduces human errors, which can cause up to 82% of asset failures. As connected assets increase at a dizzying pace due to the IoT, industrial data is overwhelming manufacturers because human beings simply can’t absorb and process all of this data. Without technology to help them, even highly skilled data scientists will almost certainly miss some critical data points. Predictive maintenance that uses data science levels the playing field by applying cognitive techniques for sensor data analysis.
As a result, every enterprise can make automated intelligence available across all levels of decision-makers, ensuring people who need the information stay in the know.
The Progress Predictive Maintenance solution automates data science using cognitive analytics, data mining that looks at historical and real-time sensor data, and machine learning to detect and predict anomalies that will cause asset failures. Seeing into the future seems like a fanciful notion, but this is no crystal ball. It’s real and it means companies can avoid unnecessary maintenance costs, shorten disruptions, gain exceptional control over the torrent of sensor data coming from every machine.
The automated approach uses a patented meta-learning solution to detect and predict anomalies, deliver machine health insights, reduce the time required to develop and operationalize models and help data scientists be more effective. Specifically, Progress provides:
The Progress Cognitive Predictive Maintenance Solution is powered by DataRPM Anomaly Detection and Prediction, which harnesses the power of machine learning, so asset-intensive organizations gain exceptional control over the torrent of sensor data coming from every machine.
Mark Troester is the Vice President of Strategy at Progress. He guides the strategic go-to-market efforts for the Progress cognitive-first strategy. Mark has extensive experience in bringing application development and big data products to market. Previously, he led product marketing efforts at Sonatype, SAS and Progress DataDirect. Before these positions, Mark worked as a developer and developer manager for start-ups and enterprises alike. You can find him on LinkedIn or @mtroester on Twitter.
Copyright © 2018 Progress Software Corporation and/or its subsidiaries or affiliates.
All Rights Reserved.
Progress, Telerik, and certain product names used herein are trademarks or registered trademarks of Progress Software Corporation and/or one of its subsidiaries or affiliates in the U.S. and/or other countries. See Trademarks for appropriate markings.