Alpha-i wins Hack the Wind 2018, the Wind Turbine Fault Prediction hackathon at the WindEurope Global Wind Summit

Time is money and anticipating failures in wind turbines reduces maintenance costs and production losses due to unavailability – paving the way for lower costs of clean energy.

The Hack the Wind challenge focused on the development of an innovative and reliable solution to reduce the overall maintenance cost.

A solution that could anticipate the failure of the most critical wind turbine components and recommend a maintenance strategy.

Challenge Highlights

  • 2 years of data:
    • 5 wind turbines
    • meteorological mast
  • 4 monitored subassemblies:
    • Generator
    • Gearbox
    • Transformer
    • Hydraulic group
  • Cost weighted evaluation

Our team was awarded the Grand Prize of the challenge as the solution that detected the most early stage failures, enabling the highest cost savings.

We achieved this by combining domain knowledge with machine learning to identify abnormal component behaviour and recommending targeted maintenance actions.

Alpha-i also received the innovation an In-Kind innovation prize from InnoEnergy for the solution with the highest business potential.

Partners

Hack the Wind 2018 was organised by:

Hack the Wind 2018 was sponsored by:

Hack the Wind was a great learning experience both in terms of technology and domain knowledge. It was 48 hours of non-stop data analysis and an opportunity to interact with new people in the industry. The EDP Renewables team was very supportive and helped us a lot in understanding the data. During the hackathon, we evaluated a two-fold approach. Firstly, the statistical properties of the SCADA data were analyzed to formalize rules for detecting anomalies. However, lack of enough fault data made it difficult to train our machine learning models. Secondly, using our visualization platform, we looked for apparent visual signals prior to a fault. Using domain knowledge, some of the faults could be easily detected well in advance using just visual signals characteristic to those faults. In conclusion, we found that data visualization and domain knowledge can help label turbine data to make machine learning more effective.

Parvez Alam Kazi – Machine Learning Engineer