Research On Reconstruction Algorithms And Compressed Sensing Of ECT/CT Dual Modality System For Multiphase Flow

Process tomography (PT), which has the advantages of being non-intrusive,non-invasive and capable of acquiring2D/3D image information of distributionparameters, is widely used in two-phase/multi-phase flow measurement. The electricalcapacitance tomography (ECT) and computerized tomography (CT) are two kinds ofimportant PT techniques. However, satisfactory images in terms of good quality anddistinguishability haven’t been achieved because of the ‘soft-field’ characteristic forECT and the limited number of measured data for both ECT and CT, i.e. the inverseproblem for image reconstruction is ill-posed. Furthermore, a single-mode system isnot suitable to monitor multi-phase flow because each modality has its own limitation.Because the characteristics of ECT and CT are complementary; additionally, thesensitivity fields of the two techniques are non-interfering, ECT and CT systems arefused as a dual-modality tomography system in this paper. The fusion sensorcombines the advantages of both ECT and CT; hence it expands the measurementscope of PT. The main work and results of this paper are as follows:1. Images are reconstructed by statistical methods. When the mathematicalmodels of ECT and CT are unified, the expectation maximum (EM) method is usedfor ECT reconstruction; the minimum cross entropy (MCE) method is used for CTreconstruction. Compared with the traditional methods, the statistical method canobtain non-negative solution and low image distortion. So the quality and accuracy ofreconstructed images are improved based on the statistical methods.2. The L1regularization method is used to reconstruct sparse image for ECT. Thesimulation and experimental results show that the L1regulation method can give goodreconstructed results to the regions of discontinuous and non-smooth, such as the edgeand abrupt change of original distribution.3. The projection method is introduced into the reconstruction process, so thatthe ill-condition of the reconstruction problem is relaxed, and the complexity ofcomputation is reduced. The dynamic experiment results show that the projectionmethod is suitable for on-line two-phase/multi-phase flow measurement.4. The ECT and CT sensors are fused to complement each other. The fusion ofreconstructed images and measured data are compared in both reconstruction quality and speed. Simulation and experimental results show that the image-fusion methodhas the best reconstruction result. However, the fusion of measured data not onlymanifests the advantages of dual-modality sensor, but also improves the real-timeperformance of reconstruction.5. Random sampling method based on the theory of compressed sensing is usedfor ECT/CT dual-modality system. L1method, which is a non-linear method andsuitable to reconstruct images through random sampling data, is used to solve theinverse problem. Compared with the systematic sampling method, the compressedsensing overcomes the bottlenecks of Shannon’s sampling theorem; both of cost andresponse time of the system can be reduced without affecting the resolution andstability of reconstruction. During the measurement of dynamictwo-phase/multi-phase flow, differential method is used for data collection of dualitysystem, in order to satisfy the restricted isometry property (RIP). As a result, both ofreconstruction quality and real-time performance are improved.