Research on measurement error calibration method of TDLAS oxygen concentration detection based on CFD and SVR

Tunable Diode Laser Absorption Spectroscopy (TDLAS) enables rapid online detection of gas concentrations through the absorption of spectral lines by gas molecules in the infrared spectral region. In order to solve the problem of hysteresis and large errors of the measured values relative to the true values when TDLAS detects gases with simultaneous changes in oxygen concentration and flow rate in real-time, the CFD simulation model of the absorption cell is established by Fluent. The relationship between the measured value of TDLAS and the actual concentration change under different flow conditions is analyzed by simulation data, and the support vector machine regression calibration model is established to reduce the measurement error. It successfully reduces the average error from 5.844% to 2.899%, which is a reduction of 50.7%. The average error is further reduced to 1.308% after a secondary calibration based on different starting concentrations. The method is suitable for concentration detection in the presence of rapid changes in oxygen concentration.