Experimental Demonstration of Multimode Microresonator Sensing by Machine Learning

A multimode microcavity sensor based on a self-interference microring resonator is demonstrated experimentally. The proposed sensing method implemented by recording wide band transmission spectra that consist of multiple resonant modes. It different from previous dissipative scheme, which aims at measuring the depth changes single mode in microcavity. Here, combining mechanism and machine learning algorithm, information extracted broadband spectrum can be efficiently fused to estimate target parameter. immune laser frequency noises robust against system imperfection, thus our work presents great step towards practical applications sensors outside research laboratory. voltage applied across microheater chip was adjusted bring its influence transmittance through thermo-optic effects. As proof-of-principle experiment, detected approach. experimental results demonstrate limit detection general regression neural network reduced 6.7% single-mode within large range.