AI & Predictive Maintenance

Part 1 of this article is here. And it covered Tips# 1 to 3.

AI projects are R&D initiatives and not simply IT projects. Organizations miss the mark in realizing value from AI as they undertake AI projects just like their network update, or systems infrastructure upgrade.

There’s a quote one of my previous managers had in his cubicle: “Being a manager is easy, it’s like riding a bike. Except that the bike is on fire, you’re on fire, and everything is on fire”. I chuckle every time I read it and then reflect upon the reality of dealing with project deadlines, deliverables, and performance. Reliability Engineering & Asset Maintenance initiatives have a project management aspect to them.

Organizations are hastily adopting AI into their operating processes to increase efficiency, raise profits, and stay competitive. Among the hustle & bustle, the effective management of the AI projects is neglected, and teams are left to figure out retroactively how a completed AI project fits into the Business’s long-term goals.

One of the common questions teams have when they first explore using Predictive Maintenance is “Is the data good enough to perform the analysis?” Answer to that question is nuanced with the reliability objective and the quality of the data available.

The falling cost of sensors for Industrial Equipment & the popularity of AI-based solutions means that Organizational teams are defaulting to using this strategy on all their Equipment, regardless of its criticality or other effectiveness. This is a strategic error.

In the previous article, P-F Curve was used to understand the Remaining useful life (RUL) of an asset. RUL can be estimated at any time during the asset’s life, but it’s opportune to calculate RUL at the time ‘t’ when the asset shows signs of an impending failure. In the P-F Curve terminology the point at which the asset shows signs of failure is called the Potential Failure Point (Pf), which can also be stated as the time of anomalous behavior. The exercise of detecting anomalous behavior is called “Anomaly Detection (AD)”.

Recently, there has been an influx of Industry 4.0 companies promising their product/application would help predict the Remaining Useful Life (RUL) of a physical asset. Each uses a mix of machine learning algorithms to estimate the RUL based on the data available. This is their value proposition. But what is this ‘life’?

Proponents of the Continuous Improvement method often quote the dictum ‘what gets Measured, gets Improved’. I’d like to modify it by adding ‘what gets Monitored…’ to its beginning. Here I’m referring to the Monitoring of the physical assets in their usage conditions and being Measured & Improved for their Reliability (Availability %, Cost $, MTBF, etc.) and Safety metrics.

‘Fleet’ is the representation of a population of repairable products that are currently in use out in the field by customers. Repairable is the keyword differentiating a ‘fleet’ from commonly known consumer product ‘units’. The differentiation is key in understanding the type of reliability program needed.

The tasks of a Reliability Engineer are long & diverse. While heavily dependent on the industry one is working in, it generally involves all aspects of the Equipment – from Design to Manufacturing to Operation to Maintenance. Even though the responsibility is wide, the resources available for a Reliability Engineer within an organization are limited. Often, there are only a few Reliability Engineers managing hundreds of Equipment. Given this current situation, the arrival of AI seems like a perfect resource to complement the work.

If you’ve ever wanted to learn more about how new Digital technologies like Artificial Intelligence (AI), Machine Learning (ML), Industrial Internet of Things (IIoT), Remote Data Sensing, and Industrial Automation apply to Reliability Engineering, then you’ve come to the right place.