Research On Acoustic Tomography Based Temperature Distribution Reconstruction

Obtaining the accurate temperature distribution is of great significance for ensuring the high-efficiency and stable operation of industrial systems,reducing pollutant emission,etc.The acoustic tomography(AT)is considered to be a promosing temperature distribution measurement technology owing to the advantages of non-invasive sensing,wide measurement range,large scale space measurement,etc.This thesis mainly focuses on the time of fight(TOF)measurement method and the AT based temperature distribution reconstruction algorithm.The main contributions can be summarized as follows:(1)The acoustic temperature measurement principles are studied,common acoustic TOF measurement methods are compared,the AT reconstruction theories are analyzed,typical reconstruction algorithms are summarized,the factors that restrict the amelioration of the measurement precision are expounded,and the influences of the selection of initial solutions and noise levels on the reconstruction precision are numerically studied.(2)A multi-scale dual-threshold TOF measurement method is proposed to overcome the problems,such as zero drift of received acoustic signals,noise interference and the difficulty in determinating the voltage threshold in real-world measurements.Experiment study validate the effectiveness of the new method in improving the measurement accuracy of the TOF data.(3)A two-stage acoustic reconstruction algorithm with the grey wolf optimizer(GWO)and the extreme learning machine(ELM)is proposed to alleviate the issues,e.g.,outliers in the TOF measurement data and the lack of information.In the first phase,the measurement domain is discretized into a coarse grid to reduce the number of the unknown variables,alleviating the ill-posed nature of the AT inverse problem.The L1 norm based objective functional is constructed to mitigate the influences of outliers in the TOF measurement data on the reconstruction precision,which is effectively solved by the GWO algorithm to get the temperature distribution on the coarse grid.In the second phase,the measurement domain is refinedly discretized to capture the details of the temperature distribution,and the ELM method is deployed to predict the refined temperature distribution.Numerical results confirm that the proposed algorithm can ameliorate the robustness of estimation,and the reconstruction accuracy is superior to common reconstruction methods.(4)A reconstruction algorithm integrating the advantages of the Tikhonov regularization method and the least squares support vector machine(LSSVM)is put forth to consider the inaccurate properties of the TOF measurement data and the reconstruction model.The objective functional is constructed to simultaneously emphasize the above inaccurate natures.The Tikhonov regularization method is employed to ensure the numerical stability.The L1 norm is used to mitigate the influences of the outliers.The alternating direction method of multipliers(ADMM)method is deployed to solve the objective functional to get the temperature distribution on the coarse grid,and the temperature distribution on the refined grid is predicted by the LSSVM method to obtain the details of the temperature distribution.Numerical simulations and experiment results indicate that the new method can guarantee the numerical stability in the AT inverse problem,and improve the reconstruction quality.(5)A reconstruction algorithm with the advantages of the Memetic algorithm and the Gaussian process regression(GPR)is proposed to increase the reconstruction precision and the possibility of obtaining the global optimal solution.The robust estimation based objective functional is constructed according to the AT measurement equation.The Memetic algorithm with the advantages of the Fireworks algorithm(FWA)and the Split Bregman(SB)method is developed to solve the objective functional to obtain the temperature distribution on the coarse grid,and the GPR method is deployed to predict the final temperature distribution on the refined grid.Numerical simulations and experiment study results validate the feasibility of the new algorithm in improving the robustness of estimation and the reconstruction accuracy.Research findings improve the acoustic temperature measurement accuracy,and make contributions for the theory research in solving the AT inverse problem and the strategy of energy saving and emission reduction in China.