Research On Intelligent Optimization Algorithm In Calibration-free Wavelength Modulation Spectroscopy

Fast and accurate gas detection is crucial for environmental protection and production safety.The emergence of Wavelength Modulation Spectroscopy(WMS),a calibration-free technology,has drawn extensive attention from researchers.Compared with other gas detection methods,WMS technology has many advantages,such as not requiring regular calibration,resistance to low-frequency noise interference,immunity to light intensity fluctuations,and supporting real-time online measurement of multiple parameters such as concentration,temperature,and pressure.Therefore,this technology has become one of the hotspots in gas detection research.In calibration-free WMS,spectral fitting is commonly used for gas characteristic retrieval.This article conducts theoretical and experimental research on the spectral fitting method in this technology,mainly including:(1)Using the Levenberg-Marquardt algorithm for spectral fitting in calibration-free WMS measurement of target gas characteristics.However,there are few reports on the relevant principles and operating effects of the Levenberg-Marquardt algorithm in calibration-free WMS systems.(2)Traditional spectral fitting techniques often require precharacterization of laser tuning parameters,which undoubtedly increases the cost and complexity of experimental setup,leading to some practical limitations.On this basis,this article conducts research on acetylene as the target gas:(1)Performance analysis of spectral fitting technology based on the LevenbergMarquardt algorithm.Firstly,the simulation platform system of WMS-2f/1f calibration-free system is built based on MATLAB R2019 b software,and the operation mechanism of the spectral fitting scheme based on the Levenberg-Marquardt algorithm is analyzed.Secondly,through theoretical analysis,the main factors affecting the fitting performance mechanism of the Levenberg-Marquardt algorithm are elaborated in detail,the degree of influence of each factor is quantified,and an improvement scheme is given based on the influencing factors.Finally,an experimental platform is built,the acetylene gas absorption spectrum at 1532.83 nm is selected,the orthogonal experimental method is used for analysis,and the experimental results verify the theoretical analysis and simulation results.(2)The spectral fitting technique based on the NID-MPSO algorithm weakly depends on the pre-characterization of laser tuning parameters.This technique uses algebraic operations to optimize the objective function,reducing the fitting complexity and improving computational efficiency.At the same time,the technique also adopts a nonlinear decreasing weighted inertia weight and a mutation elite strategy to select the optimal parameters automatically.The experimental results show that compared with the spectral fitting technology based on the Levenberg-Marquardt algorithm,this technology not only does not need to consider the comprehensive characterization of laser tuning parameters but also has strong optimization ability and fitting efficiency,which effectively improves the fitting accuracy.