Dual oxygen and temperature luminescence learning sensor with parallel inference

TL;DR

This paper introduces a neural network-based sensor that simultaneously measures oxygen and temperature from optical data without relying on complex models, achieving high accuracy and broad applicability.

Contribution

It presents a novel parallel inference learning sensor that extracts multiple parameters from optical measurements without prior mathematical models, surpassing traditional methods.

Findings

Achieves accurate simultaneous oxygen and temperature sensing.

Introduces the Error Limited Accuracy metric for sensor performance.

Applicable to various luminophore-based sensing scenarios.

Abstract

A well-known approach to the optical measure of oxygen is based on the quenching of luminescence by molecular oxygen. The main challenge for this measuring method is the development of an accurate mathematical model. Typically, this is overcome by using an approximate empirical model where these effects are parametrized ad hoc. The complexity increases further if multiple parameters (like oxygen concentration and temperature) need to be extracted, particularly if they are cross interfering. The common solution is to measure the different parameters separately, for example, with different sensors, and correct for the cross interferences. In this work, we propose a new approach based on a learning sensor with parallel inference. We show how it is possible to extract multiple parameters from a single set of optical measurements without the need for any a priori mathematical model, and with unprecedented accuracy. We also propose a new metrics to characterize the performance of neural network based sensors, the Error Limited Accuracy. The proposed approach is not limited to oxygen and temperture sensing. It can be applied to the sensing with multiple luminophores, whenever the underlying mathematical model is not known or too complex.