Applied Machine Learning to Critical Assets

LRV-TLV-Redline

When it comes to industries that rely heavily upon expensive assets and infrastructures to enable operations, there is no question that keeping assets functioning in top performance is critical to the bottom-line.

For many years organizations have included planned-maintenance activities, typically recommended by manufacturers of equipment and based on some usage parameters (hours in service, number of miles/kilometers, etc.).   While planned maintenance activities are an integral part of a maintenance regime, condition-based maintenance has also, in the past decade, become commonplace.  Pre-determined threshold values set for a given asset or asset class and based on some simple logic such as if threshold exceeds a particular amount, then take a specific action or set of steps, like schedule a maintenance activity.  The condition-based maintenance is also critical to optimizing the asset investment and reduce any downtime or impact to operations.

The Digital Replica of Assets Holds Value

As digitization continues to blur the lines between physical maintenance and the maintenance of the digital replica of the physical asset, the amount of data gathered about the asset has increased exponentially.   For example, performance data in context with other related factors such as the surrounding environmental conditions are now being collected by asset manufacturers to feedback into product development.   However, product development should not hold sole exclusivity to this data, as Operations and Maintenance (O&M) personnel are beginning to recognize a new gold mine of opportunity which, if managed intelligently, can and gains of 10 to 20% asset utilization improvements.  These are big numbers that cannot be ignored, as it represents direct competitive advantages along with improvements to the bottom-line.

autonomous-assets-self-driving
Image from Komatsu Press Release

There are powerful economic incentives to leverage real-time data and predictive maintenance. Companies can benefit from reduced costs, open new revenue streams, extend equipment life and increase production capacity.

-Martin Provencher, OSIsoft | April 16, 2020 

Examples in Industry

A recent article was written over at Mining.Com, A guide to predictive maintenance for the smart mine, highlights some key elements required to move beyond historical maintenance approaches of reactive, planned/preventative, and condition-based maintenance regimes, and includes:

  • Establish an operational data infrastructure
  • Enhance and contextualize data
  • Implement condition-based maintenance
  • Implement Predictive Maintenance 4.0

AI helps unlock value from Digitalization 4+

It can further be emphasized that new applications of Artificial Intelligence (AI), and more specifically including deep learning into the data-to-analysis workflow pipeline will continue to augment the human experience to identify anomalies and patterns in the asset data that are hidden to the human eye but hidden in plain sight of the algorithms.

The Connected Asset Infrastructure creating vast amounts of usable data

We will begin to explore more on the topic of Applied AI in the field of transportation systems in future articles, and look forward to updating our readers as the field of examples continues to expand.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top