 I'm very happy to be here today, and Ligres as well. So I'm going to talk about what I'm Ricardo, Simon Carbajo, and the head of innovation and development at Cedar. And today, I'm going to talk about what is Industry 4.0. So I'm going to just give an overview of what Industry 4.0 is, and then I will touch the area of predictive maintenance. How many of you have heard of predictive maintenance here? OK, that's good, that's good. OK, so it won't be that technical, and I will try to provide some detail of what we do. But before, allow me to introduce Cedar, which is the center where I work. So Cedar is the National Technology Center for Applied Data Analytics Research. So the story about Cedar was created around six years ago as a strategy from the government to help industry. So 15 centers were created, for example, one in pharma, another one in dairy, meat, ICT, cloud. So our center was in data analytics, sort of machine learning, statistics, all of that. So we are funded by Enterprise Ireland and the IDA. And we sit in two universities, in UCD and in DIT. UCD, we have headquarters, but we have collaborators in both universities. The fact that we sit in the university doesn't mean that we do only research. Everything we do is applied research, so more towards industry. So our goal is just to help with the challenges that industry faces. So we work in AI, obviously, but big data analytics, technologies like machine learning, big data, predictive analytics, real-time analytics, deep learning. It will be the future of AI, deep learning, so we're putting a lot of emphasis in that, and blockchain. So we have kind of like a model of member companies. So we work with a lot of companies in different verticals. We have, for example, as you see, big consulting firms. We have Accenture there. We also have Optum in there. So multinationals there. We also have Johnson and Johnson, who we'll be presenting later. Electricity, Siemens, top companies. And at the moment, I think we have 50% multinationals and the other 50% SMEs. So the idea is just that these company members come and get the benefits of CIDAR. We can work with other companies as well. But they benefit from these companies. They benefit from meetings that we have, regular meetings, regular events, so big meetings where they can do a lot of networking. We talk about data analytics. We talk about GDPR, any kind of area which is related. And they do a lot of network, which is very good. Then they also benefit from accessing funding. So there's a lot of mechanisms to get funding, enterprise and so on. So when they liaise with us, we prepare the projects and then we access the funding. We leverage this funding for them. And then we have more than 40 prototypes that we have developed over the years. And these prototypes arise as a necessity, or as a proposition of the challenges that the industry faces. So our companies vote and they say, this is the main challenge and then we go and execute them. And we build that as a prototype and other member companies can actually use them. Okay, so these are all the benefits that we provide. So now, so industry 4.0. So what is industry 4.0? I just put a collection of kind of big words there, which I think encapsulates a lot of things. You know that industry 4.0 is trying to achieve. So connectivity, sort of IoT, everything has to be connected. Sending data, monitoring what happens in the factory and also acting. Predictability, prediction is everywhere, predicting processes, predicting machine is failing. Personalization, and that's one of the projects that we were trying to work on, but we are working on, which is basically if I want this jacket to be developed with this thing here, here, or this particular item, everything has to change in the factories to actually be very agile and develop personalized products. So this is the change, this is the paradigm shift. Optimization of the whole process, for example, at shop level, manufacturing level, supply chain, so putting all of these elements together and optimize it. And that brings me to the decentralization because now we're going from decentralized to decentralized. I will touch on this later. Adaptability and obviously sustainability in a lot of areas. So this is just a typical slide. So obviously industry 4.0, for those who don't know, 4.0 is the fourth revolution. So that was one, the first one in 1784, which was based on when the steam power came, the water power, the mechanization. You all know that, that was in Great Britain. Then around 80 years later, or wow, 1870, that was industry 2.0, and that was the introduction of electricity into the factories. So everything changed, mass production, and the conveyor belt was created then. Then in 1969, industry 3.0, and that was due to the electronics. The computers, Apple was around there, and automation. And that was the first PLC was created then there, which is PLCs are like the basis of all these factories. And then finally, in 2011, at the Hanover Messe, which is this huge trade fair in Germany, the concept of industry 4.0 was introduced. And that was a German strategy for just improved high-tech in industries. So that was presented there, and it's all about digitization in the whole supply chain and manufacturing, and it relies on the cyber-physical system, so the sensors and the connectivity, the IoT, okay? Obviously autonomous robots, machine learning, and big data. So I like this definition here, industry 4.0. So it says the combination of new technologies and organization of labor. So labor is important here to push this manufacturing into a new realm of optimization. So technologies, labor, and optimization. So it's a trend that focuses on creating smart factories through innovative communication and design between machines and humans. So there is emphasis in optimization, in communication, in labor, and in the interaction of people, the staff that work there with the processes and with the factory. So the human has an important component there, even though everything is smart factories, smart defense manufacturing. These are concepts, you know, intelligent factories, factories of the future. The human or the staff has a particular importance there. And then we are going towards a more competitive, sustainable, and smarter production. That's obvious. And I was mentioning that we are changing the paradigm from a centralized to decentralized smart manufacturing. So now we have this big manufacturing companies that they do it all, but we are going to this model where you develop this piece, you develop this piece, everything is assembled here. You don't rely on that. The supply chain is super well-coordinated and Shadowland Optimize says that you know that this part will arrive here and it will be assembled here. So that actually represents a lot of tricky processes there that have to be addressed. With regards to the technology pillars that will enable this technology, we have obviously autonomous robots everywhere. Simulation, you might have heard about the zero defect prototyping. Now with simulation, you can have like, even design a whole manufacturing process, sorry, or a factory, the robots, you know, the processes, everything simulated, check if it's going to fail, and then just implement it. So zero defect prototyping is kind of an interesting area these days, same as the digital twins, which actually represent digitally what you have in the factory. So you can play with that and you don't risk money or resources. System integration, horizontal integration. So all the process horizontally and vertically integrated. So from the subfloor down to the management, everything has to be integrated to optimize the whole processes. Internet of Things, key for monitoring and enacting, cybersecurity, if we don't put an effort in that, the whole concept will disappear. That's key. Cloud computing, again, for example, what you were saying, edge computing there. So there's a dilemma here, you go fully cloud, you go edge, a little bit of edge, and then cloud. Well, we can debate about that. Additive manufacturing, 3D printing. So good that you have this 3D printing machine and you need this piece and it's broken and you don't have to depend on someone to send it to you. You can have it ready there to try different things. And then augmented reality, the new interfaces, and then obviously big data. And on big data, I'm gonna touch a little bit here, because according to McKinsey, for example, they say that in industry 4.0, the next phase in the digitization of the manufacturing sector is driven by this four disruption. The first one is the interaction with machine, augmented reality systems, like we saw. The second one is improving these transfer instructions to the physical world, so advanced robotics and 3D printing. And the last two are due to the rise in data volumes and then computational power and then the big data analysis. So there's a big emphasis in that that every company should actually embrace. So what is predictive maintenance? Predictive maintenance is a particular area within the whole industry 4.0 and it's actually a very important big area to take care of. Basically, predictive maintenance looks at predicting when any kind of machinery is gonna fail or is gonna have a fault. And obviously the goals is reduced downtime. And if you reduce the downtime, you increase the productivity and obviously you reduce the overall cost. So that's pretty logical, everyone wants that. So up until now, there were different maintenance strategies. Like from centers, you know, they were just from centers a hundred years ago. So they were just using the corrective maintenance and still they use it now. You wait when it breaks, we'll fix it. Obviously they're, this is not optimum. Then the other one is just the planned maintenance. So according to this machine, you know that it's gonna have, it might fail here, it might fail there, it might fail there. So you do this maintenance schedule over the time. But sometimes it happens before, sometimes it doesn't happen. So at the end of the day, everyone is going or everyone should be going towards the predictive maintenance approach. Predicting where it's gonna fail by monitoring and the condition monitoring. The thing is, predictive maintenance is good but the future is proactive maintenance. Which is basically, you don't predict that it's gonna fail. You predict that it's gonna change and it's gonna have an impact and then you adapt so it doesn't happen, okay? So the goal is just that there's no maintenance. Obviously at some point you'd have to replace something. But anyway, so the architecture of a project like in predictive maintenance might have the next following steps. Data collection, aggregation and big data analytics. When I approach these projects, we end up, you know, in the kind of data aggregation and data analysis in Cedar. But it is key to address data collection point. The type of sensors that you use, the sampling rate, how fast you sample because sometimes, you know, people want to sample out every millisecond or every 10 milliseconds and there's no need for that, okay? Then networking, whether it's wireless and that will depend that you will get a lot of data loss. Whether that is a requirement for real-time streaming and real-time acting. That complicates the times, if we have to process the data in real-time. Battery powered. Many people are actually using sensors, battery powered with wireless connectivity and they have to realize that they need a maintenance strategy itself on the battery to replace it and to see when it actually goes down. Security as well. And then on data aggregation, databases, which databases you're using or using the SCADA system or how is the data stored? And the other thing is just impremises versus cloud. Edge computer versus, you know, cloud computing. Different approaches there. And then in baked data analytics, we do a lot of data preparation, a lot of statistics feature engineering, which is one of the main tasks and machine learning. So for the machine learning, this is very basic, obviously, but for predictive maintenance, we use a combination of classification, regression, you can even use reinforcement learning for this, but we basically use classification for label data. So the label will be the errors and then, you know, we learn from this historical data label when there are errors and we try to predict or model the behavior of the machinery, model when it works properly, okay, and then when it doesn't work properly. But then if you want to predict how far in time you have until it breaks, you have to go more with regression and regression kind of machine learning tools, use kind of predict, you know, the time series, okay. Sorry, and finalizes the time series to predict, you know, this time to failure. So finally, as an example, this is one of the projects that we had, many of them, some of them are just a play chain, this one, for example, is in energy. So we are trying to predict when wind turbines fail and this is a crazy area because these wind turbines are actually quite complex models, sort of complex structures and they behave in a non-linear way. And then they are also subject to all of these harsh environments. So taking that and modeling this is kind of tricky. So it is important, you know, to model that right, to actually send the engineers there to have a look and maintain, you know, the system because some of them are offshore, they have to actually prepare a boat, pick the blade, go there, if there is a storm they cannot go, so the sooner they have it, the better for them. And that's what we work on. So at the end of the day, the goal is minimize downtime and avoid the faults in these components. So basically we get the data from the SCADA system, we just model it, you know, analyze every kind of sensors, some of them are not good, some of them are just a combination of them that actually provide good modeling. We do a lot of data exploration and one thing that we do, we rely on the data, we just look at, you know, what the data tells us when we model it, but we also incorporate, you know, the domain expertise. So we talk to the engineers just to discover their hints and we try to incorporate tips on that. And then we do a lot of, you know, advanced monitoring, just trying to model the behavior. So finally, as remarks, part of the future of industry 4.0 depends on this advanced data analytics and the Internet of Things, the industrial Internet of Things. In this area, the predictive and the proactive maintenance is key. Okay. And I put, you know, this image here, which is from a bridge in Honduras. There was a storm and the river changed as a metaphor of what will happen, I think, you know, to companies that don't embrace this technology. Basically, 60% of companies here, they come and they want to invest. They get very happy, they have this data. Then they might say, okay, well, don't know. And then only 5% invest. So I suppose here is just, I hope you know that for some of the companies don't happen, you know, what happened to this river. Sorry, sorry, to the river, to the bridge. So because it seems to me that the, all the technologies are changing so rapidly that there will be companies like this bridge that will serve no purpose in the very near future. Okay. I hope you like it. Thank you very much.