Image Reconstruction for ECT System Based on Least Squares Support Vector Machine and Particle Swarm Optimization Algorithm

Xia Chen, Hong­li Hu*, Fei Liu and XiangXiang Gao

State Key Laboratory of Power Equipment and Electrical Insulation Xi’an Jiaotong University, Xi’an710049, China

ABSTRACT

The task of image reconstruction for electrical capacitance tomography (ECT) system is to determine the permittivity distribution and hence phase distribution inside a pipeline by measuring the electrical capacitances between sets of electrodes placed around its periphery. In view of the non­linear relationship between the permittivity distribution and capacitances and the limited number of independent capacitance measurements, image reconstruction for ECT is a nonlinear and ill­posed inverse problem. To solve this problem, a new image reconstruction method for ECT is presented based on support vector machine (SVM) combined with particle swarm optimization (PSO) algorithm. SVM is regarded as the best special small sample theory which can avoid the issues, such as network structure difficult­determination, over­learning, less­learning, and local minimum, appeared in artificial neural network methods. However, SVM performs differently with different parameters. As a relatively new population­based evolutionary optimization technique to select the appropriate parameters quickly and effectively, PSO has advantages of global optimization, rapid convergence. Experimental results indicate that the algorithm has good generalization ability and high image reconstruction quality.

Keywords ECT; Image reconstruction; Support vector machine; Particle swarm optimization algorithm