Research On Acoustic Velocity Dispersion-based Imaging Method Of Gas Concentration Field Using Acoustic CT

Gas concentration field monitoring in certain circumstances such as furnace and spreading fire in buildings is of great significance in terms of process monitoring,system efficiency enhancement and security assurance.Among all kinds of gas concentration field imaging methods,optical technology has relatively achieved a mature state,however,the manufacture of relevant industrial products,especially those essential optical components,is dominated by foreign countries.Acoustic components,compared to optical ones,possess a lower cost,easier maintenance and a longer lifespan.An acoustic sensor array is also easier to position and scale up because of its propagation property.Therefore,acoustic gas imaging technology is now attracting more and more attention.Based on the existing acoustic velocity dispersion-based gas sensing method,combined with reconstruction algorithms for solving the inverse problem,an acoustic gas concentration field computed tomography(CT)method is presented for a gas mixture within a rectangular area.We arrange a certain number of acoustic transceivers surrounding the test area,transmit and receive sound signals at 3-5 setting frequencies and collect time-of-flight data,solve the inverse problem and get sound speeds at respective frequencies.Grounded on it,we can reconstruct acoustic velocity dispersion,then identify the gas composition by the relaxation time and calculate the concentration distribution by the dispersion intensity of acoustic velocity we obtained.Compared with existing acoustic gas concentration field imaging methods,this method avoiding the troublesome measurement of acoustic attenuation,only requires acoustic velocity,which is of high precision in measurement,to accomplish the reconstruction.In addition to quantitative reconstruction,qualitative detection of gas composition for unknown mixtures can also be attained.The simulation experiments of gas mixtures CO₂-N₂ and CO₂-CH₄-N₂,of which the reconstruction average error can be as low as 1.12%,validate the feasibility and effectiveness of the proposed method.In simulations of different concentration distributions,the factors which affect the reconstruction quality are analyzed.Three representative reconstruction algorithms,namely least squares method(LSM),algebraic reconstruction technique(ART)and basis function method(MTR),are evaluated on reconstruction accuracy and anti-disturbance ability,and the parameter selection of those algorithms is also studied in this thesis.The simulation results indicate that ART outperforms LSM while MTR has the highest accuracy and robustness among the three algorithms.With back propagation neural network introduced for a description of concentration field in sufficient detail,a gas imaging method based on LSM-NN is provided.According to the simulation results,the imaging method based on neural network is superior to those based on traditional interpolation methods on both reconstruction quality and anti-disturbance ability,and it can nicely adapt to various concentration distributions as well.