 Hello, everyone. So I'm Ricardo Cimón Carbajo, as Edward has mentioned. And I'm the head of innovation and development here in Cedar. So today I'm going to talk about this strange word, Windows 3 4.0, which I suppose you know what it is. It's just the industry 4.0 revolution in the energy section, sector, OK? So yeah, I will just mention what is changing there, particularly with the use of data science. So before starting, I will just talk about this Cedar, OK? Edward has mentioned we are the National Technology Center in Applied, Data Analytics and Artificial Intelligence. Applied is very important here, because we have been created to actually merge research into industry. So everything we do is products for industry, for the benefit of industry, to actually tackle the challenges of industry. And as Edward was mentioning, we have been recently selected as the Digital Innovation Hub in AI in Ireland. Digital Innovation Hubs are these kind of concepts in Europe now, which are centers which are created specifically for this, to help industry, OK? So we are key in helping all of this industry in Ireland to embrace artificial intelligence, making sense of artificial intelligence for them. So for that, we use a lot of collaboration instruments. So we have a bespoke service, for example. Many companies come to us with a specific project, maybe basic data analytics to very complex reinforcement learning kind of cutting-edge technology. Then also we leverage national funding from EI, IDA, SEI, so we help you get the money there. And also, as Edward was mentioning, we participate a lot in Horizon 2020 projects. So we bring our member companies into the European consortiums there, OK? So we increase their visibility in Europe. So we work in a lot of areas, but particularly in energy and in wind energy, we are quite strong. We, for example, in DIM, working in energy, we have data science architects, senior data scientists, data scientists, and also Marie Curie, a career-feed postdoctoral fellows. This is a fantastic team. They are very, very handsome, very pretty, but also they are very good at what they do, OK? They are crazy good data scientists, but also not just data science. Now they have learned how wind energy works and they are at the intersections. They are very well-equipped to actually tackle the challenging problems that we encounter in this area. So now I'm going to talk about this Windows 3.4.0, which is like a play between wind and industry, as you might have seen. So obviously the industry 4.0 revolution has some principles that are applied as well to the wind energy. These principles, for example, are connectivity. So the Ford revolution is all about this IIoT, Internet of the Things. So now in wind turbines, they are actually installing a lot of devices there, not just typical ones in the Scada, more sensors that we are actually using in our big data applications. Predictability industry 4.0, in particular, about wind, everything is about predicting. Predictive energy, predicting faults, predicting everything. And also optimization. Companies come for us to optimize these wind turbines. Then also scalability, a lot of adaptability, particularly in wind turbines. For example, in the area of industry 4.0 in manufacturing, we work a lot. You have to, there's different situations that we can not anticipate. Obviously we train our models for this. But in here, in wind turbines, it's even worse because they are subject to these harsh environments. So our models have to be well trained. And obviously, sustainability, the green kind of tag everywhere. So to put you into context, in terms of energy in Ireland, and this is important, the European Commission has established some sort of targets that we have to aim, the 2020 targets. The problem with the 2020 targets is just, the day said that 16% of our energy has to come from renewable energy in Ireland. At the moment, well, that's from 2017, we are at 10.6%. I'm not sure if we're gonna get there. But anyway, we are on the right way. And what we get is just 90% of our renewable energy, as you see, comes from wind, solid biomass, and liquid biofuels. And if you look at the other graph here, the majority, 52% comes from wind. So wind is actually the key renewable technology in which we have to invest on, particularly in Ireland, and in many areas in the world. Solar, as you say in Ireland, as you see there, it's not very good. But anyway, so because of this, we are putting a lot of emphasis and because the wind sector has had a lot of work, for example, control systems or any other predictive systems using the traditional kind of physics-based approach. Now, it's not that they are stuck, but they can complement the machine learning approaches. And by actually extracting it, it's not just with the data, that's where we can go farther. Okay, so it's key to invest in this. And because of that, we have different areas that we work on. These are all real projects where we actually do cutting-edge research, but to develop products that are actually productionized in companies, okay? So the first area, and I will mention more about that later, is wind turbine predictive maintenance, basically predicting faults, okay? Predicting when these machines are gonna fail. The other one is wind turbine farm forecasting, energy forecasting. So now there's a huge interest in knowing how much energy this wind turbine or this wind farm is gonna produce. And then another area in which we have worked is wind turbine power output optimization. These companies which come and say, like, we are the owners of these wind turbines, I want to see if we can get more energy out of this. So can you optimize the performance? So we have developed recommender systems to actually, for example, say, if you tilt the nacelle like that or the blade like that with this wind speed and this wind direction, you can get a better performance. Obviously, we are competing with Siemens and all of these companies, so it's tricky enough that they come to us. Okay, so I'm gonna mention about, you know, Project One. So Project One, in the area of predictive maintenance, as I was saying, this project is called Wind AI Project and it's funded by Enterprise Salon, so the Marie Curie program, okay? You can apply now because there is a new, a new kind of like program coming for the Marie Curie for postdoctoral research fellows. So Kristen is working there, he's over there. And this is a very challenging program because obviously we have to model this complex wind turbine system. And there's a lot of parameters there and we have to actually predict, you know, when they are gonna fail. And not just predict when they are gonna fail one hour ahead. Usually it's just more than 24 hours. For example, when you send your engineers in an offshore environment, offshore wind turbine, they have to not rent the boat, but you prepare the boat, see if the blades are available, see if they have all the parts. Also, if there's a storm coming, so if you actually predict, you know, well ahead in time, you can account, you know, you can reduce the downtime of the turbine and you can even prevent other problems there. So in there, we have a lot of a scatter data from different turbines, real scatter data from this company. And then we have a variety of parameters with the speech, you know, but also the temperature, for example, of the gearbox, the pressure, the oil temperature as well. So there's a lot of things, lots of parameters. And then the challenge here, explore all of these parameters and generate new features. But it's not as easy as it looks because sometimes we have to understand these problems. So I have to understand these parameters. So we just talk a lot, you know, with wind energy engineers to actually learn the domain. So it's not just, you know, throw it there and see what happens. Sometimes, you know, it's way more difficult than that. And then obviously, as you say, you know, we use classification and regression problems. Classification, try to model the system, for example, the wind turbine working normally and then predict, you know, when it actually is in a prefall stage. And regression, for example, is just trying to predict how much time it's remaining until default, okay? So we just jump between, you know, these kind of modeling techniques. I'm not gonna go deeper because I don't have time there. The other area is wind turbine power forecast. I don't know how many of you are aware of the integrated single electricity market, but it's very important, okay? It's at European level, in particular in Ireland, the ISEM has been created, dictated by Europe because what happens there is just, we have to just solve this energy problem in Europe. So we have to just combat, you know, the imbalance between production and consumption, okay? Generation and demand. So in there, what happens is just a regulator, Edgrid here imposed this program because of Europe, as I said, and they said now, every kind of wind turbine generator has to forecast the generated power for the next 36 hours ahead. So basically, if they don't do it right, do you say you're gonna generate X amount of megawatts per hour and you don't get it right for these 36 hours, you will get penalties there. So now all of these wind turbine owners come to us and say, can you predict it, you know, for real? Can you very, very accurate there because money is at the stake here. So then what we do is just, we just create this project with SSC electricity, SSC is our main partner, David Norona is the lead there, I don't know, he's over there, and then he was funded by the SEI Sustainable Energy Authority of Ireland. And the aim, as I say, is just predict accurately, you know, the wind turbine forecast methods, which is then used as well for trading. They actually used to say, I'm gonna buy this or I'm gonna sell this energy in the trading market, okay? Of energy. So obviously this is very tricky, we have to predict the available active power generated by these wind turbines, and we have, as I said before, a bunch of parameters, but particularly what we use is wind speed, wind direction, and temperature and so on. And one of the things, which is key is obviously the weather. So you can predict, you can model this wind turbine here perfectly, but if your forecast of the weather is wrong, you cannot do anything, everything will be wrong. So what we're doing as well is just trying to minimize the error, coming from, you know, different forecasting models from Mediterranean, Norway and things like that, which is the challenging part, okay? So basically, I just saw you there, we are actually modeling, if you understand about this, the power curve. So it's basically the active power generated the different kind of wind speeds there. It's usually a power curve like that, the theoretical one, but obviously when you have real data from real wind turbines, you encounter surprises, loads of surprises. So we have to remove noise and layers, deal with things like that, and it becomes challenging enough. And for the modeling, when we tackle this problem, in terms of machine learning and modeling, we use for example, a combination of many things, forecasting data from the meteorological systems, also the data, meteorological data from the wind turbines, but also, you know, these parameters from the wind turbines at turbine level or even wind farm level, okay? So a combination of all of that, variations, applying signal processing in all of these parameters, changing frequencies, changing lags, cleaning, all of that. And then doing aggregations as well, and applying a battery of machine learning algorithms which range from neural networks to trees, decision tree, random forest, or even ensembles of all of these, okay? So it's like a kind of like research hyperparameter tuning problem. So finally, I brought my bridge here because I love it, but the message here is just industry 4.0, and particularly everything that needs to embrace AI. Not because I say so, it's because it has to be, you know, the technology is changing rapidly, and there are certain things that you cannot do without machine learning, okay? Because there is a limitation, for example, in control theory, the algorithms that are limited and come up certain functions. Now we have to use, for example, reinforcement learning, okay? So we have to embrace AI, we have to embrace data analytics, particularly we are actually tackling hybrid modeling now, physics models, control theory, with machine learning, putting all together, creating these digital twins. That's what is working now, okay? So the message here is just to embrace the technology, only 60% of the companies are exploring to invest, but only 5% are actually investing, and see that it's here just to help them with this. So it doesn't happen what it happens to this bridge. So I see, you know, the river, as the technology is changing, and then maybe companies are developing these great products which will have no purpose in the future, like this bridge, okay? So thank you very much, and if you have any questions, later on. Thank you.