Research On PT Imaging Technology Combined With Data Fusion

Multiphase flows widely exist in the process of industrial production.However,due to the complexity of its work,the setting of partial flow parameters is still dependent on experience.In order to explore the movement mechanism and the law of the gas-solid two-phase flow.Electrical Capacitance Tomography(ECT)is used to study the phase distribution of gas-solid two-phase flow in closed pipelines.It has the advantages of non-invasive,low cost,simple structure and high safety.However,ECT system suffers from the problems of soft-field and rank-deficiency,the accuracy of reconstructed image is limited.In this paper,two fusion methods are proposed to improve the accuracy of ECT reconstructed image.The main work is as follows:1.On the basis of the above,to alleviate the problems of soft-field and rank-deficiency,a framework of Bayesian data fusion(BDF)is proposed.Five typical examples of two-phases flow are taken in this paper,and a 12-electrode square sensor is designed to measurement.In the framework of BDF,the ECT reconstructed image is the soft data,and the point data which coming from the preset flow patterns is the hard data.Then,to extend the influence of the accuracy point data to the entire space,the proper orthogonal decomposition method(POD)is taken to process the hard data.Further,the weighting factor of the two sources of data can be changed by users' preferences.The result shows that the proposed fusion method can make the edges of the image become clearer,especially for sharp edge which is hardly reconstructed only by the ECT technique.Besides,it is also good for separating adherent objects.2.Next,Bayesian maximum entropy(BME)method is also successfully applied to ECT image fusion.Simulation studies showed that the method has a good effect on improving the reconstructed image quality.Then,to further explore the impact of the distribution of the hard data on the fusion image quality,an optimization on the points'location is proposed.By contrast,the optimal method can get a higher fusion image quality when the number of hard data is unchanged.