Transcript - Sustainability in Statistics with Domna Ladopoulou
Stephanie Dickinson 0:13
Hi. This is a sustainability interview with Domna Ladopoulou from the Department of statistical science UCL. Questions are by statistical science green champion, Stephanie Dickinson, myself. Hello, Domna, would you like to say something?
Domna Ladopoulou 0:33
Hello, Stephanie, thank you very much for inviting me.
Stephanie Dickinson 0:36
That’s all right. Now this is part of a series of interviews that I’m doing for the Department of statistical science at UCL to try to connect the subject of statistical science more with the discipline of sustainability. And so that involves interviewing people who are doing research in statistics, you have a relationship to sustainability within that research. So I will ask some questions, and Domna will answer them, and so let’s get started with Question one, what is the problem you are trying to solve?
Domna Ladopoulou 1:24
I’m trying to tackle critical challenges on statistical modeling in the wind energy sector. I’m working with Professor Dellaportas, who is a member of the statistical science department at UCL, and my main research aim is to make a meaningful impact on the efficiency reliability of wind energy production. More generally, we could say that we hope in contributing to the global need for sustainable energy resources, especially, we would aim to contribute in some reliability issues in wind farms. During my PhD, I’m trying to develop statistical and machine learning modeling approaches that are appropriate for wind energy, and one of the problems I’m trying to work on is to create a condition monitoring system for wind farms.
Stephanie Dickinson 2:19
Great. Thank you. And how will your research benefit the solving of the problem?
Domna Ladopoulou 2:27
Using data collected in wind farms, which are called SCADA data, and we try to develop new, robust modeling techniques. Currently, we were working on developing a probabilistic condition monitoring system for wind farms, which should discover faults and failures happening in the wind generators early on. And this system must be appropriate for the number of data collected, which are millions, and ideally, it should be able to include an important number of different variables as inputs, so we could benefit from the amount of information available. We hope that if this system is used for conditioning, for condition monitoring of the wind generators, it will help us in the overall sustainability of the wind farms.
Stephanie Dickinson 3:16
Fantastic. That sounds really helpful. Thank you. What area of statistics is your research based in?
Domna Ladopoulou 3:24
I’m working on non parametric probabilistic methods such as Gaussian processes, and I’m also using some other machine learning models, such as probabilistic neural networks, that they offer good predictive capabilities and an advantage of over Aussie and processes in terms of computational cost needed to train them. The advantages of both methods are that they offer flexibility and adaptability in modeling complex relationship within the data. This is something very useful for us working in the wind energy sector, as the wind energy systems have a dynamic nature. And what I mean by that is that the variability of the wind speed creates the dynamic nature of the wind power output.
Stephanie Dickinson 4:14
Could you explain your methodology so that someone without a statistics background might understand what are the bones of it and how does it work?
Domna Ladopoulou 4:26
As I mentioned earlier, the scope is by using we farm data to spot problems such as failures in wind farms early. So what we really do is that we use our models, which are appropriately designed for wind farms to predict the wind power output, and then we compare the expected power which is the result of our model, with the observed power output. Using this information, we can build a condition monitoring system that can draw conclusions on whether the wind. Operators operate in a healthy way. So in a way, we are developing new techniques to improve wind energy, reliability and sustainability.
Stephanie Dickinson 5:08
Great. I can see why that would be very important. And yeah, I think improvements in efficiency and reliability is very welcome. I think if more people who are interested in sustainability understood about methodology and applied it to their own actions, they have the potential to become more effective. What do you think methodologies do for sustainability?
Domna Ladopoulou 5:36
I think that understanding what a specific methodology could potentially offer in solving a particular problem can empower people to act more effectively. So nowadays, people can make more informed decision in comparison to the past, due to the power of data analysis offered in our everyday lives. And by informed decision making, I mean that we can make choices based on a deep understanding of all the relevant information given for a specific problem, while at the same time we know the potential consequences of our action. And of course, we can evaluate the risk of each choice, which is a very important benefit when we speak in terms of science. So overall, I think that making informed decisions can give us the privilege of implementing impactful strategies for sustainability practices.
Stephanie Dickinson 6:30
That sounds very positive. Do you think it would be possible to roll out your methodology on a mass scale? What would that look like?
Domna Ladopoulou 6:42
I think it is feasible with the right infrastructure and collaboration. I mean that from a text, a technical perspective, it would involve implementing a standardized data collection process across wind farm. So the development of an automated anomaly detection system could be also feasible. Another component of probably higher importance this is more difficult to achieve, is that it might be necessary to educate, or maybe even convince the relevant stakeholders and policy makers about the benefits of such methods in wind energy production and sustainability, and I think this is the most essential part if we aim for widespread adoption.
Stephanie Dickinson 7:27
Are you collaborating with other departments across UCL or in different universities? Are there any external partners you are working with in industry?
Domna Ladopoulou 7:40
Yes, so my research is multidisciplinary, since it requires some knowledge of engineering and it also requires a statistical background. So to fill in my engineering gap, I have some collaborations with industry partners in the fields of wind energy. This helped me to stay informed about real world challenges, which I aim to solve with my research. But as for now, we don’t have any academic collaborations, but we are more than open to explore further collaboration, both in academia and industry.
Stephanie Dickinson 8:15
Great. What can you bring to the table with a potential collaboration with other academics?
Domna Ladopoulou 8:22
I think we could bring a blend of statistical expertise and some domain knowledge in the wind energy sector, and we could work on open research questions in the field of wind energy. And probably we could work also in other energy related fields that require machine learning, model suite suitable for almost the same type of complexities as these observed in wind power systems.
Stephanie Dickinson 8:52
Right. Are there any particular fields of academia that you think would benefit from a collaboration with your research field?
Domna Ladopoulou 9:02
Yes, I think so. Collaboration with scientists working on renewable energy, engineering, environment and total scientists and wind energy policy and could potentially bring some benefits in the field, in the case of engineering, in engineering and environmental science, I think using the appropriate models, we could increase renewable renewable energy efficiency, and we could get a better understanding of the environmental impacts, which I find quite important. And what I mean is that if we have a proper assessment study on the economic on them, ecological footprint of wind farms could optimize their sustainability. And from a policy perspective, I think we could make informed policy decision making for wind energy production issues which are quite important.
Stephanie Dickinson 10:00
Fantastic. When you collaborate with others, particularly thinking of your current industry collaborations, what is your main function in these projects? And how does your research change things?
Domna Ladopoulou 10:14
The projects in which we work with the industry now is to improve the efficiency and reliability of wind energy production by developing an anomaly detection system. So as I mentioned earlier, the main scope of this system is to identify anomalies in the operation of we generators. And with the system we built, we detected operate operational anomalies about a day before the existing conditioning monitoring system that is installed now in the wind farms, and this result could change something. One of them is the number of is a number and the cost of the interventions we that must be done in a wind generator to return back to a healthy operation. So if we manage to detect earlier potential at fault, then we help in the overall reliability and sustainability of the farm by reducing a lot the main maintenance cost of the wind farms, which is a very significant amount spent yearly for wind farms.
Stephanie Dickinson 11:18
Really helpful. Yeah, where in the world will this particular piece of research be of the most use?
Domna Ladopoulou 11:26
This I think this work will be most beneficial in regions with a high reliance on wind power and, of course, a significant presence of wind farms, also regions with a growing or established wind energy industry might find these research useful, and I’m referring to these regions, since they offer a higher concentration of wind farms, and as a result, they face such challenges in the optimization of energy production, the research could be also viable in markets where the development of wind farms is of growing interest, just because they will face this kind of issues on a later stage.
Stephanie Dickinson 12:09
If you could think back to your formal education in statistics, what topics covered in the standard university curriculum had the most influence on your current research and your interactions with industry?
Domna Ladopoulou 12:25
I studied in the Athens University of Economics and Business in the Department of Statistics for my bachelor degree, and I find it very hard to pick a specific topic covered, but the modules related to data analysis and statistics for environment had some influence in my work, but I believe that I learned whatever I learned in this department, apart from helping me start the PhD still helps me today, since it provided me the foundation I need to grasp new concept in my new concepts in my field, but since you asked me this question, it’s a great opportunity to thank my professors in Athens for what they taught me and in general, for what they offer to the students of this department, despite facing challenges with limited resources now concerning My interactions with industry. I started working on wind energy related topics during my master dissertation, where it was my first interaction with industry experts in the field, and they introduced me to the wind energy related problems. Along with my supervisor, they offered me some data, which I use for my thesis, and that was the starting point of my work in wind energy.
Stephanie Dickinson 13:44
Finally, the final question, let’s save the world. If you could pick anywhere or anything to apply methodology to that would have the biggest impact? What would that be? And what would the methodology change there?
Domna Ladopoulou 14:02
So that’s a great and very hard answer to answer question, but to keep the conversation in the environment, which is important, I think you would, I would try to work on the most important to me problems in the world, which is climate change. And this is an incredibly hard and complex problem to solve, but we could use AI in helping climate change, and there are currently a lot of brilliant scientists working on embedding AI to help climate change. I think applying statistical and machine learning models in such contexts would increase our awareness and understanding about climate change, which I think is an important component of tackling with this issue, or at least a good starting point. At the same time, AI can help with predicting natural disasters. They can offer plenty of other applications. For example, I watched on the news a few days ago about a system that can dive deeper than a human. And using AI, rather than normal GPS systems, to navigate itself, it can capture very high quality video and track the corals and the sea life, which is now due to the climate change, moving to a deeper level of the sea in order to survive from the increased temperature of the sea. So again, as I mentioned earlier in our discussion, and important feature of statistics is that it helps us to make informed decision making, and I believe it is a necessary step to help us act towards a more sustainable future.
Stephanie Dickinson 15:34
Great. I’m really pleased with that final answer to that question. I very much agree with you on on those points, and I think that you’re doing some really good work, and it sounds like there is a lot of hope for statistics to help solve these issues that that we’re facing right now in the entire world with climate change. So thank you very much for putting all this work into it. As a professional working in the field of statistics on sustainability issues, I think a lot of people would be very appreciative of that work. I think, yeah, there’s a lot of opportunities for people to work with statistics in the future to help solve these issues. Hopefully we can all use our minds and get ourselves out of this, this problem. But yeah, thank you. Thank you very much. Domna for this brilliant interview.
Domna Ladopoulou 16:40
I’m very glad you liked our work.
Stephanie Dickinson 16:43
Well, if, if anyone wants to contact you about about your research, how can they contact you?
Domna Ladopoulou 16:50
So I have a web page, the UCL profile web page under my name, which is Domna Ladopoulou. I also they can contact me my email, which is available, and I also have a LinkedIn profile, so I’m happy to discuss further.
Stephanie Dickinson 17:08
Great right. Thank you very much, and hope you have a lovely rest of the day.
Domna Ladopoulou 17:13
Thank you very much.
Unknown Speaker 17:17
UCL minds brings together the knowledge, insights and ideas of our community through a wide range of events and activities that are open to everyone you.