What is the relationship between a salamander and a blackout?

Vladimiro Miranda applies computational intelligence to power systems; José Carlos Príncipe studies how the human brain processes signals. The first is a professor at INESC TEC, in Porto, Portugal, as a second is a professor at the University of Florida, in the US.

Vladimiro Miranda combined the Biomedical Engineering models developed by José Carlos Príncipe with wind forecasting and the results paid off. Yesterday he received an excellence award for renewable energies in the US from the Institute of Electrical and Electronics Engineers (IEEE), the largest professional association of engineers in the world. José Carlos Príncipe was the first Portuguese scientist to receive an award from the IEEE, in 2007 and 2011, for his contributions to Biomedical Engineering.

The PÚBLICO met with these scientists in Porto and our meeting has shown that salamanders, blackouts, shellfish that lose their nervous system, and wind ramps –abrupt shifts in wind speed that cost millions to the power industry – are all connected.

An emblematic project behind this relationship, which led to this IEEE award, is the Argus, a wind forecasting model developed for the US Department of Energy according to the specific needs of the country, the second largest wind power producer in the world. Despite the gigantic wind parks, the country suffers from wind ramps and lacks support from the hydro power sector.

According to Vladimiro Miranda, combining Biomedicine and wind led to the “first industrial application of the machine-brain interface”.

What kind of research do you conduct [which have led to international recognitions]?

JOSÉ CARLOS PRÍNCIPE — As species evolved, all living organisms have developed mechanisms to extract information from signals. Man has developed sight (our most refined sense), hearing, smell and taste. I see living organisms as systems that must live in a non-stationary world where they must survive. That is where I take analogies to develop signal processing algorithms. The Mathematical principles are often restricted. I’m not saying that this is the incorrect way of seeing things, but it must be complemented with new paradigms. I was trained in signal processing, where I apply mathematical foundations to engineering and system problems. But during my PhD I became familiarised with biological processing problems, namely brain signals, and so I have an insight on engineering and biological science. I use biological paradigms and I try to find mathematical formulations for those problems.

Are signals inspired by the human body?

J.C.P. — Not the human body, but living organisms. Take the salamander for example: the central part of the human paleo brain [the oldest in terms of evolution] is [basically] the brain of a salamander; we still have the limbic system of a salamander.

Does this have to do with reusing signals?

J.C.P. — Not reusing, but processing and extracting information from signals. The information collected from a signal using a sensor is found in its time structure. It is the sequence, the variation that contains the information. The same thing happens in a wind sensor: it is the structure of a signal that varies throughout time. On an abstract level, the brain and wind signals are [just] signals. It does not matter if those signals come from the wind, brain, or astronomy, the problem is the same: extracting information from the signal, from its time structure.

VLADIMIRO MIRANDA — Information can do magic.

Is the memory of cells in living being important here?

J.C.P. — The cell itself has no memory; in the brain the cells work together and the result is interpreted by the rest of the brain as a memory of what happened.

V.M. — When we see a face, a special area of our brain is activated: from this example we can extract a parallelism with power systems. José Carlos Príncipe’s team is trying to discover how to interpret the signals which are triggered every time we recognise a face in a photo and to understand how capturing these signals is important. How do we know that the brain is indeed seeing a face? The electroencephalograph shows a number of signals but they are all scrambled. The information must be hidden in the signals because I know I’m seeing a face. It’s like having cake batter with eggs, flour, all mixed, but the flour is still there and that’s what’s fascinating with this work.

That is the chaos of information.

J.C.P. — Yes, because all the information is in there and we lack a periscope. V.M. — The mathematical instruments are our periscope; they’re the lens that allows us to look at information from a certain angle and see what’s there in the middle of the mess.

Professor José Carlos Príncipe, which international institutions do you collaborate with?

J.C.P. — I have a very fruitful cooperation with Chile because of the Alma [the largest radio telescope on Earth], where they are using my algorithms to process radio astronomy signals. Within two or three years, a lot of people will be using the models we are developing because they have proved to be more efficient than the traditional models, but they have to be tested in different [scientific] areas. I have also been working with Singapore in the nanotechnology area; with Italy in signal processing in ECG and EEG; in China, essentially because of my students, I have been working in signal processing projects connected to telecommunications and video. Wind forecasting [the goal of the Argus] is another application that helps us understand the models.

And how did Vladimiro Miranda borrow from the work by José Carlos Príncipe?

V.M. — The first thing was understanding the importance of the information in the signal, instead of using the automatic approach of the signal as a mathematical variable. I was fascinated to understand another way of handling the contents of the signal. I have worked a lot on electricity and gas consumption forecasting. Well, in forecasting we work with signals, we measure the temperature of the atmosphere, of the wind...

Ad how can you forecast consumption?

V.M. — By creating a model that is similar to that of reality and reducing the errors produced by the model.

Can your research be applied to other areas?

J.C.P. — I work on other problems, such as vision, image and video processing. I am interested in [biological] signals which are not created by man, an intelligent being with its own mathematical principles that creates systems based on those principles. The signals created by the living systems have a way of changing how that signal is generated. There it is necessary to find mathematical techniques that are less restrictive.

Is that why you brought up the salamander?

J.C.P. — I studied the evolution of species to learn about the evolution of the nervous system. Deep down, the nervous system is associated with the moving organisms’ need for their own model of the world, and that’s very interesting. There are shellfish, such as the ascidians, which move in the beginning of their life and have a nervous system. In the last stages of their lives, they will fix on a rock and their nervous system is assimilated, it disappears.

V.M. — ...Because they don’t need to move in that environment, they don’t need to recognise signals.

J.C.P. — This is a simplistic way of seeing the development of the nervous system, but it is also related to the need for movement and understanding a non-stationary environment. When people move from one place to the other, they need to adapt to their new reality. I try to understand how organisms process nervous systems in order to survive. Then I try to apply that, for instance, in image and video processing.

How do you go from that to energy?

V.M. — Information avalanches are more and more frequent: blackouts are examples of that. In some cases, that avalanche was found at a control centre that operators froze because alarm signals were coming from all over the place and they didn’t know which information was important or not. They weren’t capable of processing all that information at the same time, they didn’t make any decision or they made the wrong decision and the system collapsed. In my area of expertise, this is an important problem.

Does that provide more information on the wind ramps?

V.M. — Because of the planetary importance of energy, some of these events lead cause impacts worth millions or billions of euros. The costs of operating errors are astronomical. In wind power systems, a lot of money is spent on measures, often excessive, to prevent harmful consequences brought about by sudden increases or reductions of wind speed. If we are capable of forecasting wind more accurately and with a high degree of probability, we can more effectively take advantage of the energy available. We won’t need as much equipment or to operate expensive equipment that will not be needed because there will be no emergency.

But REN’s [national grid operator] forecasting error is low.

V.M. — The accurateness of the forecast has to do with scale. It is a lot easier for a small forecasting error to occur in an aggregated territory — Portugal or the Iberian Peninsula, for instance — than in a wind park. I haven’t worked on REN’s problem, which is a national problem. I’m more focused on a smaller scale to obtain a more correct image of wind parks, of their participation in energy markets and the effects of the parks locally.

What are those effects?

V.M. — If there is plenty of wind in one place and not a lot of it in another, power traffic in the grid will increase and that may cause problems because the grid’s capacity to transport power is limited and the distribution of the power plants is asymmetrical. Typically, Portugal has hydro power plants in the north and thermal power plants in the south, and their reaction times are quite different.

Does the wind forecasting model you developed for the US provide a lower error?

V.M. — All the experiments conducted provided better results than traditional techniques. For the US Midwest we reduced the error by 20%. That difference is quite significant and worth a lot of money. The next step, inevitable for operators such as REN and the Red Eléctrica de España [Spain], is using models which are smaller than aggregated forecasting. The Argus, the platform we developed for the Argonne National Laboratory [of the Department of Energy], allows American wind power forecasting companies to build commercial systems from that and allows universities to build new concepts.

Text by Lurdes Ferreira | Photo by Nuno Alexandre Mendes