AI Predictive Functionality: Data Analysis for Productivity

Carlos Becerra:

Hello, and welcome to this IMTS Spark presentation. Today, we will be talking about Okuma's artificial intelligence applications.

Carlos Becerra:

Okuma has a collection of high-tech tools that analyze data from a variety of sources, with the purpose of predicting machine failures and improving productivity. But before talking about the specific tools that Okuma has, let's just stop for a minute and try to define what artificial intelligence is.

Carlos Becerra:

At the most basic level, artificial intelligence basically refers to the ability of a machine to be able to make a decision without the direct input from the user. For example, if you are working on a CAM program and you are ready to produce the GN code, you go out and you basically click Go. And the computer is going to take maybe a good amount of time, but it's going to do essentially only one thing, which is to generate the GN code. And it's going to do that only because you clicked on the Go button, and that's exactly what it is supposed to do. It's a little bit difficult to come up with good AI examples from the real world, because AI is still very much a new technology being developed.

Carlos Becerra:

But we already have, for example, a few cars that will stop automatically. And so if you're driving up to a certain speed and you are driving the car (we're not talking about a self-driving vehicle), it will override your command, and it will stop the vehicle under certain conditions, because it knows that if it doesn't, it will collide with a vehicle in front, and it determines that based on a number of input sensors: its own speed, the distance to the car in front, maybe a variety of other things, as well. So it's doing something without the user telling it to do it, and it's doing so based on just analysis and computations and the computer taking over and concluding something. Although it's very basic, that is a good example of artificial intelligence.

Carlos Becerra:

Now, artificial intelligence can be classified into a number of different levels. The most basic of which is basically that the computer uses a lot of data that it has in store and only looks at that data. And based on the analysis of that data, probably in the variety of different ways, it comes to a conclusion. And that is what we call reactive machines.

Carlos Becerra:

Next, we talk about machines that not only utilize what it already has in store, but it also looks at sensors and it takes into account what happened in the past. So it can start now thinking about trends. If the vehicle was traveling at decent speed before, most likely right now, you can have a curve and extrapolate that. And so between those two, now it has more of a larger view of the world to make a decision about what to do.

Carlos Becerra:

Machines can also utilize data and learn from it. It uses the data in one way and their outcome is that. And it uses the data in a different way; the outcome is different. Whether the machine has the ability to learn and adapt from that basically determines whether it can use that outcome and put it also into its set of data that it is going to look at the next time. So far, all these first two and maybe two-and-a-half levels is where that technology resides right now. This is more or less where the state of the art is.

Carlos Becerra:

The next couple of levels are basically out in the future or are being developed in research institutions or things like that, and the first one deals with where the computer, the machine, is aware that others that it interacts with also have the ability to react and do things on their own. And it's like if you're playing a game, for example. You know that after you do whatever it is you do, you take your move, the other person will do something. You don't know what it is, but you know that that is going to happen. So that's what is called theory of mind.

Carlos Becerra:

And the last category belongs to machines that are aware of themselves, just like you and I are aware of ourselves. These category really resides, for now, in the realm of science fiction.

Carlos Becerra:

Okuma's system basically uses a set of preset data and then inputs from a variety of sensors. And then of course, it has its own algorithm to determine and make a prediction, either about the health of the particular system that it's looking at, or the opportunities for improvement, in terms of utilization of the machine. And we're going to dive into more detail about what all that is, in the coming slides.

Carlos Becerra:

So the first of Okuma's AI technologies reside within the machine, part of what we call OSP Suite. And each machine looks at only itself and the data that it, itself, generates to make the decisions about the health of a couple of very critical machine subsystems.

Carlos Becerra:

The second tool works as part of Connect Plan. Connect Plan is the tool that looks at all the machines that you have connected in your plant and is able to present to you data relating to utilization and productivity and things of that nature. And what Connect Plan does is exactly that. It is able to show you that data and then take input, not only from the machines itself, but also the user, to learn how the machine is being utilized and present that data back to you, so that you can make decisions about how the machine should be utilized.

Carlos Becerra:

So the first of the two subsystems that the OSP AI looks at deals with feed axis. It looks at data from the ball screw itself and the bearing that supports that ball screw. And depending on that, and also historical data and a set of parameters, is able to determine and present to you a trend, to basically tell you whether the machine is perfectly good condition or it is trending toward needing maintenance. In a similar way, the system can also look and do the same analysis for the spindle, and looking at trends and looking at data, it can also tell you if either the spindle is good, is in need of maintenance in the short-term, or is already in need of maintenance right now.

Carlos Becerra:

Okuma's AI tools can connect to the Okuma cloud through Connect Plan, and in that way, maintain the set of parameters that we have been talking about. Maintain those parameters, totally update, so that the model that it uses to determine the wear and the condition of the subsystem is always optimal.

Carlos Becerra:

Now, using these tools is extremely simple. Starting from the OSP welcome screen, you basically just click on the tool, and the interactive AI screen appears, where you can see, basically, the historical trend of previous analysis. You can look at them individually, and you can perform a new analysis, for which you first select the axis that you are going to work with and start with the analysis itself. When the analysis is completes, it's going to present to you the results of that analysis, one of which could be that the machine, like it shows here, is in perfectly normal operating condition. It could also tell you that the subsystem is already in need of some sort of maintenance or repair. When it gives you red, that means that there is something that you need to take a look at.

Carlos Becerra:

Or most importantly, it tells you that something is going to be needing some sort of maintenance. When it presents to you yellow, the machine is still in perfectly good condition, but because of this analysis that it is performing, it is able to conclude that if nothing is done, the trend shows that something could become broken or in need of repair in the future. So this is probably the most important result that you can see, because at first sight, judging just by the normal performance of the machine, the machine is in good condition. But this tool is able to look at the future, basically, and tell you that it is not trending in the right direction, and you should call for maintenance, call for service, and see what's going on, and maybe perform repairs.

Carlos Becerra:

Now, some of the benefits are obvious, right? Obviously, if you are able to look into the future, you're not going to have to stop the machine unexpectedly and have downtime and have to run around trying to get service all of a sudden, because you didn't know that something was going to break.

Carlos Becerra:

But it's not just the annoyance of having it down. It's also the ability to be able to schedule, ahead of time, when is the perfect time to do the maintenance. When you get the yellow, it's not that something is going to break immediately. Depending on the utilization, you may have a number of days. So now you can say, "Alright, I don't want to stop the machine right now, but I do want to stop it. I can stop it tomorrow at 12 o'clock." And you plan that way.

Carlos Becerra:

Generally speaking, today, maintenance is performed based on a set schedule or a fixed period of time: every six months or every time the plant is on shutdown, things of that nature. But that may mean that you're going to do repairs and maintenance or even have downtime when it is really not needed, and so this tools allows you to do and to spend the money and have the resources down only when it is really necessary. So obviously, you save money, because resources remain productive and are not down unnecessarily. And obviously, also because you don't have to pay somebody to come over and do any kind of maintenance on your machine when it doesn't need it, or spend money on parts that the machine does not need.

Carlos Becerra:

The other AI tool that Okuma has available works through Connect Plan. So as I said before, Connect Plan collects data for the different status that the machine goes through, throughout the day, and presents the data to you in a graphical form, color-coding the different status. Things like stopped or cutting or alarm, operating, etc, etc. When you have the AI plugin as well, it has a lot more detailed visibility about the different status that the machine can have and presents them to you. And it also allows you to do very detailed drill-down analysis on each of them.

Carlos Becerra:

But the most important portion of all this is that it allows you to enter a specific input for a specific period or periods of time that the machine cannot know by itself, so that that analysis is even more detailed. So the system uses that input from you and plugs it in, going forward. And that then becomes part of the data set that the system utilizes to tell you what the machine has been doing. So this allows you to have a better set of data to determine what the machine has been actually doing, and whether you need to take action to improve your productivity.

Carlos Becerra:

This concludes our presentation today. Thank you very much for your time. If you have any questions, feel free to ask now, or send us an email or contact us through our website. Thank you.