CloudUp: Getting Started With Machine Learning For Predictive Maintenance

Predictive maintenance is a growing buzz word in the industry, but how many companies are actually making progress? Some companies are reporting a reduction of equipment downtime by up to 50 percent with predictive maintenance using IoT. The main takeaway is that you can save a lot. On this episode of CloudUp, we’ll be getting past predictive maintenance as a buzzword and get into the what, why, and how your company can make progress.

Meet the Speakers

Rick Erickson from CloudUp headshot

Rick Erickson

Co-Founder and Chief Cloud Strategist

Mark Brose Cloudup Headshot

Mark Brose

VP of Software Engineering

Transcript

Mark

What we’re seeing is there just hasn’t been quite as much pick up of it as we would have expected, so a lot of interest, a lot of value, perceived value, but it’s moving a little bit slowly. I think what we’re seeing, even for single-use cases, you’ve got clients that are seeing hundreds of thousands of dollars that can be saved even with single-use cases so, the take away there is there’s a lot of stored value here in doing something with predictive maintenance, so definitely worth taking a look at and seeing if there’s any value you can add to your company.

Rick

On this episode of Cloudup, we’re gonna be getting beyond predictive maintenance the buzz word and focused on the what, why, and how you can make progress quickly.

Rick

So welcome to Cloudup, the series where we explore the coolest things built on the cloud today brought to you by Agosto. Today’s topic we’re gonna use machine learning to predict maintenance events and there’s a ton to dig into here so let’s jump into it. First off let’s talk about how companies perform maintenance today. Is it proactive, is it reactive, and what’s the impact of that?

Mark

Yeah, we definitely see companies doing both reactive and even proactive maintenance. You know I think the more advanced companies we’re seeing are doing proactive maintenance in the sense that they’re really doing maintenance on a schedule right, so preventative maintenance. Over time you can develop some experience with when you should replace things, but what we see there is that’s not optimized, right? So that’s not, in some cases you’re gonna replace things too early, in some cases you’re gonna wait ’til it fails, so it’s difficult to get it just right. So there’s some cost to that. You do it too early, you’re losing some material usage that you could’ve gotten and at volume and scale, that can be a lot of money and it you wait ’til it fails obviously you might have, if we’re talking tires in a fleet, trucks out of service, maybe has an accident, liability issues, so there’s cost of waiting too long. What we can get to with predictive maintenance using machine learning is you get that a lot tighter. Not saying we’re gonna ever get 100%, but you’re gonna get a lot tighter inside that window so what we’ll be able to save on material, but also prevent more of those failures from happening. That’s to me is the big impact.

Rick

Yeah, so going back to your point about just reacting and not being able to predict the impact of failure, not good, right? Sounds kinda like how I rolled in high school with my Malibu. Let’s talk a bit about how using predictive maintenance can help avoid some of those unexpected costs.

Mark

Again, it’s really being able to optimize that time window. We’re doing that by having a lot more data. We’ve got kind of a wider range of data that we can take advantage of so we can use somewhat static data like making models of a tire, a piece of equipment which tells us something about how it’s constructed, and there’s some predictive value to that, to how long it’s been running, what conditions it’s operating under, to real-time telemetry data like temperature, tire pressure, vibrations, that kind of thing. All those things together then really can be used to build a good model.

Rick

So these are key attributes that ultimately humans can even understand the basis of those key attributes, but I imagine that at scale when you have millions and millions of events, it’s really hard to understand what’s happening, how to use that data and create classified information that fits into categories that we understand.

Mark

Yep.

Rick

I imagine by using machine science and Cloud ML, we can use some of that information to train models so how does that all work?

Mark

Yeah, so that’s a good question. We find still in machine learning there’s a lot of value still to human input and the primary value of that is in this area we call sort of feature engineering. It’s the fancy word for knowing what data elements will be predictive of failure. So it’s still helpful to have domain knowledge to sort of pick those data attributes that should be included, but then what we can do is we can take advantage of the machine learning technology to take that data and create the algorithm for it based on machine intelligence, so it’s not something the human has to spend all this time engineering the algorithm. Their focus is on getting high quality data in place that we can use to be predictive. With the cloud, the cloud really brings to us the platform to run all that on so a lot of time the data scientist that’s working in this space won’t have infrastructure background or development background. So what cloud platform can give us is a lot that just as a service so we don’t need to spend a lot of time or having broader skill sets in that team to build out these models.

Rick

Sweet. So now we’re gonna take a quick break to hear from our sponsor and then we’ll come back and do some jamming on the jam board.

Sarah

More and more we’re seeing organizations wanna be really strategic with their success and part of that means they’re moving to the cloud. At Agosto, we’re seeing a big uptick in clients using Google Cloud platform for their online business operations. So whether you’re thinking you need a boost with AI, machine learning, or you just wanna build something new and fast, Agosto is an award-winning partner who would be happy to help you with your needs. We’ve been Google’s Cloud partner of the year multiple times. We’ve got a few other awards as well and we would love for you to get to know more about us, okay, can’t take anymore, by visiting agosto.com. You can get a free guide to your company’s strategic plan for heading into the cloud. We’d love to see you there. Now back to the show.

Rick

Welcome back to Cloudup. In today’s deep dive session, we’re gonna focus on how you can get started on predictive maintenance and

Mark’s gonna take us through our approach.

Mark

Yeah, so definitely where you usually wanna start is pretty basic just framing your business problem. Essentially that’s just thinking about use cases that make sense for your business. Pretty simple approach, we essentially start out with ID’ing the business area we’re focusing on. An example of that is gonna be, so let’s take the tire example we’ve talked about a little bit. You’re spending too much money on recovering from failures of tires in our fleet. So just framing up that’s an area you wanna focus on and then what we wanna do is dig in a little bit deeper in that and create an actual business question that we’re trying to answer. So in this case, that’ll be something like can we use predictive maintenance with machine learning to better predict failure than we’re doing right now with our preventative maintenance program? What is a more specific thing that we’re trying to target?

Rick

And this is really an impact right? So we’re trying to frame this up as high value, low risk, but a problem that’s going to impact this organization in a material way that’s considered financially viable.

Mark

Yep, but it’s a high-level problem, it’s a framing of the problem in a way that we can target, is it something we can do? You have to think about here you’re typically thinking about well, what does failure mean? Tire case is pretty easy, tire blows up. But with a machine, does it just slow down, does it actually shut down the line, you have to frame what is failure for us, what is the thing we’re trying to detect and some sense of what’s different than what we’re doing today? That leads you to selecting a business metric. Say we’re trying to target tire failure. Business is really our success metric that we’re trying to get to. Say we’re trying to get to, for us right now we’re predicting at about 70%. Can we get to 90? So we have a target. In this first phase, you’re just trying to narrow down the focus of what it is that you’re trying to do.

Rick

And percentages obviously matter when you frame it up like this, but in some cases, if the impact, even if it’s 75 or 80%, so let’s say it’s 5% or 10% better, in some use cases that can be really impactful.

Mark

You can turn that into we’re moving from 80 to 90 and you’d be able to turn that into dollars. That might be for you 50 grand, maybe it’s like hundreds of thousands of dollars depending on what that looks like for you.

Rick

And so the way to think about this is I’ve got humans today that maybe can react to a problem, but when I’m using a system that can handle massive amounts of information and then predict an outcome and I can alert on that prediction. You can also think about sort of downstream as I’m framing the business problem, what’s the ultimate impact if I can go from this sort of narrow use case to something that adds scale?

Mark

Absolutely. You should think about this is a place to start and a lot of times what we’ll do is we’ll look at a whole bunch of use cases. You might start out with maybe 10 or 12 ideas, run this process through with those 10 or 12 ideas and out of that we’ll get to hey, if we can hit these success metrics, these things logically bubble to the top, so we’re looking for hey we found some high value potential here and then what we’d move into is how do we start to test that out and see if we can actually do it.

Rick

Cool. All right.

Mark

So that’s framing where we start and then where we’re going to here is really we’re trying to get to somewhat of a iterative process of learning. We’re gonna frame this up as kinda think about this as a circular pattern here where we’re doing a continuous process and what we’re starting is where all the really needs to be is in data prep. So data prep is really, talked about this a little bit already, but it’s essentially pulling in data from a bunch of different places and doing some data exploration, making sure we can suss out what are predictive variables? So again with the tire example, we’re talking about things like make and model of tires, talking about how long the tires have been running, we’re talking about temperature, tire pressure, all these kinds of things. We’re pulling that together and then this is the place where domain knowledge is important. You build this stuff up, you explore it, you do some visualizations maybe, you’re doing some graphing, you have some thoughts on what’s predictive, but in this phase you’re really spending time determining whether it really looks like it’s gonna be predictive and in this case, you’re sometimes maybe you’re creating some absolute values, you’re looking at median values, or moving averages of data of variables, so this is a part where you’re spending probably the most of your data engineering and science time really is pulling that data together and shaping it in a way that you can use it.

Rick

Are we trying in this phase do we try to understand what a representative amount of data or a relationship means to humans yet, or are we just trying to make sure that the data is in a form that’s sort of repeatable and consistent?

Mark

It’s a little bit of thinking about what you can actually operationalize and then from a model-building perspective, it’s about throwing out things that aren’t useful, shaping things in a way that look predictive, it’s really kind of all of that. You hear terms around data engineering, data normalization, that’s kinda what’s happening in this phase.

Rick

Okay.

Mark

We mentioned a little bit earlier sort of your feature engineering and you’re really pulling together the things that look like they’re gonna be valuable.

Rick

Okay, and throwing those things out that maybe won’t be.

Mark

Exactly.

Rick

So it’s not just getting everything, it’s also trying to be smart about what you are using because we’re trying to again, move relatively quickly to get to some validation of the business problem.

Mark

That’s right. In some cases, you’ll already have this, like in a data warehouse. In some cases, there’s gonna be some work to get the data. This is the foundation really. We have to have good data and lots of it.

Rick

Cool.

Mark

So then from there, we move into what we call model involvement. So here, this is really where the machine learning part kicks in and sometimes there’s straightforward answers to what type of ML approach we’ll use here, but you may experiment with a few different options and kind of see how that susses out. There’s a little bit of data science to this part, but here’s a lot of our letting a machine now use all this data that we got and building models from it. And so we’re doing this typically with a subset of our data as we’re engineering models that look like they’ll be valuable.

Rick

Okay.

Mark

This is the machine part and then we’ll move from there into an evaluation and review. In this phase, what we’re doing again we’ve taken a subset of our data to build a model, now we’ll run that model against a test holdout set of that data. So we’ll have trained a model and here we’re running that against some test data to see is it actually predictive? Sometimes the model may get overfit to the data you used if you have other data that doesn’t look exactly like that data, maybe it isn’t quite as good as you thought. You wanna do some work here to make sure that it is as predictive as it looks like it was when you were building that model.

Rick

Okay, so you’re actually looking at how well this information, this model is predicting the outcome that you’re hypothesis is expecting?

Mark

That’s right, is it performing as you thought? At this point we’re all operating here very much in the cloud, cloud platform tooling, usually a couple of data engineers, data science people can do this all this work. Here already, you might have some iteration. If this is bad, you may go back here and like, all right, we need to take a new approach. That can happen.

Rick

Sure.

Mark

But ideally, from here then we’re moving on to looks like we’ve got something that’s useful, now we’ll get into deploying that model. This can be straightforward if you got a lot of the things in place, but this could involve okay now we’re putting sensors in the field, maybe we have some edge processing that we need to do, so we might engineer a mini data center close to that edge where the data is.

Rick

So you’re gathering more data?

Mark

You’re actually gathering more data, but definitely creating the infrastructure to wherever the feedback loops that we want are available. So put a model out there and now what do we want to have happen? If we’re predicting tire failure, we wanna alert somebody that hey the tire’s about to fail so somebody does something. So here is the work we do to get the model out there, get whatever tech deployed to do the alerting. If we’re alerting a driver directly, maybe it’s notification immediately in the vehicle or maybe it’s to a dispatcher that lets them know, depends how we wanna architect it, but this is when we’re operationalizing the whole thing. And in the first rev of this, this may be a small group, you’re gonna wanna not affect your whole fleet ’cause when you go to the field you inevitably figure out things you didn’t think were gonna happen. At this point, you’re again validating all this stuff works in the real world.

Rick

Okay.

Mark

And making tweaks that will fit for your business.

Rick

when we go back to this first process, which is framing of the business problem, understanding these key components, defining some sort of targets of success that are important for the stakeholders in your organization, and then we start iterating on this sort of loop of processing. How long does this usually take us?

Mark

We can typically do a small run of this in a short time period. You might be able to do something as fast as four to six weeks. A lot of it will depend on where you’re at with your data. Sometimes there’s work that needs to be done to get that data in place so that can take a little longer so that data engineering piece, that will drive a lot of how quickly the rest of it will go.

Rick

In our experience, we typically will run workshops and help the executives and the engineers that we’re working with at organizations that our organization partners really help understand what’s possible so that they’re not stepping into mistakes that typically because of inexperience and not understanding this process real well, that they won’t do that. They’ll make good choices and we’ll help them so that they’re focusing on the right problem with the highest value and ultimately is gonna have the highest impact to their organization.

Mark

That’s it.

Rick

All right.

Rick

Thanks for watching this episode of Cloudup where we focus on the coolest technology delivered on the cloud. We’d love to hear your feedback comments on how you can use predictive maintenance in your own industry, how you’re making progress in this space and if you have questions or challenges that we can address here around predictive maintenance, leave a few comments, leave a few notes asking questions, and we’ll give you some swag. Thanks again.

Sarah

Cloudup is brought to you by Agosto, a leading Google Cloud platform partner. Like this episode and subscribe to our channel on YouTube to learn more. We would love to help you out. Visit agosto.com to learn more.

Overcoming Manufacturing’s Customization Barriers to Predictive Maintenance Solutions

Welcome to the holy grail of maintenance solutions – Predictive Maintenance (or PdM).

If you’re reading this post, it’s likely you’re already familiar with the concept of Predictive Maintenance solutions. You’ve begun your due diligence. You understand it’s one of the most effective ways to optimize the maintenance and performance of your assets (not to mention your bottom line).

Management consulting firm McKinsey & Company notes its industry-transforming power:

“We estimate that predictive maintenance could reduce maintenance costs of factory equipment by 10 to 40 percent. Such savings would also be possible in health-care settings. Additionally, better predictive maintenance using IoT can reduce equipment downtime by up to 50 percent and reduce equipment capital investment by 3 to 5 percent by extending the useful life of machinery. In manufacturing, these savings have a potential economic impact of nearly $630 billion per year in 2025.”

In his book, The Plant Engineer’s Handbook, Keith Mobley further documents that predictive maintenance can:

predictive maintenance capabilities

Even if you assume these to be “best case” results, generating a fraction of them can mean millions of dollars in increased performance for any manufacturer.

And if that weren’t enough, predictive maintenance offers an important associated benefit. It increases overall workplace safety; reducing risks associated with malfunctioning equipment and boosting worker efficiency.

These are no small impacts for any manufacturing business.

They are benefits that simply can’t be had from reactive or preventative maintenance programs.

Adoption of PdM Has Been Shockingly Slow

Despite these documented financial, performance and safety benefits, predictive maintenance has been slow to catch on. A recent study by Bain & Company revealed this unexpected trend:

“Two years ago, predictive maintenance was forecast to be one of the most promising uses of the industrial Internet of Things (IoT). … So it’s somewhat surprising that predictive maintenance has failed to take off as broadly as expected. A recent Bain survey of more than 600 high-tech executives has found that industrial customers were less enthused about the potential of predictive maintenance in 2018 than they were two years earlier.”

Why would any business be slow to adopt a technology with such incredible ROI potential?

At Agosto, we believe this reluctance among manufacturers is the result of a combination of factors.

In this article we will briefly explain those factors. Then offer solutions manufacturers can consider to overcome them when implementing a proof of concept installation.

The Biggest Barrier to PdM Implementation

At its most basic level, the requirements for an effective predictive maintenance program are fairly straightforward. You need sensors, data, and a processing platform.

But that’s where the simplicity ends. Predictive maintenance gets its “super powers” from a level of sophistication unlike any other maintenance strategy.

It leverages the power of big data along with artificial intelligence (AI) and machine learning (ML).

That means, by default, there are no “cookie cutter” solutions. No one-size-fits-all monitoring system you can unpack, plug in and start generating same-day results in your factory.

And that is compounded by another reality of manufacturing businesses.

Top manufacturers, who can benefit the most from predictive maintenance, often have the most customized manufacturing processes.

It’s what sets them apart in their market. And gives them a competitive edge.

That means that not only are there a large number of factors to consider regarding the solution itself, but all those factors need to be customized to every manufacturing environment.

These “customization barriers” can often seem overwhelming to any manufacturer thinking about a predictive maintenance solution.

However, by understanding four fundamental considerations, you can simplify the process greatly.

We’ll examine those four considerations in the rest of this article.

First Things First: Creating a PdM Strategy

The first thing any business considering a predictive maintenance strategy must understand is that implementing it is a process – NOT a project.

The goal of any implementation should be to start small while developing a “repeatable playbook.” One that can be cost-efficiently scaled for the other assets in your company.

Here at Agosto we recommend our clients employ a “lean” implementation strategy. One that will let you implement at the lowest possible barrier. Validate your proof of concept. And then scale up from there.

Here are four areas where you’ll need to consider your customization needs.
1 in red circle graphic

Consider the Assets You Want to Monitor

2 in red circle graphic

Determine the Failures You Want to Predict

3 in red circle graphic

Assess the Data You’ll Need to Collect

4 in red circle graphic

Define the Models You Need to Run

1. Consider the Assets You Want to Monitor

The good news where predictive maintenance goes, is that you don’t have to implement it across your entire business.

In fact, you shouldn’t.

In many cases, more expendable production assets may actually be better served with a preventative or run-to-failure maintenance strategy.

For PdM, you want to focus on your most critical assets. In general, determining these is pretty straightforward.

Equipment where an unexpected breakdown would slow or stop production entirely is an obvious consideration. Especially equipment that needs to run on a near constant basis with little to no downtime afforded for maintenance.

But there are other factors you should consider when selecting customized equipment for your proof of concept.
clock graphic

Assets with higher failure rates or maintenance needs.
These can be ideal candidates as well. Predictive maintenance can not only reduce downtime, it can also help uncover the root causes of frequent failures.

crane in use graphic

Equipment with expensive or hard-to-source replacement parts.
The goal of any business is to minimize inventories of replacement parts. Fixing or maintaining these machines becomes especially costly when these parts have to be ordered on short notice.

caution graphic

Consider remote assets.
For equipment in hard-to-reach or even dangerous locations, predictive maintenance can allow for more frequent and safer monitoring.

crane graphic

Difficult to replace assets.
Would replacing a roof top fan require bringing in heavy equipment like a crane? Any asset whose replacement carries added costs or planning time can be a good candidate for your predictive maintenance trial.

percentage increase graphic

Finally, don’t limit your exploration to your most expensive equipment.
Critical assets come in all shapes and sizes. A $200 fan motor that is essential to the operation of an entire production line may be just as important as the million dollar machines on the line.

2. Determine the Failures You Want to Predict

The goal of predictive maintenance is generally understood to be preventing catastrophic failures that grind businesses to a halt. A better way to understand it, however, is as a means to manage failure.

Every piece of machinery you have is eventually going to fail.

One of the biggest challenges to manufacturers’ maintenance programs is determining when to replace components or entire assets. This challenge increases with the complexity and customization of a manufacturer’s production system.

When components of an asset operate in unison to produce an outcome, it becomes exponentially more difficult to determine which and whether a component within the system should be repaired or replaced in a timely manner (not prematurely) so the whole system stays up and running.

The goal of predictive maintenance is to quantify the risk of a specific failure at any moment. And then use this data to manage the effects of that failure

If you can effectively plan for equipment failures in advance, you can achieve all the benefits predictive maintenance offers.

Understanding that, there are 3 basic failure outcomes predictive maintenance can predict for:
Clock graphic

Estimating the remaining useful life of an asset:
This failure projection involves making predictions about the time remaining before a component or piece of equipment will need to be replaced.

stopwatch graphic

Predicting whether an asset will fail within a specific time frame:
Again, predicting this doesn’t reduce the incidence of failure. It does,however, give you time to organize resources, order replacement parts, and minimize downtime.

red switchbaord graphic

Detect and predict anomalies within a complex system that may lead to failure:
Anomaly detection is unique in that its purpose is to discover mechanical behaviors that are not part of the equipment’s historical data. In essence, it’s trying to predict the unknown.

This type analysis can take one of three courses:

It can recognize established anomalous patterns within your assets and signal when they occur.

It can attempt to establish “unrecognized” spikes and dips – deviations from the normal range of operating data – and report them as anomalies.

Finally it can track minimal changes over time until they become anomalies. This analysis presents a challenge of its own as it is not “pattern based” like the previous 2 models. Here performance degradation takes place over time making the resulting anomalies more difficult to spot.

You should start by selecting only one failure outcome to manage. Ideally those that best match the needs of your current production.

3. Assess the Data You’ll Need to Collect

Effective predictive maintenance is a function of machine learning. The key to effective machine learning is access to data. Lots of data.

But don’t let this requirement put you off.

Like we’ve said before, no one’s data is perfect, so start with the data you have.

“If you’re like most manufacturers, you already have a rich store of data for machine learning to use, including MES, PMM or CMMS data, your ticketing system and other databases across the company.”

In general, specific data requirements will depend on the type of failure you want to monitor for. All of it, however, should be obtainable from one of 4 sources:

Historical failure data: Failure data is one of the data types critical for predictive maintenance. Yet it can be one of the most difficult to come by.

A lack of this type of data is not uncommon. It’s often due to preventative maintenance systems which over-perform maintenance (while increasing costs).

So what to do?

To overcome this obstacle, it’s possible to simulate failure data to train your algorithms.

Consultants with extensive systems knowledge can leverage strategies such as Failure Mode Effects Analysis (FMEA). Following these steps, they can adjust factors like temperature, vibration, and flow rates under certain scenarios to simulate potential failure scenarios. This data can validly be used in place of actual failure data.

Historical maintenance and repair data: These data include records of components that have been repaired or replaced. Recent inspections should be noted as well.

Machine operating data: This is data collected under normal operating conditions. There should however be enough to contain variances that capture the aging process of the equipment being measured as well as any potential anomalous data that could also lead to degradation. In other words, it’s data collected under both normal and faulty operating conditions.

Equipment meta-data: This refers to “static” data about the actual equipment being monitored. It includes information such as model or version numbers, manufacture date, operational settings and the like. This data never changes relative to the performance of your equipment. But it can add a powerful predictive advantage to your data.

One final note: Data calculations at scale are typically done by leveraging a cloud implementation. This does not need to be the case when testing. Data computations can frequently be done “at the edge” – on a small amount of local infrastructure deployed for inference purposes. This can be a preferred solution to start out. Especially when data transfer costs are high.

4. Assess the Data You’ll Need to Collect

When considering the type of models that will best suit your needs, it’s first important to understand the basic types of calculations predictive maintenance uses.

Predictive maintenance math essentially falls into one of two categories:

      • Condition monitoring
      • Predictive analysis

Condition monitoring models use data obtained directly from machinery. These include factors like temperature, vibration, fluid pressure and so on. Its goal is to assess a machine’s condition at any point in time and diagnose possible faults.

To do this, it tracks condition indicators which are data of the system that can predictably be measured as the system degrades. It groups similar clusters of performance data together. It then diagnoses potential faults by comparing incoming data against fault benchmarks.

Predictive analysis forecasts when a failure will occur based on past performance data. This model looks to predict future values of condition indicators based on the current and past states of the machinery.

Predictive maintenance uses a combination of these calculations to analyze and report on the data collected from the system.

Depending on the failures you want to predict for, you’ll need to consider implementing one of two main models.

Regression models: These models are best used for estimating remaining useful life (RUL). They look at dependent variables in conjunction with independent variables to measure the correlation or “strength” of the relationship between them.

They return a specific value for the likelihood of an outcome occurring.

Classification models: These models come in two flavors: “Binary” and “Multi-class.” They are typically used to predict whether an asset is likely to fail within a certain time period. Rather than returning the likelihood of a single outcome, these models seek to answer a “yes or no” question – will a piece of equipment fail with X time frame – with a particular degree of certainty.

The model you choose will be largely determined by the results that best serve your manufacturing process.

The Bottom Line: Leverage the Expertise You Need

As you can see, there’s a lot to consider when thinking about a predictive maintenance program.

And much of it depends on the complexity and degree of customization of your manufacturing processes.

But these considerations don’t have to stand between your company and an effective predictive maintenance program.

This article is not meant to be a comprehensive guide in solving these dilemmas. In truth the vast number of possibilities are beyond the scope of any online article.

What it hopes to do, is to shed some light on the customization barriers that must be considered when implementing a predictive maintenance proof of concept solution. Also, to step you through the major points you need to be aware of when designing a predictive maintenance solution for your business.

There are any number of hurdles a business can face when seeking to leverage a predictive maintenance solution.

But the only real barrier is when implementations are done without the requisite expertise to properly integrate them into a highly customized manufacturing processes.

Leveraging the expertise of your in house team – as well as external resources – to determine the specific needs for your business will make implementing your proof of concept, and ultimately a full blown predictive maintenance process into your custom business environment a snap.

If you have more questions, we’re happy to answer them.

Ready for predictive maintenance? It’s easier than you might think

Manufacturers can build on nearly any PMM, MES or CMMS platform with machine learning and computer vision capabilities

Downtime.

It’s an eight-letter word certain to make manufacturers cringe. If your systems are idle, you’re likely losing money.

How much money? The International Society of Automation estimates that globally, manufacturers lose $647 billion each year to machine downtime. Gartner sets the cost as high as $540,000 for every hour that assets, workers, infrastructure, systems or networks are unavailable.

That cost—coupled with the sheer unpredictability of idle systems—are two reasons the global manufacturing sector is eager to embrace machine learning. Harnessing your manufacturing data in new ways is a game-changer for productivity, maintenance and of course, your bottom line.

Machine learning moves mainstream

The opportunity for manufacturers is to move from reactive to proactive. In this case, from unexpected repairs and unscheduled downtime to predictive, planned maintenance that minimizes disruptions, reduces costs and streamlines operations. Machine learning provides the tools to make it happen.

While it sounds complex and expensive, companies have deployed machine learning for years; today, it’s the widespread availability of cloud computing power that’s bringing the technology into the mainstream.

We interact with machine learning every day:

    • Navigation suggestions that avoid traffic jams
    • Fraud detection that knows if “you” are really making those online purchases
    • Vehicles with lane departure warnings and accident avoidance alerts
    • Customized playlists based on your personal musical tastes

The term applies to any computer system that leverages algorithms and statistical models to perform specific tasks—without requiring explicit programming or instructions.

    • First, data scientists teach the system to recognize patterns and make inferences, or conclusions and predictions based on data.
    • Next, a supervised training period occurs where staff actively monitor and adjust the system’s performance.
    • Finally, the system moves to unsupervised training and truly lives up to its name: a machine that learns on its own. The more data it consumes, the stronger the insights it produces.

Employing machine learning to solve manufacturing challenges is a logical extension of its capabilities.

Preventive techniques lead to over- or under-maintenance

Today, most manufacturers rely on preventive maintenance management (PMM) to keep equipment operating smoothly. Preventive maintenance employs time, usage and event data, such as hours in operation or replacement dates of a component. Tracking those milestones enables manufacturers to perform maintenance activities on a set schedule.

Dozens of platforms exist, including manufacturing engineering systems (MES) and cloud-based maintenance management systems (CMMS), to help manufacturers monitor their equipment. However, analyzing PMM data and making decisions still falls to humans. Operators must log information. Analysts must crunch numbers. Managers must make decisions—often weighing the value of PMM data against their gut instincts.

Preventive maintenance also excludes numerous important inputs. For example, issue logs from the company ticketing system, images of wear-and-tear on internal components or vibration patterns from machines in operation. All are key variables that affect performance and maintenance timing—but difficult to track and integrate in a typical PMM platform.

As a result, preventive maintenance often leads to over- or under-maintenance. Both situations can be costly. Unnecessary maintenance adds up with the cost of replacement parts and idle systems. Infrequent maintenance may extend the usable life of a component, but raises the risk of unscheduled outages. Neither is ideal.

Predictive maintenance increases uptime, reduces costs

Predictive maintenance using machine learning and computer vision is the next step forward for manufacturers. It builds on the foundation established with preventive techniques, with several key differences.

Predictive maintenance:

    • Leverages more data sources
    • Requires less human intervention or analysis
    • Allows customization to your specific manufacturing environment
    • Yields far greater precision in maintenance recommendations

For example, predictive maintenance systems can analyze qualitative data, such as ticketing system entries, alongside quantitative metrics from sensors and operation logs. Examining multiple data points provides more accurate maintenance insights.

These smart-systems can also detect correlations faster than their human (or PMM) counterparts. Machine learning intelligence may spot previously undetected patterns that lead to equipment issues or failures, such as vibration levels in equipment.

Computer vision is a particularly useful element. This tool combines one or more cameras mounted on critical equipment, with a machine learning system that’s been trained to recognize minuscule differences in images. The applications are numerous. For example, manufacturers can inspect product quality and detect defects in areas where it’s not feasible for human viewing. Computer vision provides high-speed operation in multiple locations on a manufacturing line.

Computer vision can be especially valuable for medical device manufacturers or those in regulated industries, where any equipment modification triggers a costly and time-consuming recertification process. Innovative medtech companies can now identify visual changes on a manufacturing line without requiring recertification. This approach accelerates the ROI for predictive maintenance systems.

Reduce maintenance costs by 25 percent

Predictive maintenance shows great promise across a wide spectrum of manufacturing sectors. McKinsey estimates that manufacturers with predictive capabilities can increase asset availability by 5 to 15 percent, and reduce maintenance costs by 18 to 25 percent.

Predictive maintenance can improve quality, reduce defects and scrap, and detect suboptimal performance in manufacturing equipment.

The technology can also complement the work of human quality inspectors. For example, every hour, a quality inspector may review hundreds of finished products for subtle differences in finish quality, color, hue, or other qualitative attributes. The inspector then flags a few dozen each shift for further review.

Computer vision with machine learning could then examine the flagged products at a deeper level—and with greater precision—to identify the exact source of the discrepancies. In this scenario, quality inspectors gain more time to focus on higher-value concerns.

Advantages of predictive maintenance:

    • Increase uptime and availability of key equipment
    • Decrease maintenance and repair costs
    • Reduce defects and scrap
    • Improve overall quality of production

Predictive maintenance tools have manufacturing applications beyond day-to-day operating performance. Imagine having a near real-time knowledge of your inventory levels in the field—without needing an expensive RFID system or changes to your packing model. You could configure a computer vision and machine learning system to measure inventory use, expirations and redeployment triggers across the supply chain. The resulting insights would increase efficiency and provide a competitive advantage.

It’s another reason the global manufacturing community (and we at Agosto) are so excited about machine learning’s potential.

Early adopters will gain a competitive edge

Despite all its potential, some manufacturers are hesitant. Common concerns include complexity, data quality and system integration.

The good news? Getting started with predictive maintenance is easier—and less complicated—than you may think.

We recommend these four steps:

  1. Take an iterative approach.You can “go big” with machine learning and an enterprise-wide business transformation that takes years—or you can start small, prove the technology’s effectiveness and gradually grow machine learning’s footprint across the organization.At Agosto, we’re big fans of an agile approach where we work quickly and collaboratively, learning as we go. In this fashion, manufacturers can accelerate value and see results in weeks or months, rather than years.
  2. Start with your existing data.Data is another perceived obstacle that doesn’t need to be a hurdle. No one’s data (repeat, no one’s) is perfect! A good technology partner will assess your data readiness and help build out from your current foundation.If you’re like most manufacturers, you already have a rich store of data for machine learning to use, including MES, PMM or CMMS data, your ticketing system and other databases across the company. Machine learning can even access unstructured data that’s “trapped” in PDFs, emails or other documents.Another key point, you don’t need to move your data offsite to deploy predictive maintenance. Today’s flexible, cloud-based systems can typically access your data from anywhere, which helps you get up-and-running quickly.
  3. Deploy a structured proof of concept (POC).A POC uses just enough data and functionality to show stakeholders (and skeptics) the value of a machine learning application. It’s a great way to see the tool in action, in your own environment, without a large investment of time, money or resources. From the POC, you can identify requirements and build the business case for a full-blown predictive maintenance project.
  4. Enlist the experts.There’s an art—and a science—to structuring your data for optimal machine learning. Outside resources with experienced data scientists and proven manufacturing expertise are a wise investment. They’ll pay for themselves in strategy, best practices and hands-on implementation that’s efficient and intelligent.

Machine learning shows great promise for manufacturers—especially those who start now. The earlier you harness your data, the more time your machine learning application has for supervised and unsupervised training. Early adopters will gain a definite competitive edge.

Learn more about Predictive Maintenence 

Privacy Settings
We use cookies to enhance your experience while using our website. If you are using our Services via a browser you can restrict, block or remove cookies through your web browser settings. We also use content and scripts from third parties that may use tracking technologies. You can selectively provide your consent below to allow such third party embeds. For complete information about the cookies we use, data we collect and how we process them, please check our Privacy Policy
Youtube
Consent to display content from Youtube
Vimeo
Consent to display content from Vimeo
Google Maps
Consent to display content from Google