← Back to portfolio

Korea Univ. DSBA × Hanwha Systems (ICT) industry–academia project · Jan 2020 – Sep 2021

Predictive Maintenance for Chemical Processes

Role · Anomaly-pattern analysis, anomaly-detection algorithms

01Background & Goals

  • An industry–academia project to detect warning signs in equipment, find their causes and let engineers act before a failure.
  • We applied anomaly detection, time-series and pattern analysis to chemical-plant sensor data (vibration, sound, rotation speed, temperature, pressure).

02Contributions

Detection and forecasting models
  1. Built an RNN auto-encoder anomaly detector to keep equipment running
  2. Forecast future patterns from historical data with an RNN time-series model
Explanation and monitoring
  1. Found similar past patterns with LSH and separated normal from anomalous patterns for analysts
  2. Built a real-time monitoring system driven by sensor status updates

03Tech Stack

Framework / Platform
PyTorch, scikit-learn
Methodology
Anomaly detection, RNN auto-encoder, Locality-sensitive hashing

04Results

  • Published in Applied Soft Computing 125 (2022), third author
  • Presented at the 2021 KIIE Spring Joint Conference

05Reference Material