IIoT in Manufacturing: A More Intelligent Approach

Share on twitter
Share on facebook
Share on linkedin

If you’ve been anywhere near manufacturing or capital-intensive industries in the last several years, you’re sure to have heard the term Internet of Things (IoT).

More recently, the term Industrial Internet of Things (IIoT) has come into usage. But how are these terms different? And what’s the best way to think about how you can use these new technologies to improve safety, conduct cost-effective maintenance, and ensure product quality?

In this post, I’ll explain what the difference between IoT and IIoT is. I’ll also talk about why there’s a better way forward than either of those two options—the Intelligent Industrial Internet of Things (IIIoT).

Before we talk about industrial or intelligent internets of things, let’s start by examining the normal Internet of Things (IoT), both as background information and because this kind of technology is becoming common even in our daily lives.

The Internet of Things (IoT) — Impacting Our Daily Lives

The basic idea of IoT is that we can equip things—the technology around us—with the ability to record, store, and exchange data with other things. It’s the simplicity of this concept that makes it so powerful, and we’ve seen IoT features popping up with things like security systems that automatically know when you leave your house, irrigation systems that adjust how much they water based on moisture and predicted rainfall, and thermostats that process data from different sensors to learn how best to heat your home.

One particularly illustrative example of IoT can be seen in Vorwerk’s Thermomix, probably the best known IoT device in Germany. The Thermomix is a kitchen machine that integrates mixing, chopping and heating functionality with a suite of temperature and weight sensors.

It also connects to the Internet. This means you can download a cake recipe directly from the cloud onto your kitchen machine. The Thermomix then helps you make the perfect cake by telling you what to add when and when you’ve added enough. Then, it takes care of all of the mixing so that you end up with the perfect cake every time.

The IoT devices that are finding their way into our homes are helping to make us more efficient and take some of the grunt work out of our daily lives.

The Industrial Internet of Things, on the other hand, helps us outside the home by supporting industrial processes.

Are you responding properly to opportunities & pressures presented by Industry 4.0?

Don’t get left behind. Follow this blueprint to learn how to optimize every segment of your manufacturing business with AI.

The Industrial Internet of Things (IIoT) — BI for Industrial Processes

If we can connect sensors to the Internet and get information out of a kitchen machine, we can do this with industrial processes as well. This is happening in factories and workshops all over the world every day. IIoT is used to monitor machines and their outputs so that on-the-fly adjustments can be made. However, the environments and requirements for industrial processes mean that there are unique challenges present that don’t occur in typical IoT domains like your kitchen, including:
  • It can be difficult to take measurements in hot, wet, or oily environments.
  • Machines may be old, so sensors and communications might need to be retrofitted to work with other technologies.
  • There’s no guarantee that a given site will have a strong network or cellular connection for transmitting data (although this can be addressed with edge computing technologies).
  • Models used to process and analyze the data can drift over time.
With industrial processes, it’s also much more critical to ensure that the measurements taken are accurate than it is if you’re making a cake. Both profits and health and safety can be on the line if your sensors aren’t providing accurate information about what’s happening on the floor. A fascinating example of IIoT comes from Achenbach Buschhütten, who presented their system at our Industrial Data Science conference in 2018.

Achenbach Buschhütten (AB) is a 500+ year old company, and they’re one of the market leaders for building machines that produce aluminum foil. Recently, AB equipped their machines with devices to stream data to the cloud. This way, they can offer additional digital services as an add-on to their machines, while simultaneously increase the reliability of the machines via constant performance monitoring.

From a data science standpoint, it’s interesting to see what is being done with IIoT technologies. However, a lot of the data analytics being done with IIoT reminds me of business intelligence, which is to say that it’s backwards looking—it tries to understand what has happened in the past, not what is going to happen in the future. For example, data might be analyzed using approaches from statistical process control to understand if a machine is still operating correctly and how it behaved on the last x-runs.

But to really get the most out of the IIoT revolution, we need to change our thinking. We should be turning to the past only to get data that we can use to predict the future.

Enter the Intelligent Industrial Internet of Things.

The Intelligent Industrial Internet of Things (IIIoT) — The AI for Industrial Processes

The Intelligent Industrial Internet of Things (IIIoT) brings your industrial data analysis to the next level. For example, on a shop floor, data is usually captured at various processing steps. An action taken in the first step, like drying the product longer, may have an impact on a later processing. You need to connect the data from different locations so that you can get a full picture and take action.

But the data captured by IIoT is notoriously noisy due to a combination of factors, including:

  • Every sensor measurement has its own level of uncertainty.
  • Some sensor measurements might not even be available for some parts of your workflow, either because you cannot measure a relevant property or because the thing influencing the creation of the product is not part of the process—for example, a delivery delay from a supplier.
  • Depending on the bill of material you are processing, sensor values may be interpreted differently.
  • Two products you create may come from the same batch of ingredients which means they are, statistically speaking, not independent of each other anymore. This needs to be carefully taken into account.

The good news is that data scientists are used to working with these challenges, and data science algorithms are made to work in noisy environments. The RapidMiner platform is uniquely positioned to use the domain knowledge of process engineers and include it in their analytics process in the most impactful way possible. Machine learning is essentially the discipline which combines information from disparate sources, thickens it, and returns a condensed, precise insight.

So what’s the best way to use these algorithms to drive value for your organization? Let’s look at two examples of how being intelligent about your IIoT implementation can benefit you.

Find better recipes or conditions

Let’s go back to our kitchen machine and making the perfect cake. This example is actually easily turned into an industrial use case. But instead of the quality of cakes, we’re looking at the quality of our product. What will the NaCl concentration be? What will the viscosity be? What will the throughput be?

Note that these are forward-looking questions. They’re not asking what the viscosity is right now, but rather what it will be like later, so that you can make any adjustments you need ahead of time and ensure that you’re maximizing both outputs and profits.

You can model all these questions with respect to the sensor data you read from your IIoT devices, including temperatures, currents, or even images and audio files. Using a versatile platform like the RapidMiner AI Hub, you’re able to capture all of this information from different sources and use them in one unified, forward-looking process.

Combining information from different processing steps

Another use case for IIIoT is the combination of information from different sites and processing steps. Most products run through many different steps during their creation, and often consist of different materials. But the people on the shop floor can’t know everything what’s been done to those materials in previous steps of the process. Plus, it’s unlikely that everyone on the floor is an expert on the materials used.

But with intelligent algorithms, you can help the people on the ground. Using machine learning with data from distributed sensors, you can predict the properties and distill the important information about your bill of materials. This empowers folks on the shop floor to make the best decisions possible about what’s happening and what actions they should take. This leads to better quality, less spoilage, and even higher security standards.

Wrapping Up

If you’re thinking about experimenting or expanding your use of the Internet of Things in your business, you’re on the right track—these technologies can have real impact on your efficiency and bottom lines. But if you’re thinking about how to use or improve your use of these technologies, make sure you’re doing it the intelligent way, and not just applying traditional BI methods to the treasure trove of data you’re generating.

If you’d like to read more about some of the ways that AI and ML are impacting the manufacturing industry—as well as others—check out 50 Ways to Impact Your Business with AI.

Looking to drive real business impact with AI?

Sometimes the most difficult thing is simply knowing where to start. Identifying impactful use cases is one of the most cited roadblocks for organizations seeking to leverage AI.

Learn from these 50 use cases across all industries.

Additional Reading


RapidMiner Virtual Optimizer

In this webinar, we cover the topic of process optimization through the lens of RapidMiner’s virtual optimizer—a real-time prescriptive dashboard that lets workers understand the current state of your operation, experiment with potential changes, and then implement the best solution.

Read More
Martin Schmitz

Martin Schmitz

Martin Schmitz, PhD is RapidMiner's Head of Data Science Services. Martin studied physics at TU Dortmund University and joined RapidMiner in 2014. During his career as a researcher, Martin was part of the IceCube Neutrino Observatory located at the geographic South pole. Using RapidMiner on IceCube data, he studied the most violent phenomena in the universe like super massive black holes and gamma ray bursts. Being part of several interdisciplinary research centers, Martin dived into computer science, mathematics and statistics and taught data science and the use of RapidMiner.