Hack the Wind image

Boldare Machine Learning Team in the final 3 of Hack the Wind

With the goal of resolving serious problems in the wind energy industry, Boldare’s team of five developers and a product designer took part in a prestigious hackathon. For 48 hours, they worked relentlessly to provide a machine learning-based product to predict wind turbine failures.

Hackthewind 6158155782 E 5 A 3 B 61 A 76 C 5 D 5162 D 9 Ff 4

About
Hack the Wind

Hackathon team

This is our chance to make an impact in the future of renewable and clean energy - the future of our planet.

Paweł Capaja

Hack the Wind 2018 is a second edition of a hackathon hosted by InnoEnergy and Wind Europe during the Global Wind Summit. More than 20 development teams competed in a challenge to build a product that uses machine learning to forecast wind turbine breakdowns. It wasn’t an easy task as judges not only looked at the machine learning model the teams created but also the user experience and user interface of the product that they have built.

The teams from around the world worked almost non-stop for two days and nights in a unique environment which allowed them to focus only on the task at hand: come up with new ideas, test them, and innovate at full speed.

The Boldare Machine Learning Team was one of the crews taking part in the competition. You can follow their journey via our Twitter feed @boldarecom and #BoldareHacksTheWind.

Team

Machine Learning Team

Deliverables

A machine learning-based solution
to predict wind turbine failures

Timespan

48h

Cost

Pro Bono

Organised by

Challange sponsor

Boldare Machine Learning Team

The Boldare team has a year and a half’s experience in machine learning. On a day-to-day basis, they consult our clients on the best solutions for their business needs as well as develop and implement artificial intelligence solutions.

Tomasz Bąk photo

Tomasz Bąk

Machine Learning Developer

Paweł Krynicki photo

Paweł Krynicki

Machine Learning Developer

Maciej Papież photo

Maciej Papież

Machine Learning DevOps

Sławomir Zagórski photo

Sławomir Zagórski

Machine Learning Developer

AI, Machine Learning and Deep Neural Networks can help you with a number of problems

The team has really special qualities: they are analytical, business goals-oriented, lean, agile, cross-functional and knowledge-hungry.

Damian Kozar photo

Damian Kozar

Frontend Developer

Paweł Capaja photo

Paweł Capaja

Product Designer

Boldare Machine Learning Team

The Boldare team has a year and a half’s experience in machine learning. On a day-to-day basis, they consult our clients on the best solutions for their business needs as well as develop and implement artificial intelligence solutions.

The team has really special qualities: they are analytical, business goals-oriented, lean, agile, cross-functional and knowledge-hungry.

  • Tomasz Bąk photo

    Tomasz Bąk

    Machine Learning Developer

  • Paweł Krynicki photo

    Paweł Krynicki

    Machine Learning Developer

  • Maciej Papież photo

    Maciej Papież

    Machine Learning DevOps

  • Sławomir Zagórski photo

    Sławomir Zagórski

    Machine Learning Developer

  • Damian Kozar photo

    Damian Kozar

    Frontend Developer

  • Paweł Capaja photo

    Paweł Capaja

    Product Designer

We'd love to help with your product. Let’s talk

Problem

CHALLENGE 1

Sponsored by: EDPR

Wind Turbine Fault Prediction

Predicting components failure in wind turbines is key for the wind energy industry. It decreases time and costs related to maintenance and cuts the disruptions to energy production to a minimum. It has also an impact on life-span of components and wind turbines as a whole. This enables wind energy suppliers to stabilize their costs and optimize and level the energy production.

The teams taking part in the Hack the Wind received a year-worth of data regarding failures of 5 components of a five wind turbines from a real wind farm. The winners have to detect the greatest number of failures by building up machine learning models for effective errors prediction and thus made the highest maintenance cost savings, which in case of wind turbines, are counted in millions of Euro. Finally, the teams were asked to to present the solution in form of a product that is not only useful but also user-centered, and thus the design and technical solution work in harmony.

Ml Team In Helicopter B 581 B 85993311 A 1287622 A 1 E 358 A 0639

We were in TOP 3 in Hack the Wind Hackathon

For 48 hours, Boldare Machine Learning team worked relentlessly to provide a machine learning-based product to predict wind turbine failures. You can read the summary of their approach to the problem bellow.

Solution

The Boldare Machine Learning team approached the problem from a business perspective. After a short product workshop they came up with an initial product canvas, however, they decided that they need a proof of concept. The team went straight to the target group using to their advantage the Wind Energy Hamburg trade show, which was taking place in the same building as the competition. What they found out from people in the industry change their idea for the product and made them pivot. It turned out that the real problem for the wind farm owners was lack of product that would seamlessly interconnect the predictive maintenance data within the maintenance flow. The team designed a product that would not only inform the owner about future malfunction but also allow him to react to it either by sending an inspection crew on site or ordering an immediate part replacement.

Wind Europe Organizers

Boldare proposes to unify the process of maintenance with the help of an application and AI!

@InnoEnergyIB

Ml Team In Action 2 E 8 Fb 4 A 36723 E 8 Fa 73099 Ea 4 A 896 A 7 Bf 7
Ml Team In Action 055081 C 3368748515 A 9 Ba 572 D 750 F 3 Da

Real impact

We delivered this solution during the time frame of

48h

Boldare Machine Learning team proposed individual machine learning models for each component within a wind turbine, that can predict failures up to 60 days before they occur.

The results and alert are displayed within the app, showing which particular turbine is likely to fail, the likelihood of a breakdown within the next 24 hours, and information, which part will malfunction.

Ml Team Group Photo Ec 2 C 0829 B 20 C 633560 D 851 F 2 C 4102 B 6 D
Hack The Wind Mockups 93 Ab 1 Fc 952 D 8 Fc 9 Cfb 5 Fc 7633 E 260 Fe 4

In the next releases the team proposed that the app could be expanded with inventory updates as financial reports.

See our case study.

Let's talk

Do you want to build a product, hire a development team or simply have a question? Leave us a message.

Phone:

+48 32 739 09 00

hello@boldare.com

career@boldare.com