How Industry 4.0 Is Changing Our Way of Developing
Industry 4.0 is a digital transformation journey that needs vision, commitment, and value creation along the way. Each organization is different, and so is its journey. In this transformation, engineers must manage the growing complexity of software and an ever-increasing amount of data to create new business models and become market leaders.
The outline of the talk includes these topics:
- How to make the transformation effective
- What are the associated people skills, processes, and technology
- How to build mathematical models of multidomain systems
- How to deploy on edge systems/IT/OT infrastructure to fit in your organization's infrastructure
- How to use machine learning algorithms for predictive maintenance to derive actionable insights
- How the evolution of predictive maintenance is driving the change
Published: 9 Nov 2021
So I will be sharing some of my experiences in terms of learning from Industry 4.0 and what exactly is digital transformation from my perspective. And I also share some thoughts on what actually it takes if we are going to implement Industry 4.0 or digital transformation. What are the things that we would need to consider while designing and developing these systems, I'll share some of my thoughts on that, right.
So and wearing different hats as an engineering leader, how do I see Industry 4.0 as an engineer? How do I see Industry 4.0? We are talking about some of those things. And then finally, I will talk about how an organization, what kind of solutions MathWorks to offer when you are looking at Industry 4.0 and digital transformation, OK.
So first thing when I started, so thinking about digital transformation, Industry 4.0, one thought that came to my mind is OK, this is going to be something. A system would look something like this. When the word Smart comes into my mind typically, a few majors just run through my mind. It could be a Smart robot, or something like this or an autonomous car or it could be altogether a drone, right. So drone, which has some intelligence and which can act like a system, right.
Then I thought OK, instead of having my own perceptions OK, let me go and ask, right. So then, obviously like many of you, right-- so I just went to Google. And then asked OK, what is this Industry 4.0 and what is this all about, right? Digital transformation, and what the industries are trying to do here, right.
And then what Google told me is-- so a couple of sites in that Google told me gave me two keywords after I spent some time understanding the Industrial Revolution One, two, three, four, and then I finally got two keywords, right. So one is cyber-physical transformation, right. That means there is some transformation that industry is trying to digest and adapt, which would lead to digital convergence in the ecosystem.
As some of the previous speakers identified, timed, and gave some thoughts on what actually this ecosystem. And then how do we integrate multiple players involved in this ecosystem with this digital transformation, right. But this didn't clarify all the questions that I had. Then I started wearing, OK, as an engineering later if I see, what is in it for me in this, right. So why should I care something called Industry 4.0?
Then again, it's a lot of text and so on. But I would like to highlight a couple of things here, right. So if you are seeing this whole thing as a winner or a leader of this industry, right-- so basically two, a couple of highlights here. It will actually help us to manage and optimize manufacturing processes.
And if we can draw the line all the way back to design, it's a bigger opportunity to actually transform the whole scenario. And with enough data and insights that are available that would come to us from the assets, if I can call them smart assets, and then what kind of decisions that we can make based out of tons of data that would emerge from these assets, right.
And then the second way of looking at this is what it means for engineers, right. So why do engineers care for this, right? What kind of opportunities Industry 4.0 or digital transformation deals to engineer, right. So in a way, this is like the complete-- I don't say complete, but almost a full list of technologies that are involved in this, starting from the simulations, right, or developing the models and doing the system-level simulations.
And then with various IoT technologies and having sensors placed, collect the data, and stream them onto your OTR ID infrastructure and then process that data, right. But it's not going to be as simple as this, learning this technology and getting onto the field, right. So it comes with its own challenges here, right.
So what are a couple of challenges? One is, how do we manage as the systems are getting more and more complex, with a lot of software getting onto, it right? So how do we manage these systems through the lifecycle, starting from capturing the system requirements, implementation, doing R&D, and deploying into the field these assets?
And then maintaining these assets in the factory environment, right. So how do we manage the complexity involved in these systems and what are the associated processes, right? So that would help us to optimize these things, the entire system here. And then third thing is invariably there is a push, right, so from the industry and from the customers.
So sample size 1, so there that is kind of customization that people are asking for. And but when we just take one step back and in a lot of industry is this kind of customization possible, right? Then it is something like actually, no, one size doesn't fit all, right. So stay tuned for some more minutes I'll walk you through what exactly that means and what kind of precautions that we need to take care.
But to address, right, so some of the challenges that we discussed, and for the sake of simplicity I broadly classify these challenges into two categories. One is related to complexity of the system and second is related to data handling, right. And then later I will show you where exactly these things fit in Industry 4.0 ecosystem in terms of the manufacturing processes.
So now let's start with the complexity, right. So first question here, so how do we handle complexity of these systems involved here, right? Say, considering the picture that you're seeing on your screen, a typical standard software or system development lifecycle, right. So if it change that comes to us, right, so once the systems are deployed and now a change coming to us-- so starting with the system requirements.
How flexible our existing design methodology is to absorb that change and then propagate it all the way to the subsystems teams and implement that change and once again take it back to the systems team and analyze whether that is implemented or not, right. And are we able to establish thread all through this process, right. And continuing that, so how is this connected to the systems that are deployed in the field, right.
So is the agility or flexibility that is built into our methodology, can it be extended all the way to the field? And are we able to bring back some of those insights back to the development team in a way to improve the performance of that design process of the system, right? So now considering this, I will show you some of the steps that as an organization MathWorks supports, what we have observed based on our experience with some of the global large volumes, right.
So there are a couple of steps how people do simplify this whole process, right. So the first thing is starting with system modeling, right. So design the plans, design the controllers, and make use of the library blocks that are available, either in Simulink Simscape, or various related tool boxes that we basically offer. So design your whole plan, design the complete system, including all the entities.
And if you have CAD drawings, you input those CAD drawings, and then you'll be able to do a quick simulations incorporating your CAD drawings into the loop here, right. And if you already have a system that is defined and you will be able to understand what kind of behavior the system is going to give, right. So this is kind of like reverse engineering, I would say.
No once you are done with the complete system design, the next step would be doing the desktop simulation, right. So what are the functionality that you're building, plus what are your existing functionalities that combine the whole thing, and then do the simulations. And of course, you're expected to do the reentry aspects, there, right.
But most important point here, right, so which would lead us towards transformation, our Industry 4.0, whatever you are developing on your desktop, it's not going to be closer to your performance on the hardware. There is going to be gap, right. The question is, can we simulate some of those aspects in the simulation world, right. So that our designs are pretty closer to the hardware results.
That's where something like virtual commissioning comes into picture, right. So what our designs that you have done, so create all those scenarios and try to virtually commission all the assets that you created. And then see how these systems are behaving. In this case again, just to as a bottom line, you are still in the simulation world, right. We haven't gone to the hardware yet.
As a natural next step to move much closer to the hardware, we redeploy the whole thing in the hardware, using hardware in the loop kind of simulations, right. And then put the plant in the hardware and then still run the controller from your software and then make sure the systems are working, right.
And then after that, once we are happy with all the results, then we generate the EXEs and then deploy into the hardware, and then see how these systems are working. So this is one way to handle the complexities of the system, right, the complexity associated with the system. But this only solves half of the puzzle. The other half related to the Industry 4.0, Is how do we deal with this data, right?
So that's where I think some of the previous speakers shared a couple of examples. I really like those examples related to the predictive maintenance aspect. Dr. Satish Reddy, in his inaugural talk, talked about AI, so the significance of AI, right. So all those things you know originate from, how do we capture, and how do we analyze this data, right. So as I am today, if you see in the world, the maturity of the industry. We have tons of data that's available in case if you don't have data, we can actually know synthesize that data using various tools, including MATLAB and associated products.
We can generate that data and with the availability of existing powerful algorithms that are available, we should be able to get insights from this data. So that's where using this data. So we can develop the concepts of digital twin, so where you can actually get the field data and trying your digital models, right. So to tune those models and then you will be able to reproduce some of the errors that you see in the field, right.
So you'll be able to introduce those errors in your digital models and then you will be able to see how to control those errors and of course always onboarding new people is also an important factor. So this is one side of the coin, right. So, so far what I was trying to talk here is what are different challenges associated with Industry 4.0, but when we actually move one step towards implementation, where exactly this is going, this will leave, right.
So what are all the different phases involved in dealing with complexities of Industry 4.0, right? So from our perspective, there are four major phases, are four major blocks in the ecosystem, are in the architecture of Industry 4.0. I would put them this way, starting with the Smart assets that we deploy in the field or and the manufacturing on the floor.
And these Smart assets, right, so you will do the design development we envisioned all those things for the creation of the software to define the behavioral aspects of the Smart assets. And sometimes, considering the processing ability of these Smart assets, all the processing you'll not be able to do on the asset alone. So you would like to dispose some part of the processing onto your different device, that's what here I'm calling it as a system.
So its system would be sitting pretty closer to your asset, right, so the computational abilities between Smart assets and your system could be distributed. Or sometimes you may want to do the complete processing only in the data system or on the Smart asset, based on what problem you are trying to solve. What's the business case for you?
And now, once this processing is done and then we would like to upload this data and to work the infrastructure operational technology, often your work infrastructure could be on premises. For example, let me take one simple example of an elevator. We closely work with Schindler, well known for their elevator products, right.
So what they have done in this example is, so, the voting infrastructure resides in the same building, which means from the Edge system they upload the data into what infrastructure, either using, you know, they may have their own servers or it could also be a cloud. But when you are looking at thousands of these kind of buildings all over the world, then they bring that data to one central system, upload onto the cloud, and calling it as IT system. Information technology system, right.
So how do we see these things when we look at from a factor viewpoint, right? This is how I would like to simply differentiate these things, in terms of speed versus scope, the significance of this data. When you are trying to control a Smart asset, we need to control it in real time. So that decision should happen then and there right. So it's going to be very hard, real time control, right.
So as you move from left to right, so when you move towards IT systems, you would be capturing that data for multiple days, multiple weeks. And then after gathering enough data, you would like to look for some patterns, and then push back some changes all the way back to Smart assets. This is where the initial part I was trying to say-- afterwards in the IT systems, if you are looking for some patterns or if you want to push some change in their design methodology of Smart systems, right.
Our existing design methodology should be agile enough to capture that change quickly. And then as you move from left to right, on the picture here, the importance or significance of that data would actually increase. So then how the world is attempting in order to implement different phases in Industry 4.0? So I would like to take a lot of some of our third party partners as well here.
So the first two categories, Smart assets and IT systems, a lot to do with the development. And once you are done with the development, based on your target hardware you will generate the code, either in CCPP or HDLM and so on, accordingly you will select the right platform for you. And then from there, when you are planning to stream the data, into either what your IT systems you would be using some streaming services either Kafka or the equivalents, Event Hub, right.
So you're using one of the streaming services. And from there you will upload the data all the way into cloud using either your own private cloud or maybe using some of the well-known cloud systems, like Amazon or Microsoft, right. Then where does-- actually we can help you in this process, right. Looking into this ecosystem, what is our role, where MathWorks can actually really help you, right?
And it's not new to MathWorks, right. So we have strong solutions across this ecosystem. From left to right here, I will try to highlight some of the things here. Talking about the first two segments here, this is widely used and well-adapted in the industry. Starting with the model, using model as a source of truth, doing the code generation based on your target hardware, and then deploying into the hardware, right. And then establishing different interface protocols between multiple systems, right.
And as you move slowly towards right of this, so we actually interface with all the stream processing. So we have a couple of products. So if you are interested, please do get in touch with us, we would be happy to walk you through some of those solutions. But the bottom line here, so what's the point that I'm trying to convey?
So assuming you are doing the development in the world that I just mentioned, you should be able to carry that design, and export that data into different systems being in the same environment, right. So to enable that seamless integration, that's where the value proposition of the solutions that I just mentioned would kick in.
So using the same workflow, using MATLAB Production Server to deploy onto your production hardware, and using MATLAB Parallel Server to speed up your computations as in when you're actually pulling tons of data, DBs of data, right. So I would like to show you a couple of examples, not necessarily coming from aerospace and defense industry, but I thought I would pull different examples from multiple industries so that you get a real sense of how the world is actually adapting Industry 4.0.
This example is actually coming from mining organization in Australia. Again, you know, all the examples that I'm showing here we closely worked with all these organizations, right. What these guys have done, as usual, just as I mentioned, using model-based design MATLAB and Simulink, they have done the complete the design and development. And used Jenkins for continuous integration, and with MATLAB Production Server deployed onto the field.
And what they have done for OT, Operation Technology, they have kept hardware on premises. And for information technology, they have used Microsoft Azure, of course, along with their respective databases here. And for the customers, they enable the Power BI reports. This is one such example.
And in the second example, as I was telling you in the beginning with Schindler, so pretty similar, right. Again, a combination of multiple technologies, multiple tool suites, right. But MATLAB and Simulink was actually helping them to set up the entire framework here.
Another example from Atlas, a well-known company for the design of compressors here, the framework is pretty similar, if you look at the screen here, right, using model-based design and then various things involved here. But what I would like to highlight here, is end of the day, what they have achieved hear.
With that framework, they were able to connect 120,000 machines worldwide. And again today, if we see, they're actually scattering data from these many machines worldwide and pulling this data into the information technology system and then analyzing that data and looking for how to make business decisions out of that data insights.
One more example, so for as Doctor Satish Reddy identified in his early talk, so bringing in EDO is very important for us in this ecosystem, at least in this latest technology, right. So this is one of our collaboration with IIT Delhi. Of course, there are many other players in this, so IIT Delhi is actually setting up a lab Facility for Smart Manufacturing, FSM, right.
And we are closely collaborating with IIT Delhi, with the Dr. Sunil Jha there. And I have provided the link here. If you are further interested, go through that link and you can get more insights on exactly what is happening here, right. So how they are trying to come up with this technology demonstrator project.
And then let's take one example from aero. This is coming from Lockheed Martin, how they have actually leveraged the whole digital transformation. So this is actually coming from one of their base. So how they have used their digital transformation techniques to improve the fleet analytics. So they have a number of aircrafts in their base, and they have taken multiple variables, including the availability of the equipment, right.
Availability of the pilots, the availability of the system, considering various factors, they were able to come up with an algorithm to figure out or to improve the performance of the fleet there, right. And it is implemented and are using the solution, right.
So with that, I would stop here. So as a summary, if you see, right-- so Industry 4.0 can be seen different by different people, right. And based on my learning, my study, my understanding, one size doesn't fit for all, right. So if you are trying to implement Industry 4.0 in your organization, please be clear about what is the problem that you are trying to solve, right.
Are you going to take all the way to information technology, and are you trying to solve a bigger problem? Or are you trying to bring in some kind of transformation on your production floor, right? So based on the, stringency the complexity of this problem could actually vary, right. So with that, I would stop here. Once again, I sincerely thank Aeronautical Society of India for giving me this opportunity to share some of my thoughts. Thank you so much.
Thank you Mr. Satish for the interesting talk. You have covered within the short period very specifically covering Industry 4.0 technology, covering systems requirements, functionality, architecture, design, et cetera, model-based engineering design, then finally helping the Lockheed Martin Warrior software development. Thank you very much, once again. Now--