Physics-Guided Concentration Inference from Resistance Transients in a Mixed-Phase SnO-SnO$_2$ Carbon Monoxide Sensor with p-n Switching
Researchers have developed a physics-guided machine learning framework to infer carbon monoxide concentrations from gas sensor data. The system analyzes resistance transients in a mixed-phase SnO-SnO2 material, utilizing physically interpretable descriptors and signal processing techniques like FFT and DWT. The study found that fused features generally performed best, but physics-guided descriptors were highly competitive, indicating that transient dynamics already encode significant concentration information. The framework demonstrated distinct behaviors for p-type and n-type sensing regimes, with p-type excelling in classification and n-type in regression. AI
IMPACT This research demonstrates a novel application of physics-guided machine learning for sensor data analysis, potentially improving accuracy and interpretability in environmental monitoring.