LØRN Case #C0045
Digitale tvillinger
I denne episoden av #LØRN snakker Silvija med VP Applications, Analytics and Visualization i Kongsberg Digital, Michael Link, om hvilke aktuelle problemer man løser med Big Data. Kongsberg Digital leverer tjenester innenfor energi og maritim industri, med Kognifai som plattform for å samle inn og behandle data. Michael forteller om Kongsberg sitt arbeid med digitale tvillinger. I denne sammenhengen er begrepet digitale tvillinger ensbetydende med syntetiske modeller av produksjonsplattformer for olje og gass. Videre i podcasten diskuterer Michael og Silvija potensialet til digitale tvillinger, og hvordan de kan brukes sammen med Big Data, til å lage realistiske simulasjoner som kan gi informasjon på områder man ønsker få innsikt om.

Michael Link

Forsker

Kongsberg

"Det som skaper mest verdi er når vi klarer å analysere dataene og lage gode verktøy for brukerne. Her kombinerer vi mange analytiske metoder hentet fra statistikk, kyberteknikk, machine learning og deep learning."

Varighet: 19 min

LYTTE

Ta quiz og få læringsbevis

0.00

Du må være medlem for å ta quiz

Ferdig med quiz?

Besvar refleksjonsoppgave

Hva gjør dere på jobben?

Kongsberg Digital leverer digitale tjenester innenfor energi og maritim industri. Kognifai er vår plattform for å samle inn og behandle data fra tusenvis av skip, oljerigger og vindturbiner som har utstyr fra Kongsberg om bord.

Hva er greia med big data?

Det hele startet med internett og brukerdata. Men ettersom andre industrier satser på digitalisering, blir det generert enorme mengder data fra brukere og sensorer hver dag. Big data dekker mye – både teknologier for å transportere, behandle og lagre dataene, samt verktøy for å forstå og ta i bruk all informasjon som finnes i disse dataene.

Hvorfor er det spennende?

Det som er mest spennende er alle mulighetene som ligger i dataene. Vi kan kombinere data fra vindturbiner, avansert analyse, ekspertise fra industrien og programvare. Resultater blir nye tjenester som resulterer i mer effektiv drift av systemer som produserer ren, fornybar energi.

Hvorfor er det skummelt?

Når vi snakker om personlig data, så kan dette gi organisasjoner og stater stor makt. Da håper jeg at dataene brukes til gode formål som bedre helsetjenester og ikke misbrukes til å påvirke eller diskriminere.

Ditt beste eksempel på big data?

Jeg var initiativtaker til et prosjekt hos Opera for noen år tilbake. Våre brukere leste cirka 10 milliarder nettsider hver dag, som serverparken komprimerte. Ved å analysere disse dataene kunne vi forstå nyhetsbildet globalt og i forskjellige land, og benytte dette til å anbefale innhold og alternative kilder.

Dine andre favoritteksempler på big data, internasjonalt og nasjonalt?

Hos Kongsberg jobber vi nå med digitale tvillinger. Dette er syntetiske modeller av produksjonsplattformer for olje og gass. Det er mye data tilgjengelig fra slike plattformer som sier noe om driftstilstanden.

Hvordan funker det egentlig?

Våre produkter innenfor fornybar energi, som heter EmPower, samler sanntidsdata fra vindturbiner. Disse dataene prosesseres for å forutse feil i produksjonen og for å optimalisere driften.

Hva gjør vi unikt godt i Norge av dette?

Teknologien har vært tilgjengelig i noen år, så det gjelder å ta det i bruk. Mitt inntrykk er at Norge er fremoverlent når det gjelder digitalisering av tjenester. La oss bruke disse dataene til å skape nye produkter og løsninger som bidrar til eget samfunn og som kanskje kan eksporteres.

Et kort big data-sitat?

Big data – much more than just technology.

Hva gjør dere på jobben?

Kongsberg Digital leverer digitale tjenester innenfor energi og maritim industri. Kognifai er vår plattform for å samle inn og behandle data fra tusenvis av skip, oljerigger og vindturbiner som har utstyr fra Kongsberg om bord.

Hva er greia med big data?

Det hele startet med internett og brukerdata. Men ettersom andre industrier satser på digitalisering, blir det generert enorme mengder data fra brukere og sensorer hver dag. Big data dekker mye – både teknologier for å transportere, behandle og lagre dataene, samt verktøy for å forstå og ta i bruk all informasjon som finnes i disse dataene.

Hvorfor er det spennende?

Det som er mest spennende er alle mulighetene som ligger i dataene. Vi kan kombinere data fra vindturbiner, avansert analyse, ekspertise fra industrien og programvare. Resultater blir nye tjenester som resulterer i mer effektiv drift av systemer som produserer ren, fornybar energi.

Hvorfor er det skummelt?

Når vi snakker om personlig data, så kan dette gi organisasjoner og stater stor makt. Da håper jeg at dataene brukes til gode formål som bedre helsetjenester og ikke misbrukes til å påvirke eller diskriminere.

Ditt beste eksempel på big data?

Jeg var initiativtaker til et prosjekt hos Opera for noen år tilbake. Våre brukere leste cirka 10 milliarder nettsider hver dag, som serverparken komprimerte. Ved å analysere disse dataene kunne vi forstå nyhetsbildet globalt og i forskjellige land, og benytte dette til å anbefale innhold og alternative kilder.

Dine andre favoritteksempler på big data, internasjonalt og nasjonalt?

Hos Kongsberg jobber vi nå med digitale tvillinger. Dette er syntetiske modeller av produksjonsplattformer for olje og gass. Det er mye data tilgjengelig fra slike plattformer som sier noe om driftstilstanden.

Hvordan funker det egentlig?

Våre produkter innenfor fornybar energi, som heter EmPower, samler sanntidsdata fra vindturbiner. Disse dataene prosesseres for å forutse feil i produksjonen og for å optimalisere driften.

Hva gjør vi unikt godt i Norge av dette?

Teknologien har vært tilgjengelig i noen år, så det gjelder å ta det i bruk. Mitt inntrykk er at Norge er fremoverlent når det gjelder digitalisering av tjenester. La oss bruke disse dataene til å skape nye produkter og løsninger som bidrar til eget samfunn og som kanskje kan eksporteres.

Et kort big data-sitat?

Big data – much more than just technology.

Vis mer
Tema: AI- og datadrevne plattformer
Organisasjon: Kongsberg
Perspektiv: Forskning
Dato: 181012
Sted: OSLO
Vert: Silvija Seres

Dette er hva du vil lære:


Big data Digitale tvillinger Eksempler på anvendelse av big data

Mer læring:

Nvidia sin siste GTC-konferanse

Del denne Casen

Din neste LØRNing

Din neste LØRNing

Din neste LØRNing

Dette er LØRN Cases

En LØRN CASE er en kort og praktisk, lett og morsom, innovasjonshistorie. Den er fortalt på 30 minutter, er samtalebasert, og virker like bra som podkast, video eller tekst. Lytt og lær der det passer deg best! Vi dekker 15 tematiske områder om teknologi, innovasjon og ledelse, og 10 perspektiver som gründer, forsker etc. På denne siden kan du lytte, se eller lese gratis, men vi anbefaler deg å registrere deg, slik at vi kan lage personaliserte læringsstier for nettopp deg. 

Vi vil gjerne hjelpe deg komme i gang og fortsette å drive med livslang læring.

En LØRN CASE er en kort og praktisk, lett og morsom, innovasjonshistorie. Den er fortalt på 30 minutter, er samtalebasert, og virker like bra som podkast, video eller tekst. Lytt og lær der det passer deg best! Vi dekker 15 tematiske områder om teknologi, innovasjon og ledelse, og 10 perspektiver som gründer, forsker etc. På denne siden kan du lytte, se eller lese gratis, men vi anbefaler deg å registrere deg, slik at vi kan lage personaliserte læringsstier for nettopp deg. Vi vil gjerne hjelpe deg komme i gang og fortsette å drive med livslang læring.

Vis

Flere caser i samme tema

More Cases in the same topic

#C0044
AI- og datadrevne plattformer

Heidi Dahl

Forsker

Sintef

#C0043
AI- og datadrevne plattformer

Sverre Kjenne

Leder

BaneNor

#C0042
AI- og datadrevne plattformer

Geir Engdahl

CTO

Cognite

Utskrift av samtalen: Digitale tvillinger

Silvija Seres: Hey and welcome to today’s version of podcast edition of Lørn.tech. We are trying to address different kind of technology in a very popular way for people who want to join the future. Because we believe you need to know some tech in order to have future literacy. Today’s topic is big data. I am Silvija Seres and my guest is Michael Link who is the VP of application analytics and visualization at Kongsberg Digital. I lost my breath.

 

Michael Link: It's a long title.

 

Silvija: Hi Michael. I've been a fan since you were in opera software. So, we are going to talk about that. But Can you please tell us what you do?

 

Michael: I joined contracted to 18 months ago. I've been building up team which works with data 

AI machine learning and analytics and the data from the Maritime industry.

 

Silvija: Say something about Kongsberg as well.

 

Michael: Kongsberg is a large Norwegian high-tech industrial company. With a large presence in the Maritime energy space. There is another part of Kongsberg which is a large defense business which I don't work.

 

Silvija: So Maritime, energy and industrial processing. But where is the digital?

 

Michael: So I work in a division of digital where we produce the software services that use all the data generate out of about all the Kongsberg systems on the different source of industrial assets, such as ships or oil platform. 

 

Silvija: So you gather data from ships and energy system and put it on one data platform?  

 

Michael: We use the system Critical Kognifai which transmits the data in securely to our system, where we can process and store the data to make applications.

 

Silvija: So a lot of data, that is big data? 

 

Michael: Mostly big data. A ship or an oil platform could contain thousands and thousands of sensors which send data every second. 

 

Silvija: Can I ask a stupid question, where are the sensors?

 

Michael: They are physically on board on these industrial assets. They can be on land or sea.

 

 Silvija: What do you measure? 

 

Michael: Eventually depends on what it is being used for. The typical things to measure on the last ship, would be the engine or the use of fuel, temperature and pressure. But you do have other kinds of sensors which are you used to map the surface under the sea. 

 

Silvija: Have you been involved in this autonomous ship project?

 

Michael: Yes. I do work on that. 

 

Silvija: So that needs to measure the bottom of the sea and the coastline, and laser sort of stuff. How does that work?

 

Michael: So that's a very special kind of big data project. Where you gather data from radars like maps, charts, cameras. It's a large amount of data which are been continuously gathered. 

 

Silvija: ..And you need to speak images, sound, weather data etc.? 

 

Michael: You need to go to process all these kinds of data. Which is something that machine learning and deep learning particularly have unlocked for us. But there are also other single processing methods to work with this data. And all of this has to be combined first to build a situational awareness. A “wessel” in this case understands what the surroundings are. And then you can begin building intelligence about how to navigate in that environment. 

 

Silvija: So basically you are making industrial assets smart through their gathering of data, analysis of data, cross connection of data. Is that the sort of high-level idea?

 

Michael: The goal is to make things smarter, to make them more automated. So that they can do more things themselves or to make them more officiant and safer.

 

Silvija: Can you can you talk about how the prosess flow is? 

 

Michael: I can give you one example we have from the wind industry. Where we had a software called “impower”. At the windfarm there is equipment.

 

Silvija: So a windfarm is a set of winter bines?

 

Michael: Yes, and maybe a transformer substation, and some electrical equipment. That we have a piece of equipment close to there, Kognifi Edge. That will transmit all of the relevant data, to our software which runs in the cloud in this case. So that process is just a digital process of plumming. To move the data from the windfarm to a place we could work with it.

 

Silvija: And you see how much energy is being produced and where there are wrong kinds of vibrations? 

 

Michael: So the goal is to both have an overview of the current production, to predict the future production. To optimize how you produce, depending on the demand for electricity in the grid. And a topic which is very interesting is that where able to detect impending problems. With equipment, many months before they happen. So that you can prepare to do maintenance on them. And of course, if you know a piece of equipment are about to fail, you can also reduce the amount you use it. And adjust your productions so it all fit together. So, the practical implication is that you can offer it more efficiently and you can also begin to plan what you need to repair, or the new equipment you need to sort out the problem, preeminently.  

 

Silvija: I was involved in a renewal company some time ago. And we produced some of the biggest Hydro power “plans” in Europe. And I remember there was a situation with one of the biggest power “plans” in Norway. Where one of the water turbines simply started to rotate the wrong way. And it took forever to get that thing replaced.

 

Michael: They are huge, difficulty to transporter and in the case of ”winter ..” they may be geographically very difficult locations. So, the more months of notice you can have an advance if the problem is likely to happen, the better.

 

Silvija: So how does it look for the users. Somebody looking at this sets of screens, there are green blips and red blips. How do you deal with the output of the big data system? 

 

Michael: You have a slightly different users in any company. With slightly different needs. So, each one will get consuls which are appropriate to them. Typically, you will see an overview. With the most important data. Such as the production or status now. And you will also see alarms about these impending problems of the system. And the system will normally use many methods to form a confident decision that the problem is about to happen, rather than just relying on one.

 

Silvija: So you are making the humans that runs these systems smarter and more officiant. But in some cases, you are also making the system self-driving. So, they need to understand in time when something is about to go wrong and be adjust. And this is really important in your “autonomous” ship example. 

 

Michael: In this example. In almost all of these industrial examples we are using the data to empower humans or to make decisions. Either help humans to make decisions or systems to make decisions them self. 

Which means it's important that we know what data we are receiving. What quality the data is and how it is used to form decisions. 

 

Silvija: So, the problem about big data is to know what to do with the data. And to know what you want the data to do for you. How can you make people do that right?

 

Michael: You need to figure out if the data contains the answer to the question you going to ask from it. You can make a mapping of the data you have. You quickly see what data you are missing. And then look for alternative methods of creating that data. Not all problems can be solved buy data driven approaches. In some cases, we must use other kinds of models such as physics, mathematics to build the complete picture of a situation.

 

SS This is a really difficult point. Basically, big data can give you a lot of information about all the usual stuff.  But what you want is the unusual stuff which fortunately happens rarely. Like a turbine breakdown. You say you can simulate things. You don't have to break things in the real world you can break them in the digital world, and then you can figure out how to recognize that the next time it’s happening in the real world. 

 

Michael: Yes. For example we have a simulator for the industrial processes. And if you can a imagine situation, we have a valve in a real “plant” that is used to control the flow of a liquid and are used to control the pressure and the temperature on either side of the valve. And what you probably would like to do is to know what would happen if you turn the valve to know how to prefect the production. Just having data from the industrial plant, you may never have any data about the other positions that vale has been to and it may never have ever been moved and that's when we need to use his physics space stimulators. To calculate using physics space method. 

 

Silvija:Physics space method means you use fluid dynamic flaws to build a model?

 

Michael: Yes.

 

Silvija: You also use something called digital twins. Just the name is very cool. Can you tell us about that project? 

 

Michael: Definitely.  Big data projects in my world is often about combining data from many sources. So additional twin is a complete synthesized representation. In this example of an all gas industrial assist platform.  

 

Silvija: And you build a real platform in the digital world? 

 

Michael: Yes. you combine both 2D models, 3D models, live data from sensors. Live date from simulations which are running synchronist with the productions. 2D drawings, diagrams, Video if you have that.  

 

Silvija: You trying to make a digital version as live, precise and real as possible. 

 

Michael: Exactly. And you can build this into application, and it can be very useful before the real physical assits even exists as part of the planning process. It can be very useful when you own an assist for optimizing the production or training. Or familiarizing stuff with equipment before they're there. And of course, when you have all this data together. You can begin to do these AI machine learning type of things to either automate and predict problems.

 

Silvija: One of my favorite examples from computer science. Has been a project run by a lady called Marie Rognes. At the similar research lab. She does fluid dynamics to simulate flow of liquids in our brains and hart to predict heart failure, or to predict long term nerval degeneration. Witch are the stuff you will be bother by once we get really old. 

 These are the sort of experience you can't to do on real people. But it is amazing how much you can learn when your model gets really good.

 

Michael: Its incredible. I really believe the combining simulators and calculations with real life data from the real operational systems. It is it is a good way forward.

 

Silvija: Could you tell us about the Sikom example? 

 

Michael: Sikom is a big data product that we have been using in Kongsberg digital for many years. And it is used to broadly worldwide, for oil and gas drilling operations which are very large and costly operations. Performed by very many companies simultaneously. And what this software does is gathers the data from the rig in this case, where the drilling is happening and streams it all into one platform. Which is then used as a collaboration platform for all of the people involved to in real time troubleshoot but are also to documents to advice to detect similar situations and problems to provide advise.

 

Silvija: So it is a planning tool for oil and gas project? 

 

Michael: Yes. For drilling. 

 

Silvija: I also want to touch on your time in opera software. It is a company we both loved. And you told me that is where you got into big data. Can you tell us about that situation?

 

Michael: Sure. I have worked in the company fore two longer periods over the last 10 years. And the first time I ran into big data was when opera mini. Which is a mobile phone browser, first became very popular. So, the data started to flood in about how our users are using our products. And we need to start counting basically about how many users we had and what they were doing. So that we could invoice our customers. And that was before these well-known big data company’s like “..” and “..” where stable and workable.

 

Silvija: What do you think Norway is uniquely good at?

 

 Michael: So, I'm not Norwegian as you know. But I have been here for a very long time. My imperious is Norwegian is very forward looking. Very willing to adopt technology. Very curious and are interested in successfully use technology to find the strengths in what it is to be a relatively small nation. 

 

Silvija: If people want to learn more about big data. Where would you tell them to go?

 

Michael: It is a huge field. If you want to get inspired, I would really recommend going to a company like Nvdia who make the hardware which is used to AI machine learning and look at the keynote speech from a few months ago, where they have a good hour section on how data can be used to make new kinds of services in our society.

 

 Silvija: The funny thing about that company is that they started with graphic processors for the gameing industry. But this is also a really interesting company. They had to process thons of graphics, and then they realized the work could be used on other types of data as well. 

 

Michael: They turned out to be brilliant for all the complications of AI. 

Another thing to get inspired of is the large cloud operation systems have a lots of tutorials and information about how to build big data systems. 

Go to the Google cloud technology websites and look at how they propose that technology can be used in different industries and do the tutorials. Another place to get free good education is just to go to the course area find a tutor you like, find the course you like hop and jump a bit between them because they vary. And get on with it. 

 

Silvija: What's the one image people should have in mind listening to this podcast.

 

Michael: I think it's important to realize we are already doing this. It is big data that a bening used in industry, as well as in the Internet. It's much more than technology. Is this beautiful soup of technology, people, data, AI. All of the things you need to combine. And the gold is almost always to work more safely and officiant.

 

Silvija: Michael Link, thank you so much for helping to see the beauty of data at Kongsberg digital.

 

Michael: Thank you. 

 

 

 

 

 

Quiz for Case #C0045

Du må være Medlem for å dokumentere din læring med å ta quiz 

Allerede Medlem? Logg inn her:

Du må være Medlem for å kunne skrive svar på refleksjonsspørsmål

Allerede Medlem? Logg inn her: