Monitoring Kyle Saltmarsh The Outline The Identity Crisis - PowerPoint PPT Presentation

Young Persons’ Lecture Competition Acoustic Based Condition Monitoring Kyle Saltmarsh

The Outline • The Identity Crisis • Acoustic Based Condition Monitoring • The How • Applications in Industry • Conclusion

The Identity Crisis (About Me)

Acoustic Based Condition Monitoring • Sounds tell stories • You do this all the time • On-site technicians do this • Technology does this

An Example in Action

Problem Statement How do you use acoustics to monitor equipment in noisy environments?

Beamforming Robust Adaptive Beamforming - Jian Li & Petre Stoica

Beamforming Applications

Signal Processing Fourier Techniques and Applications - Joseph Fourier

Extracting Acoustic Features Feature #1 Feature #2 Healthy Bearing Unhealthy Bearing

Model: Neural Network Feature #1 Output #1 Equipment Type Feature #2 Output #2 Health Status

Problem Solved

Unearthed Mine Hackathon 2017 How can we more effectively use sound in mining?

Industry Data: Woodside Fin Fans Healthy Vs. Corroded Bearing

Microphone Array

Beamforming

Signal Magnification

Time Series

Acoustic Feature Extraction Feature #1 Feature #2

Model Performance: Confusion Matrix True Condition Corroded bearing Fin Accuracy: 99.5% Normal Fin Fan Fan Normal Fin Fan 1 176 Prediction Corroded 2 167 bearing Fin Fan

Conclusion • Sounds tell stories • Acoustics, signal processing and predictive modelling • Application in Industry (Rio Tinto) • Easy Implementation

Thank You Questions?

Contents • The Identity Crisis • Rio Tinto RailBam • Beamforming • Signal Processing • Feature Extraction • Predictive Model • Unearthed 2017