How the Industrial Internet is Going to Save Trillions of Dollars

Like the name says, the Industrial Internet (of Things) or IIoT is a field that has a strong industry focus. So it’s not about smart watches or smart homes like the more general Internet of Things. It is all about increasing profitability through asset management and optimisation. What that means exactly, I will explain to you in this article.

The 3 key components

Machines, analytics and people are the key components of the Industrial Internet.

1) Intelligent machines that are connected and use advanced sensors, controls and software to make use of data for other tasks then just to operate the machine.

2) Analytics create new insight using three distinct methods. Predictive analytics for instance uses machine learning to predict equipment breakdowns before they happen. Physics based analytics solves problems where a formula can be written to describe the physical context. Also domain knowledge is important. It has been gathered by the company in the past because of the experts working on the products. For instance measurement data from development and production where the experts know exactly how to interpret it.

3) People are also a key part of the Industrial Internet. Wait what? People? Yes, they are the ones who work at facilities to support operations, do maintenance or design products. Without them the Industrial Internet does not work.

From data to insight

Machines create data while they are running. This data is then transmitted to storage systems, for instance in the cloud, that rely heavily on big data systems to handle all the produced data. The problem is: Raw data is quite useless on its own because the needed insight into what is really going on is hidden in the data. Someone needs to extract it. For humans it is an impossible task because of the sheer size, velocity and complexity of the data.

This is where advanced analytics comes in to save the day. Advanced analytics through the afore mentioned three methods machine learning, physics based analytics or domain knowledge. You should definitely look into machine learning because it is extremely powerful and can produce insight that was never available before. I have planned an article in the near future that will show you on some examples what machine learning is, what you can do with it and what the future of machine learning could look like.

A quick recap. Data has been produced by intelligent machines, stored in the cloud and processed by clever top notch analytics to create insight like never before. Now comes the part where someone needs to do something with that data.

Connect data with the right people and machines

Now that the analytics data is created, you need to make it available to someone who will act on it. In the hyper connected world of the Industrial Internet it will be someone within the network of people and machines. For that, you need to create a platform for data visualisation and communication. This platform needs to visualise analytics and machine data for people and also must have interfaces for machines to connect and share information. Basically, it is about getting the right information to the right asset (a person or a machine) at the right time.

The feedback loop of continuous improvement

When you put all of this together, you create a feedback loop where data flows from the machine to analytics, through the network of people and machines and then back to the machine. The machine then uses this feedback to improve its own operation. This feedback loop is something totally new. It is in my opinion why the Industrial Internet business is supposed to help companies save trillions of dollars world wide.

I would recommend you to deploy your IIoT solution in two steps:

1) Build a platform to connect machines and people, to get the right information to the right asset at the right time
2) Get the feedback loop going to enable the machine to self improve based on data

About the Author

Andreas Kretz

The IoT and Big Data is my daily business and my passion. How can I help you today?

Comments 2

  1. I have been following your posts from almost 4 months.
    I would like to ask you about a study advice
    Is it better to study Python and the packages deals with Big data such as Skitlearn etc… or should I study the Hadoop echo system with all this tools and solutions.
    The reason I’m asking that I can see a very rapid change or new tools and old tools that vanish in less than a year
    Everything is being take over by Spark and you can use R or Python or even SQL to deal with Spark.
    Hope you come back to me soon.

    Best regards
    Mohammad Rabie

    1. Post

      Hi Mohamad from your comment I figure that your want to focus on doing analytics. Not actually building big data systems.

      As a data scientist you should learn what big data tools do. Like Hadoop (mainly HDFS), NoSQL Databases, Kafka and Spark.

      Not in depth, just to the point where you understand how it works and for what you need it. Because you will use these tools to get data to analytics, or store output data from analytics.

      As for the actual analytics and machine learning, it depends a bit on the platform you are working on. A safe bet is to learn how to use python and packages like Numpy, Pylearn2 and Pandas.

      Did that help?

Leave a Reply to Mohammad rabie Cancel reply

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