High-precision microwave photonic pressure sensor assisted by machine learning post-processing

A high-precision microwave photonic (MWP) pressure sensor assisted by machine learning (ML) post-processing is proposed. The proposed sensor operates by virtue of MWP interrogation and advanced ML post-processing based on least-squares support vector regression (LSSVR) and, hence, eliminates temperature effect and achieves high-resolution pressure measurement. The theoretical analysis and experimental results of the proposed MWP sensor are presented and discussed. When the temperature varies from 35°C to 39°C, the mean absolute errors (MAEs) of the sensor are less than 0.9 kPa. The estimation result can satisfy the body pressure monitoring at different temperature conditions.