Spectral fits with full broadening for kHz-iteration high-accuracy gas-phase spectroscopy

Abstract This work presents methods enabling real-time multi-species direct absorption spectroscopic fitting at kilohertz iteration rates with high accuracy of spectral line shapes on embedded systems. Precomputation of multi-dimensional polynomials and effective pressures allows high accuracy approximation of HITRAN Hartmann-Tran absorption coefficients with full matrix broadening and low residual error, replacing computationally slow line-by-line simulations and inaccurate fixed-basis functions. A novel Levenberg--Marquardt solver augmented with Taguchi orthogonal array exploration accelerates fitting convergence and escapes local minima. Together, these methods enable simulated implementations that fit 1000-point spectra with simultaneous etalon fringe, temperature, and pressure estimation at 1~kHz using 0.8~W simulated computational power for a single laser system. This work further simulates a simultaneous dual-laser single-detector architecture exploiting co-prime spectral lengths, fitting six concentration parameters at 0.82~kHz using 1.6~W simulated computational power. Mid-infrared TDLAS simulations targeting CH4, H2O, C2H6, CO2, and N2O provide an approach for possible atmospheric sensing deployments and chemical kinetics studies.