Efficient Bayesian inference of absorbance spectra from transmitted intensity spectra.

High-resolution absorption spectroscopy is a promising method for non-invasive process monitoring, but the computational effort required to evaluate the data can be prohibitive in high-speed, real-time applications. This study presents a fast method to estimate absorbance spectra from transmitted intensity signals. We employ Bayesian statistics to combine a measurement model with prior information about the shape of the baseline intensity and absorbance spectrum. The resulting linear least-squares problem shifts most of the computational effort to a preparation step, thereby facilitating quick processing and low latency for any number of measurements. The method is demonstrated on simulated tunable diode laser absorption spectroscopy data with additive noise and a fluctuating fringe. Results were highly accurate and the method was computationally efficient, having a processing time of only 2 ms per spectrum.

Data analysis and uncertainty estimation in supercontinuum laser absorption spectroscopy.

A set of algorithms is presented that facilitates the evaluation of super continuum laser absorption spectroscopy (SCLAS) measurements with respect to temperature, pressure and species concentration without the need for simultaneous background intensity measurements. For this purpose a non-linear model fitting approach is employed. A detailed discussion of the influences on the instrument function of the spectrometer and a method for the in-situ determination of the instrument function without additional hardware are given. The evaluation procedure is supplemented by a detailed measurement precision assessment by applying an error propagation through the non-linear model fitting approach. While the algorithms are tailored to SCLAS, they can be transferred to other spectroscopic methods, that similarly require an instrument function. The presented methods are validated using gas cell measurements of methane in the near infrared region at pressures up to 8.7 bar.

PolySpec: polynomial spectrum models for fast and light-weight spectroscopic evaluation.

For the analysis of spectroscopic data, model fitting approaches are commonly applied. The spectrum model applied in these fitting processes significantly influences the performance of the spectroscopic evaluation, which can be critical in real-time process diagnostics and control. In this work a spectrum model is introduced that uses a polynomial description of absorbances, transmittances, or similar in dependence on parameters such as temperature, pressure, and mole fraction. Using this approach, either experimental spectra or spectrum databases can be compressed into a matrix of polynomial coefficients. The evaluation of this model consists of a single matrix multiplication and, with a slight modification, derivatives with regard to specific parameters can be calculated in the same way. Both these points are important to model fitting methods for spectroscopic data, as the simple evaluation method allows for a fast analysis and the direct calculation of derivatives simplifies the application of gradient-based fitting methods. Additionally, the easy parallelizability of the matrix multiplication promotes the application of this method in real-time evaluations on programmable logic devices.