Leveraging AI and Pattern Matching to Automate Fault Analysis in Power Systems - Webinar

Leveraging AI and Pattern Matching to Automate Fault Analysis in Power Systems - Webinar

Title history
  • Leveraging AI and Pattern Matching to Automate Fault Analysis in Power Systems - Webinar — observed on 20261027-1005 UTC (YouTube)

by Qualitrol · View the channel on PSI

Applied Sciences & Engineering Energy THEnergy technology and engineeringTHRElectrical engineering
🎙 Qualitrol 👥 609 📅 October 27, 2026 ⏱ 39 min 👁 6 📄 expert opinion 🧭 2026-10-27
Available in: English (current) Français

Keywords

AIpattern matchingfault analysisCOMTRADErandom forestpower grid

Summary

This webinar, hosted by Power Systems Technology and presented by Qualitrol product managers Mark Diamond and David Cole, addresses the challenge of automating fault analysis in electrical power systems. The speakers highlight the high cost of outages ($100k-$200k per hour) and the fact that 50% of faults lack root cause. They introduce a machine learning solution trained on 1.5 million COMTRADE records, using a random forest algorithm to categorize fault events with 97% accuracy. A 10-month pilot at Scottish Power validated the approach on 23 devices, correctly identifying all but two events and even flagging a VT failure from earlier. The presentation includes a live demo of the Grid Assist software, which merges waveforms from both ends of a line, calculates fault duration and breaker timing, and maps events geographically. They also compare the before-and-after scenario of a storm, showing how automation reduces analysis time from hours to minutes. The webinar concludes with a Q&A session and emphasizes a supervised learning model where user corrections improve the algorithm over time.

170 words

Critical Evaluation

Value of the Information & Strength of the Argument

The value of the information is high for utilities facing overwhelming fault record volumes. The presenters provide concrete statistics, real-world pilot results, and a logical flow from problem to solution. The argumentation is solid, relying on quantitative evidence (97% accuracy, 300 records/day at Scottish Power) and visual demonstrations. They effectively explain why rule-based systems fall short and how pattern matching handles diverse fault patterns. The cost-benefit analysis strengthens the case for automation. However, the presentation is vendor-driven, lacking independent validation, which slightly weakens the overall value.

Scientific Rigor, Source Quality, Title Accuracy

The rigour is moderate: the methodology is described transparently (data labeling, common sample rate, class imbalance handling), and the pilot is well-documented. Sources are mostly internal (Qualitrol’s own data and engineers), with no external references beyond NERC and DOE statistics. The title accurately matches the content, and the webinar stays focused on the claimed topic. No significant discrepancies between title and content were noted.

165 words

Title / Content Match

The title accurately reflects the content: a webinar specifically about using AI and pattern matching for power system fault analysis, with a demonstration of the Grid Assist software.

Quality & Reliability

7/10

The webinar provides specific quantitative claims (97% accuracy, 1.5 million records, pilot details) but these are vendor-generated and not independently verified. The presentation is professionally structured and transparent about methodology, but lacks external peer review.

Chapters

Cited Sources

Contribution & Novelties

The main contribution is demonstrating a practical AI system for power system fault triage, trained on an unprecedented 1.5 million COMTRADE records, achieving 97% accuracy in categorization and extending to fault cause identification. The 10-month pilot at a major utility provides real-world validation, and the software demo illustrates integration with legacy recorders. The approach of using patterns to distinguish lightning, tree contact, crane contact, and VT failures offers a scalable way to automate root cause analysis. The proposed feedback loop to improve the model globally is innovative.

Pour aller plus loin :

  • COMTRADE (C37.111 IEEE standard) — Standard file format for power system transient data exchange, foundational to the described solution.
  • Random Forest — The machine learning algorithm used for classification, known for handling high-dimensional data.
  • NERC State of Reliability Report — Source for the 50% root cause claim, provides reliability trends.
  • Power System Protection and Fault Analysis — Background on protection relays and fault analysis concepts.

157 words

Radar Profile

The profile shows high quantities of information and moderate technical depth, with overall reliability supported by specific claims but limited by lack of independent validation. The strongest dimension is the quality of information presentation, while the lowest is reliability due to vendor bias.

Reliability 6/10