Image Reconstruction Algorithm for Electrical Capacitance Tomography Based on Multi-dimensional Support Vector Regression
A new method based on multi-dimensional support vector regression (MSVR) is presented to solve the ill-posed image reconstruction problem in electrical capacitance tomography (ECT). The MSVR with a hyper-spherical insensitive zone and IRWLS algorithm is firstly introduced to solve this problem. The neural networks have been reported to be applied to this kind of inverse problem. However, this method is known for serious over-fitting. MSVR has been proven to have all the advantages of neural networks, and can overcome the over-fitting problem. The proposed MSVR method in this paper is verified through typical flow patters image reconstruction. The results show that this method is an effective approach to solve image reconstruction for ECT, which is faster compared with the iterative methods and more accurate compared with the neural networks.
electrical capacitance tomography two-phase flow image reconstruction algorithm multi-dimensional support vector regression similar iterative re-weight least square
Yang xiaoguang Li Jianwei
Province-Ministry Joint Key Laboratory of Electromagnetic Field and Electrical Apparatus Reliability School of Computer Science and Software Hebei University of Technology Tianjin, China
国际会议
太原
英文
73-76
2010-10-22（万方平台首次上网日期，不代表论文的发表时间）