Evidence Demo
Switchgear AI-Olfaction Multimodal Fusion Sensing System
Front-end simulation of a laboratory step-voltage test (6 kV → 20 kV) accelerating insulation degradation. Not live field data.
Channels CH1–CH9 measure the real-time resistance of metal-oxide-semiconductor sensors, while the trend chart displays RAF—the rate of change relative to a baseline that remains at 1 when stable. When insulation degradation releases characteristic gases, the channels respond in a cross-sensitive pattern to generate an electronic "electronic nose" fingerprint, triggering visual warnings and audible alarms first. Partial-discharge channels (AE, TEV, UHF) already show amplitude readings in the early stage; however, both count and amplitude remain below thresholds until degradation becomes severe, causing alarms to trigger sequentially only then—significantly extending the early-warning window. Lead time varies by condition; this demo is just one example.
- Test sequence: 6 kV → 9 kV → 12 kV → 15 kV → 18 kV → 20 kV; 60 sampling cycles per step
- Olfaction channels: CH1–CH9 display live MOS sensor resistance R(Ω); RAF is the rate of change relative to baseline and remains 1.000000 when unchanged.
- PD alarm condition: discharge count ≥ 15 and amplitude ≥ 20 dB; AE, TEV, and UHF trigger sequentially as degradation progresses.
- The alarm tone is synthesized using Web Audio. It activates after the first click and can be muted at any time.