Design of a high-precision capacitive dew point meter based on physics-guided residual learning and four-phase differential TDC
Measurement Science and Technology · 24.04.2026 · 37 (17) · 175101
DOI: 10.1088/1361-6501/ae5fbc ↗Kurz gefasst
Fachartikel in der Zeitschrift Measurement Science and Technology, online erschienen am 27.04.2026. Verfasst von Baicheng Chen und 3 weiteren Autor:innen. Die Arbeit nennt 25 Literaturangaben.
Abstract (Originalsprache)
Capacitive dew-point measurement systems are essential for industrial humidity monitoring, but achieving high precision is often hindered by the coupled challenge of parasitic capacitance and sensor nonlinear hysteresis. To address this issue, this paper proposes a high-precision measurement system based on hardware-algorithm co-design. On the hardware front, a four-phase differential time-to-digital converter measurement method is presented, which systematically suppresses both external and internal parasitic capacitance through dynamic switch matrix timing reconstruction, thereby enhancing the extraction capability for picofarad-level micro-capacitance signals. Algorithmically, a physics-guided residual learning framework is constructed: the Magnus equation serves as the physical benchmark to ensure extrapolation consistency of the model, while a lightweight XGBoost regressor is introduced to learn the nonlinear hysteresis residual, compensating for the sensor’s inherent nonlinearity and hysteresis effects and effectively mitigating inversion error amplification in low-humidity regions. The system is deployed on an STM32F411 microcontroller, with a single inference requiring only 330 μs. Experiments demonstrate that within a dew-point range of −40 °C to 5 °C, the mean absolute error is 0.1217 °C and the maximum absolute error is 1.6843 °C, significantly improving the consistency and reliability of full-range measurement. This work provides an integrated solution for industrial dew-point monitoring, featuring high accuracy, strong anti-interference capability, and edge computing performance.
Zitieren
APA
Chen, B., Gu, M., Shi, J., & Shen, K. (2026). Design of a high-precision capacitive dew point meter based on physics-guided residual learning and four-phase differential TDC. Measurement Science and Technology, 37(17), 175101. https://doi.org/10.1088/1361-6501/ae5fbc
Vancouver
Chen B, Gu M, Shi J, Shen K. Design of a high-precision capacitive dew point meter based on physics-guided residual learning and four-phase differential TDC. Measurement Science and Technology. 2026;37(17):175101. doi:10.1088/1361-6501/ae5fbc
BibTeX
@article{Chen2026,
author = {Chen, Baicheng and Gu, Minming and Shi, Junlong and Shen, Kaikai},
title = {Design of a high-precision capacitive dew point meter based on physics-guided residual learning and four-phase differential TDC},
journal = {Measurement Science and Technology},
year = {2026},
volume = {37},
number = {17},
pages = {175101},
doi = {10.1088/1361-6501/ae5fbc}
}
EUVIMED