| Capacitance Tomography(Electrical Capacitance Tomography Technology,ECT)is used to study area to be tested with different dielectric constant material distribution of imaging Technology,a process of ECT system sensor has the advantages of simple structure,low cost,quick,and the advantages of non invasive,after many years of many experts and scholars to research has made great breakthroughs,but due to the soft field characteristic problem,and this problem is very difficult to avoid and solve completely,Therefore,it greatly affects the imaging effect of ECT image reconstruction system.In order to improve the existing problems in this technology,it is of great significance to further study the ECT image reconstruction algorithm technology.According to the above problem,this paper on the basis of reading a large number of related literature at home and abroad,and the combination of ECT system,optimization was studied for the ECT sensor structure parameters to improve its sensitivity field is nonlinear,and on the basis of ECT image reconstruction theory,combining ideas generated against network,puts forward a new image reconstruction algorithm,this paper mainly completed the following several tasks:1.Firstly,this paper introduces the research background and significance of ECT technology and the development status of ECT system at home and abroad,classifies and theoretically analyzes several traditional image reconstruction algorithms which are widely spread nowadays,and expounds their advantages and disadvantages.Then the structure and principle of ECT system are introduced.2.On the basis of the research on the working principle of ECT system,the mathematical model of capacitance detection is derived by using the maxwell equation of electromagnetism,and then the value of detection capacitance is solved.Finally,COMSOL is used to build a 3d model of 8 plate capacitance sensor and conduct simulation research.3.Aiming at the nonlinear problem of sensitivity field in ECT system,a methodbased on quadratic approximation boundary optimization(BOBYQA)algorithm is proposed to improve the nonlinear problem of sensitivity field.BOBYQA algorithm is an optimization algorithm for complex objective function problems without calculating the derivative of the objective function.Firstly established according to the characteristics of the ECT system sensitivity field,sensitivity field is the optimization of objective function,and then use BOBYQA algorithm optimize the sensor structure parameters,finally has carried on the simulation,the simulation experimental results show that the optimized ECT system sensitivity of the indicators are improved,and the sensitivity field has dramatically improved nonlinear problem.4.To solve the problem of low accuracy of ECT image reconstruction algorithm,a new ECT image reconstruction algorithm based on generating antagonistic neural network was proposed.Emergent against network(Generative Adversarial Networks,GAN)is a kind of deep learning model,the model of model and the discriminant model was generated by the framework of game learning each other produce fairly good output.In this algorithm,Landweber algorithm is used to obtain a set of reconstructed images,and the reconstructed images and real images are formed into a training data set.Then this data set is used to train the generative adversary network.Finally,the test image is input into the network to get the reconstructed image.5.Finally,the simulation experiment through the simulation experiment results can be seen that the reconstruction of the algorithm in this paper the results compared with the traditional algorithm,image error is smaller,the correlation coefficient is bigger,it shows that the image reconstruction result accuracy is higher,it also proves that based on the generated against the validity of the neural network for image reconstruction of ECT system,as well as the future ECT image reconstruction algorithm of this paper proposes a new way of improvement. |
PDF Full Text Request