Research On Boundary Compensation Algorithm For Electrical Capacitance Tomography System Based On GAACO Algorithm

Electrical Capacitance Tomography (ECT) is a new computer assisted tomography technology, which is mainly applied to the detection of the multiphase flow in the industrial pipeline. It also has many merits, such as simple system structure, high dependability with excellent security and non-invasive in particular, which makes a broader space of application of this technology. While images reconstruction is a kind of solving inverse problem, the electrode measures the capacitance between themselves outside the pipe, namely observational data. After that we use the image-reconstruction algorithm to rebuild the gray scale images of the distribution of the elements within a certain section in different phase flows. Because of the limitation of the electrode’s quantity, the capacitance is limited as well. Compared with the number of the required capacitance, that’s not nearly enough. Employing so few capacitance values to back calculated the distribution of different phase flows in the pipeline is a kind of ill-posed problem. The result we get is not the true solution. Therefore in order to make the gray scale images we reconstruct become closer to the real images in an actual production environment, we need to compensate the edge of the gray scale images of reconstruction. The main research contents are as follows:1. This paper introduces the system composition and basic operation principle of electrical capacitance tomography in detail and analyzes the ill-posedness of the inverse problem in ECT. Aiming at this issue, we propose several common optimization algorithms.2. Pointing at the ill-posedness of ECT with the elaboration of the progressive optimal-order theory, we can prove that it is necessary to compensate the edge of the gray scale images of reconstruction. Meanwhile, we utilize convex hull theory to determine the scope of the edge, which can demonstrate the feasibility of edge compensation. On this basis, we propose an ECT boundary gray compensation algorithm based on the GAACO. After that we analyze and compare the results of using edge compensation and no edge compensation based on the GAACO, we find out that edge compensation could improve the stability and quality of the reconstructive picture. Furthermore, we come up with a preferable algorithm prior to PSO algorithm, which has advantages after edge compensation.3. Concerning the flow pattern identification problem of the inverse problem in ECT, after we analyze several kind of characteristic parameters, with the comparison of BP neural network applied on pattern recognition technique, this thesis put forward to apply the revised weighted gauss Newton neural network (RW-GN) to ECT flow pattern’s identification.