Research On Temperature Reconstruction Algorithm Using Acoustic Tomography Technology

Acoustic tomography(AT)is a non-destructive testing method with great development potential,which has the advantages of wide detection range,non-invasiveness and low cost.It is widely used in chemical industry temperature measurement,marine climate monitoring,agricultural production storage and other fields.This topic conducts in-depth research on probe arrangement and acoustic temperature measurement reconstruction algorithm the factors that affect the accuracy of AT reconstruction.The main tasks are as follows:1.The effects of the division of the measured area and the arrangement of the probes on the reconstruction accuracy are studied.The number of probes can be selected from 8,12 and 16.In addition,two probe arrangements,f our-sided and fourcornered,are used,and 5 different discrete area division schemes are set up.Through numerical simulation of three typical temperature distributions,the temperature reconstruction results under different distributions are obtained,by comparing the reconstruction results of different distribution methods,it is found that the results of the four-sided 16 probes distribution method are better than other distribution methods in the edge and peak shape reduction degree of the measured are a.If it is limited by the space of the tested area and other conditions,it is necessary to arrange a small number of probes,8 probes can be preferred,and the four-sided and fourcorner layout schemes can be selected according to the characteristics of the tested area.When 12 probes are used,the reconstruction quality of the quadrangular distribution is higher than that of the quadrangular distribution,so the quadrangular distribution is preferred.2.A high-resolution reconstruction algorithm of temperature field distribution based on acoustic tomography is proposed.In order to reduce the influence of the limited number of discrete regions on the reconstruction accuracy of the temperature field,the time-of-flight(TOF)data was pre-reconstructed to obtain the temperature reconstruction results in the sparse discrete regions,and then the obtained results were pre-processed smoothly,the processed data is brought into the RBF neural network output to obtain the refined temperature distribution reconst ruction result.Comparing the reconstruction results of the Tikhonov regularization-RBF(TR-RBF)algorithm with the reconstruction results of the ART algorithm,SVD decomposition method and Tikhonov regularization,the results show that the reconstruction accuracy and restoration degree of the TR-RBF algorithm are the best among the tested algorithms.On this basis,noise signals of different levels are added to the original TOF data to verify the anti-noise and stability of the algorithm.The TR-RBF algorithm can achieve higher resolution temperature distribution reconstruction,effectively improving the problem of low resolution of reconstructed images due to the limited number of discrete regions.3.An optimization algorithm for AT image reconstruction is studied.By transforming the underdetermined equation system in the AT inverse problem into a planning optimization problem,since the objective function is a vector,the elements in it will generate the multi-peak and multi-extremal multi-modal problem under the2-norm representation.The cuckoo search(CS)algorithm is used to solve the sound velocity vector in the original equation system;then the initial population of the CS algorithm is optimized to enrich the diversity of the population and improve the reconstruction accuracy.The numerical simulation results verify the feasibility and noise immunity of the algorithm.