Mid-Infrared Spectrum Sensing Algorithm Applying Micro-Resonator Arrays

A mid-infrared (mid-IR) integrated photonic sensing module using the machine learning (ML) technique has been illustrated. The sensing wavelengths were determined via a wavelength selection model trained by a classification algorithm according to the mid-IR absorption characterization of the analytes. The spectrum of the device output ports was identified via the algorithm-enabled spectrum selection method, where two artificial neural networks (ANN) were trained to optimize the device structures. The module with the micro-resonator array can precisely monitor various gas mixtures by altering the micro-resonator geometry. Our miniaturized mid-IR sensing module enables a new technology for accurate and efficient gas detection.