20 May 2024 in IEEE Transactions on Instrumentation and Measurement
This paper presents a novel gas concentration detection approach based on wavelength modulation spectroscopy and linear convolution (LC-WMS). The linear convolution is first time utilized in the WMS measurement system to demodulate the absorption spectrum, and then to obtain the absolute harmonic signals related to the gas concentration. Theoretical and experimental results show that when the absorption signal is convolved with a whole-cycle reference signal, the harmonic signals can be acquired without the need of lock-in amplifier (LIA) and low-pass filter. The linear convolution calculation process is regarded as a combination of demodulation and low-pass filter as in the conventional demodulation system (e.g. LIA). Simulations demonstrate that appropriately increasing the cycles of reference signal (not exceeding five cycles) can help to decrease the noise of the harmonic signals. Experimental results show that the proposed LC-WMS approach can obtain a comparable accuracy as the conventional LIA in CO 2 and H 2 O concentration measurement with concentration errors less than 5 ppm and 0.05‰, respectively. The long-term measurement shows that this method can be used for real-time detection with large concentration fluctuations. Since the linear convolution process gets rid of the problem caused by the magnitude gain response in low-pass filter, the proposed LC-WMS method can not only obtain the absolute harmonic signal without calibration but also simplify the detection process. Thus, it has potential applications in industrial field measurements.
24 July 2023 in IEEE Transactions on Instrumentation and Measurement
Wavelength modulation-based tunable diode laser absorption spectroscopy (TDLAS/WMS) is widely used in various manufacturing processes. On the aseptic preparation filling production line, the seal quality of pharmaceutical glass vials should be strictly inspected in situ, where the 2 nd harmonic signal is usually selected for headspace residual oxygen measurement due to its prominent amplitude natures. While, higher-order harmonics still contain reliable concentration information but are often ignored. In this paper, we first design a laser injection signal with dynamic modulation depth and then propose a new methodology of multiple harmonic information fusion (MHIF), which is simple but promotes the mining of more effective concentration information from the absorption spectrum. Based on the fusion analysis results, an MHIF-guidance gas concentration inversion model is designed. To verify the performance of the proposed inversion model, a TDLAS/WMS-based residual oxygen detection platform is built. In this system, we use the harmonic signals from the 1 st to 8 th cycles together to perform precise measurement of the oxygen concentration in pharmaceutical glass vials. The results show that the stability of the new MHIF-based feature can exceed that of the traditional 2 nd harmonic features by 24.7%. Moreover, the new inversion model can increase the detection accuracy by over 10% compared to the traditional inversion model.
20 November 2025 in Information Fusion
26 January 2026 in IEEE Transactions on Instrumentation and Measurement
Tunable diode laser absorption spectroscopy (TDLAS) based on wavelength modulation spectroscopy (WMS) is widely used for gas concentration measurement due to its rapid detection, high accuracy, and non-destructive nature. However, traditional WMS produces unreliable results for residual oxygen detection in pharmaceutical vials due to laser attenuation caused by glass walls. Although advanced methods such as ln-WMS and WMS-2f/f attempt to overcome this limitation, they suffer from algorithmic inflexibility or high computational cost. This paper introduces ln-Hilbert-WMS, an improved technique that integrates natural logarithm processing with the Hilbert transform. This method effectively isolates gas absorption components from laser intensity variations while simultaneously enabling harmonic extraction, laser intensity elimination, and offset noise suppression. A detection system for oxygen within pharmaceutical vials is established for validation. Experimental results demonstrate that the method can accurately measure the oxygen concentration without being affected by variations in vial transmittance. Furthermore, comparative analysis under different offset-noise conditions, together with FPGA implementation results, shows that the proposed method provides greater algorithmic generality than ln-WMS while requiring lower computational cost than WMS-2f/f. In pharmaceutical vial testing, ln-Hilbert-WMS achieves the best detection performance, with a standard deviation of 0.45% and an RMSE of 0.51%, outperforming traditional WMS, ln-WMS, and WMS-2f/f techniques.