Application of Hybrid Conjugate Gradient Algorithms in Inverse Problems of Electromagnetic Tomography

Based on the existing conjugate gradient algorithm, a new MN conjugate gradient method for inverse problems of electromagnetic tomography is proposed. The composition of the electromagnetic tomography system and the characteristics of the positive and inverse problems are described, and the solutions of the inverse problems are given. The existing typical conjugate gradient algorithms and the hybrid conjugate gradient methods using convex combinations are summarized. Based on four typical models of EMT system, PRP, HCG and new MN methods are compared in terms of image reconstruction accuracy, iteration times and iteration time. It is concluded that all three conjugate gradient methods can satisfy the termination conditions. The MN method can reduce the number of iterations and the calculation time, and has excellent numerical performance in practical calculation. (paper)