Transfer-learning-based multi-wavelength laser sensor for high fidelity and real-time monitoring of ambient temperature and humidity

Abstract

Multi-wavelength laser absorption spectroscopy has the advantages of superior sensitivity, accuracy, and robustness for gas sensing applications, offering an opportunity for the development of high-performance laser-based hygrothermographs. However, accurate and fast determination of gas parameters from multiple spectral features can be quite challenging in the presence of large numbers of features, measurement noise, and increasing demands for real-time measurements. To address this challenge, we propose a transfer-learning-based multi-wavelength laser absorption sensor for the quantitative and simultaneous measurement of temperature and concentration of water vapor, with a focus on real-time monitoring of ambient temperature and relative humidity (RH). A spectral simulation based on the most-updated HITRAN database was employed as the dataset for model pre-training and transfer learning. The experimental dataset was obtained from absorption measurements using a distributed feedback laser that probed multiple water absorption features within the band of 7179−7186  cm−1{{7179 - 7186}}\;{\rm{c}}{{\rm{m}}^{- 1}} . To evaluate the sensor performance, mean absolute error, error distribution, and linearity were selected. In the presence of an insufficient experimental dataset for direct data training, the proposed transfer learning approach outperformed the traditional deep learning method with a lower prediction error of 0.14°C and 0.42% for temperature and RH, respectively, as compared to the values of 0.84°C and 0.66% obtained using the traditional deep learning method. Finally, the fast data post-processing performance of the proposed transfer learning approach was demonstrated in a field test against the conventional baseline fitting method.

Full text

5932 Vol. 62, No. 22 / 1 August 2023 / Applied Optics Research Article
Transfer-learning-based multi-wavelength laser
sensor for high fidelity and real-time monitoring
of ambient temperature and humidity
Liuhao Ma,1,†Weifan Hu,1,†Wei Wang,1AND Yu Wang1,2,*
1Combustion and Laser Sensing Laboratory, School of Automotive Engineering, Wuhan University of Technology, Wuhan, 430070, China
2Foshan Xianhu Laboratory of the Advanced Energy Science and Technology Guangdong Laboratory, Xianhu Hydrogen Valley, Foshan,
Guangdong 528200, China
†These authors contributed equally to this work.
*yu.wang@whut.edu.cn
Received 12 May 2023; revised 9 July 2023; accepted 12 July 2023; posted 12 July 2023; published 25 July 2023
Multi-wavelength laser absorption spectroscopy has the advantages of superior sensitivity, accuracy, and robust-
ness for gas sensing applications, offering an opportunity for the development of high-performance laser-based
hygrothermographs. However, accurate and fast determination of gas parameters from multiple spectral features
can be quite challenging in the presence of large numbers of features, measurement noise, and increasing demands
for real-time measurements. To address this challenge, we propose a transfer-learning-based multi-wavelength
laser absorption sensor for the quantitative and simultaneous measurement of temperature and concentration of
water vapor, with a focus on real-time monitoring of ambient temperature and relative humidity (RH). A spectral
simulation based on the most-updated HITRAN database was employed as the dataset for model pre-training
and transfer learning. The experimental dataset was obtained from absorption measurements using a distributed
feedback laser that probed multiple water absorption features within the band of 7179 −7186 cm−1. To evaluate
the sensor performance, mean absolute error, error distribution, and linearity were selected. In the presence of an
insufficient experimental dataset for direct data training, the proposed transfer learning approach outperformed
the traditional deep learning method with a lower prediction error of 0.14◦C and 0.42% for temperature and RH,
respectively, as compared to the values of 0.84◦C and 0.66% obtained using the traditional deep learning method.
Finally, the fast data post-processing performance of the proposed transfer learning approach was demonstrated in
a field test against the conventional baseline fitting method. © 2023 Optica Publishing Group
https://doi.org/10.1364/AO.495482
1. INTRODUCTION
Accurate measurement of temperature and humidity is of
paramount importance in many fields such as manufacturing
of electronic products [1], meteorological observations [2],
indoor air quality control [3], and the storage of sensitive goods
(e.g., medicines or fine arts) [4]. Working environments hous-
ing electronic devices such as computing facilities and precise
instruments also have stringent requirements for temperature
and humidity [5]. Deviations from optimal working tempera-
ture and humidity ranges may cause performance deterioration
and, in extreme cases, irreversible damages. Precise monitoring
of these parameters is also necessary during manufacturing of
temperature- and humidity-sensitive IC chips [6] and fun-
damental fluid mechanical study in high-performance wind
tunnel and shock tube experiments [7]. In this regard, the
need for the development of precise sensors for simultaneous
measurement of temperature and humidity is clear.
Laser absorption spectroscopy (LAS) is a promising method
for temperature and humidity measurement; it utilizes the
fractional transmission of incident laser intensity to deter-
mine the gas properties (e.g., concentration and temperature)
when the laser interacts with gas molecules [8–10]. By fitting
the measured spectral features, calibration-free and quanti-
tative measurements can be achieved. This method has the
significant advantage of a calibration-free property against
conventional multi-sensor fusion measurement methods that
typically require a separate temperature sensor [e.g., ther-
mocouples [11], mercury thermometers [12], and negative
temperature coefficient (NTC) thermistor [13]] and a humid-
ity sensor (e.g., psychrometer, electronic humidity sensor, or
chilled-mirror dew-point hygrometer [14]). This multi-sensor
configuration reduces system robustness; perhaps more impor-
tantly, both of these conventional sensors have the drawback
of long-term drift and require frequent calibrations [15,16].
Over the past decades, LAS has been successfully used for
1559-128X/23/225932-14 Journal © 2023 Optica Publishing Group