Usability in the Driver’s Seat: Simplifying Drive Data

Trend Detection Podcast ·

Show notes

Welcome to the Trend Detection podcast, brought to you by Senseye Predictive Maintenance – the platform which enables predictive maintenance at scale across all of your assets, across all of your plants.

You can find out more about how Senseye Predictive Maintenance can reduce unplanned downtime and contribute towards improved sustainability within your manufacturing plants, by visiting: www.siemens.com/senseye-predictive-maintenance

Transcript

Welcome to the trend detection podcast, powered by Siemens, and now your host, Liz McGinn. Hi everyone, I'm Liz McGinn. I am here um to talk about simplifying your data um and kind of, uh, that was a bad intro. I am sorry. Hi everyone, I'm Liz McGinn. Today we're gonna talk about um how Siemens simplifies the implementation and use of drive data. And I'm joined here by Dominic, our expert. Dominic, do you want to introduce yourself? Hi, this is Dominic. Thank you for having me here and talk about this topic. Um, I am, uh, at Siemens for 14 years now and I'm coming from the drives department, so you already saw it in the title. Um, I started my career as a, um, system test engineer for drive systems. I went over. To field service and, and, and commissioning engineer as well. And then I started to get into digitalization. So I'm working as a project manager for R&D projects in digitalization regarding drive technology. And now, meanwhile, I'm a product manager for an industrial edge app, um, also addressing digitalization for drive technology, especially in regards of predictive maintenance. Awesome. Before we get too far into it, I'm from the marketing world and I've heard the word drives in a couple of different contexts, usually, like, you know, a one, you know, your one drive. Can you explain exactly what you mean when you're saying drives? Yes, so in industries whenever something is in motion, normally it comes from a drive system. So, um, the easiest thing is you have a motor, an electrical motor, and in many cases you need that motor to run in a certain speed. So then you attach to it a frequency converter. So that you can run the motion like you want to have it. So mainly spoken drive technology is frequency converters and electrical motors in our case. OK, so it's both the motor and the thing that makes the motor go, or the, you know. Yeah, it tells you what you could call it like that, yeah, OK, at least that's what makes sense in my brain. So, um, why is usability such a critical focus for Siemens drive solutions? Yeah, so when you look at the uh topic usability, I have a great example in mind, um, an awesome example for usability, and that would be, for example, chat GPT. So it's, it's a quite, uh, I would say difficult technology, but it is packed in a way that everybody can use it. It is absolutely easy to use because you do the most natural thing in the world. You talk to the Um, uh, to your, uh, chatbot or whatever you are using. So it's, it's pretty easy. And when it comes to drive technology, um, of course, there are not that many experts in this, um, field of technology. And, uh, it is very important that you bring, um, the knowledge and bring all the benefits that, um, knowledge about this technology can have to the people, to the people that normally do other things like a maintenance manager who's caring about the, um, uptime of a machine. Um, he doesn't want to be a drive expert. He wants to be an expert in maintenance, and he don't want to get too deep into it. He wants to have a system that he can use easily. So in this regard, our target group are just the normal people in a, in a factory that want to keep it up and running. And furthermore, there is another aspect. Also, when you look at the demographic changes, so experts and well-experienced persons um are getting less and less because our, um, people get older and older and in these, these cases also, it is very important that you have easy to use solutions for all the younger people, all the unexperienced people also to um have the um benefits of this, all these, all this knowledge. And put in a technical solution. That makes a lot of sense from the, you know, demographic changes to sometimes like old technology can just prove extra difficult and you have to jump through many hoops to make it work. I remember like old video games where you had to like blow on the cartridges to get the dust out before it even started and I can't even imagine people trying to use that these days. So I can definitely see updating it to make it easier to use, not only from the new people perspective, but also it's annoying to have to keep, you know, Wiggle it just a certain way to make it work or however you have to do it. Awesome. So what makes Siemens so usable? Like how, how are we, are we chatbotting in that chat GBT function, or is it something, um, more simple? Yeah, sure. Uh, also chatbots are, um, often used in, in our solutions, um, especially when you look at the, um, Sensei predictive maintenance solution, a chatbot is navigating and helping you in any cases. Um, but not only chatbots are important, also the usability of the software itself. So when you're using an analytic software, um, not everything is done by a chat chatbot. Also, um, the, the commissioning and so on. has to be done by someone and the more easy it is, the better it is. And as I already told you, we have quite some experience in drive technology, so we have decades of experience and we want to use this experience and provide it to a customer and whoever, whoever is using our solutions in the easiest way. So the first point is we know our devices, so we know how to preconfigure. solutions, we know how to avoid complex setups and also we know what kind of data is within our drives, our equipment, and Wherever we can use this data even to avoid additional sensors, so to make it as easy to implement as it gets. Is the Siemens solution, do you need an like a Siemens expert to implement these drives, or can um customers use them on their own? So our goal is that everybody can use it. So when you want to analyze the drive data, um, it should be as easy as possible and of course, um, usable for anybody, for everybody. So, um, Yeah, you would have to have a cut, so I forgot your question a little bit. You're good. I was just um kind of implement like uh focusing kind of on this like usable outside the box, like, do you need a Siemens expert to um implement these, these drives? Um No, you don't need an expert. The solutions are designed that they are easy to use. For example, we made for the commissioning of the solutions a video, and this video takes 15 minutes. So then the complete solution is set up and running. Of course, in your first try, you will need some more time because you will have to search for some buttons or something like that. But when you have looked at this video and you take it as a guide, I'm pretty sure you can, um, do the commissioning on your own within a very short time. That's awesome that there's a video in the background because I know a lot of people use videos to do things. I know my dad, um, he fixes cars and he has to look up every time where to find my oil filter or whatever on my car. So, um, it's great that there's a video that you can just follow along. Yeah, I think that's that's the new way we have to think things and to also enable trainings for anybody. Everybody is using YouTube for when he wants to fix something, when he wants to get knowledge about something, and of course also in industries we should use this medium because it can bring very much information into a very attractive. A way of showing it and in a very compact way of showing things and you, yeah, just bring it on much faster than if you would only read a manual. Yeah, and there's a visual aspect of it that makes it easier, so you can say the button here versus trying to figure out a manual of the images and be like this isn't even, I can't even tell. It's like building furniture from IKEA. I'm like I have no idea what's going on. Yeah Awesome, I know you mentioned, um, a lot of the experience of Siemens and knowing the drives. How, how long have we actually been doing that? Do you know? Um, So it started with analog drives, I think in the 80s. I'm, I'm not really sure, so I would have to look up on my own. Um my, my first boss I had at Siemens, who was one of the first guys actually dealing with this analog drives, having to play around with resistors and so on to bring speed controls in, in, in an optimized way. So I'm, I'm pretty sure. 40, 50 years. Yeah, so basically the idea is we've been doing it pretty much as long as there's been some type of drives. In existence kind of does that sound about right? Much longer because we have, we have different generations, of course. So starting with the, with the analog drives and then we, we went into the generation of digital drives and and um starting with SIMmadune Simo Vat going over to master drive, micromaster, and then the newest generation, the cinemics and also in the cinemics we are already. Uh, not in the first generation anymore. So, of course, this is also an evolving topic, but being at Siemens, um, you have always these, these, uh, guys working with the old equipment and guys with the new equipment, and we bring them together to, um, exchange the, the knowledge there. And of course, this brings us the decades of knowledge that we can put into our solutions. Like I said, my boss, my former boss was uh in the 80s, a drive engineer, so he could also give some, some experience and some knowledge to me. It's good to know that we're not, um, kind of, I mean, obviously the goal is to make older equipment obsolete, but we still have experts that can help on those older equipments that are still running. Yeah. Exactly. You need also these experts, for example, when you want to run retrofits afterwards because someone has to know how the old machine was working and how you can switch it to a new type of machine. That's a really good point. I didn't even think of that. Can you provide a, a kind of a real world example of how, how we're using this technology? Um, yes, so we had recently a very Uh, yeah, I will start once again. You're good. Yeah, Um, we recently had a very happy customer that came to us and said he is using our solution and has now made the jump into predictive maintenance. It's a company in located in Austria. Um, and this is the biggest spice manufacturer of Austria, and they are using our solutions to have a look at their conveyor belts. They had the problem in the past that these conveyor belts got dirty from time to time, and they had to stop production, tidy it up, make it clean again. And then proceed with production. Now, nowadays they can observe the gratitude or the grade of um Of the dirt in the conveyor belts. And so that they, that they can react before this conveyor belt gets too dirty. So that's predictive maintenance in the clearest way how you can see it. So prevent a problem before it occurs. Yeah, is that using Sensi predictive maintenance or another Siemens technology? Um, no, actually they are using predictive service analyzer, so that's the product where I'm responsible for as a product manager. OK. And can you explain a little bit into what is PSA or predictive Service analyzer? Yeah, predictive Service Analyzer is an app for the industrial edge ecosystem, so it is an app that is running on premise directly in your factory. And, um, yeah, it consists of these benefits that I, uh, already told, um, at the beginning of our conversation, um, that it is a very easy to use solution, um, very easy to, to, um, commission, and it is especially designed for drive systems. So that's also, um, One difference compared to sensei predictive maintenance. Um, in Sensei, you have the holistic overview about your whole um factory and you can attach a lot of different applications. In our case, we're really looking very deep into drive systems. So that's our field of expertise where we have developed the solution and to give you the deepest insight in your drive system. Of course, based on our knowledge that we have gained the last decades. Oh, that's interesting that there's an expertise versus an overarching thing. I know Sensai is overarching, but it's interesting to learn the different solutions that are a little bit more specialized as humans. Exactly, and they come together perfectly, so they fit together perfectly because when you want to have the holistic overview, Sensai is the perfect fit. And when you want to go very into detail into a certain drive because this drive is crucial for for your production or Um, a downtime means very high costs. Then you can have the specialized solution on top and um have a deeper look inside it. Even extending the functionality of Sensi in this regards. Yeah, that's awesome to know. It's just another way of getting information for, for maintenance teams to make sure that everything's running smoothly because the last thing we want is anything to be shut down. Yeah. Awesome. So how does usability kind of go into that? You mentioned that it's easy to use. Is it, I know like Sensei prides itself on not needing data analytic like analytics or, you know, obviously it's data analytics, but you don't need to be a data analyst to use it. Is that the same with a predictive service analyzer or um exactly, exactly. So that's the same goal we had there. Um, we wanted to that everybody can use it. You don't have to be a drive expert. You don't have to be an analytics or AI expert because we have set so much preconfiguration, easy setup that you don't need that knowledge. We put all that knowledge already inside it. Is uh, PSA using AI? Yeah, it's, it's to 95% using AI and to 5% real-time simulation. OK, I didn't realize that it was using AI as well. Can you explain like what are they using AI for? mainly for anomaly detection. So we are looking especially at all parts of the drivetrain beginning from the grid where the frequency converter is attached to. So we have also observation, observation AI that is looking at the grid grid quality. We have AI looking at the status of the drive itself. So for the frequency converter. We have also artificial intelligence to look at the motor, especially, and also one very powerful functionality which is also bringing the benefit to the spice manufacturer is an artificial intelligence to observe the whole application. So to find out if there are anomalies in the behavior of the machine that could lead to a standstill or to a failure and to find it. In a very early stage to have the chance to react to it. OK, and what would, I guess, for an example, what would that kind of anomaly look like in in the spice, um, like conveyor belts? Would that be like a spice jar tipping over or something, or is it something? Something else. Um, so the AI is working in a way that it learns the normal behavior of the machine. So you're just in a in a first stage you're just recording data from this machine where the machine was running fine, where everything was good, where you don't see any problems, and when you say, OK, when it stays in this status or in this. Um, in these boundaries, then everything is fine. Then my day as a production manager or a maintenance manager would be just perfect. And you take this data, give it to the AI, and the AI is learning what is the normal behavior of the machine. So every working point, every recipe, every single scenario that this machine is seeing is then learned by the artificial intelligence. And in future, the AI gives you then the hint if there are deviations to the normal behavior. So in the spice manufacturing, For example, when the conveyor belt gets clogged or something like that, um, you could see it in different values of the drive, for example, the torque that drive is needing or the speed points that it is driving. So with these data points you can then see if there is something animal in the system. And then the story goes further. So when you know, OK, there is an anomaly in my system, I have to find out what, what is causing that. Um, you have something what we call an explainer. It gives you an indicator which value is causing this anomaly. So is it an animal torque or an animal speed value? And with this knowledge you can go and find out what the machine. What caused this behavior in the machine and the great thing is afterwards when you have gained this knowledge, what was the problem, you can also give this knowledge to the artificial intelligence once again and the next time the same scenario appears, the artificial intelligence will give you a hint. Hey, please have in mind in the past I've seen something like that. Please check this as a first measure. And also you can put some some more knowledge into it. So for example, how you got rid of this situation. So maybe you have to go there just by wiping with a with a cloth and have it clean again. Or sometimes it's even a little bit more complicated. Then you can also put in a step by step manual how to get rid of this situation. And here we come again to this topic of knowledge sharing. So when you have some very experienced persons in your maintenance team and some newbies, also these very experienced persons can give their knowledge over this database to the new bees that they are also capable of doing the job. Yeah, that's awesome that you have, uh, kind of that built-in knowledge and then being able to share that for future, future issues. I know that's really important, um. And obviously avoids a lot more shutdowns and stuff if you have an idea. And I know you kind of talked about costs there. How does um having this kind of implemented in a usability kind of format help the ROI of uh manufacturers? So, often AI driven solutions have the problem that they have a quite long, um, commissioning phase, a quite intense and costly phase where you have to implement everything, we have to set up everything, um, where you have to train your models and to retrain the models and, uh, go on and on until the solution is up and running. So this causes a lot of time. A lot of efforts and of course costs in these ways. And when you have a solution that is easy to implement and is set up in 15 minutes by some experienced people, um, and even by some unexperienced people in 1 hour, um, this is of course a very big factor when it comes to an ROI at the end. So I had a customer, um, that said, OK, if we even can find 2 failures a year, this solution has already paid off. When you see it in comparison with what um unexpected downtimes cause, um, as, as cost, and when you then see a very um easy to implement and, and, uh, cost attractive solution, uh compared to it, then often, uh, the decision is quite easy to uh implement it. Yeah, how long does it take to train the AI model with these solutions? I know in like uh getting it installed is easy, but how long does the actual process take for it to learn your equipment? This is depending on how complex your data is. So when you have a machine that is producing product A one day, the next product the next day, and the third product the 3rd day, then maybe you need 6 days or 9 days so that you have every product some data points so that the artificial intelligence can learn from it. Of course, if it is an, an easier uh application that only knows, um, let's say daytime and nighttime uh production, um, then the period of time that you need for training is, is shorter. But in fact, um, what you have to do at the end is uh exactly the same, it needs the same time, it is exactly the same work, um. In this time where the training data is recorded, you have to do nothing. You just have a listener there, so the AI is just listening what data is coming in there. After the listening time, the recording time is over, you have to go there as a user, set the time intervals where your production was running fine and where everything was smooth. And afterwards push start training button. That's all you have to do. You don't have to sit next to it for 2 weeks. You just install it, leave it there for 2 weeks, come back, give some time intervals where production was, was fine, and push start training. That was everything you have to do. …

Full transcript (WebVTT)