Deep Autoencoder Acoustic Reconstruction
This dashboard is powered by a Deep Learning Neural Network trained exclusively on the sounds of a perfectly healthy machine. It doesn't know what a broken machine sounds like; it only knows normal operation.
When live sound enters the system, the AI tries to reconstruct it. If it hears unexpected rattling or grinding, it fails to rebuild those sounds correctly. We track this reconstruction error on the top chart. If the error spikes above the yellow dashed line (Threshold: 0.0036), a mechanical failure is flagged!
The bottom heatmap is the machine's "acoustic fingerprint." It breaks sound down into frequencies over time. A healthy machine shows smooth, consistent blue/purple waves. A failing bearing or wobbly fan will appear as bright yellow/white vertical spikes, highlighting the exact frequencies that are failing.