October 25 Asset Health Community - Predictive Maintenance of Utility Assets with Agile ML & GenAI

When:  Oct 25, 2023 from 14:00 to 15:00 (CT)

Join the Asset Health Analytics Community on Wednesday, October 25, 2023 at 2:00 PM CT for a presentation with Q&A and open discussion on, "Predictive Maintenance of Utility Assets with Agile Machine Learning and Generative AI," led by the following speakers Sapient: 

  • Supratik Chaudhuri, Director and Industry Principal - Publicis Sapient
  • Claus Agerskov, Lead Data Scientist - Publicis Sapient
  • Ramon Perez, Director, Data Science - Elder Research

Session Description: Publicis Sapient's client, Sira-Kvina, generates 5% of Norway's energy from seven hydro power sites. It wanted to build predictive models using its IOT and sensor data to determine hydro generator component failure risk and do preemptive maintenance. 
The team was able to leverage time series SCADA data + control systems + protection systems + maintenance data, across 16 generators. 
- Visual abstract model of the generator systems and components.
- Machine learning sprints to create ML and survival analysis models, 6 years of historical data.
The solution reduced complexity from 1000 to 10 most critical sensors. It also enabled better planning of periodic maintenance schedules and improved predictive performance 5x times.
Next steps:  Prototyping the use of large language models (Gen-AI) to determine if operator logs can serve as indicators of upcoming equipment problems
Session Takeaways - Attendees will learn how to:
1. Apply AI/ML to asset component performance and failure
2. Use the insights to align maintenance cycles to improved asset performance
3. Use AI/ML to decompose complex asset models to critical points of failure
4. Deliver AI/ML projects through agile model
5. Discuss findings of LLM (Gen-AI) model implementation

Members of the Asset Health Analytics Community will receive a Microsoft Teams link prior to this meeting. Not a member of this community? UAI Utility Members can register for any of our communities by using our Request To Join link. 


Dial-in Instructions:
Microsoft Teams link will be provided to UAI Community Members prior to this meeting.


Kevin Praet