改进牛顿–拉夫逊电阻层析成像图像重建算法

为了满足电阻层析成像图像重建算法应用于两相流领域的精度与实时性要求,提出一种改进牛顿–拉夫逊算法。针对牛顿–拉夫逊算法局部收敛的特点,将线性反投影算法成像结果作为新算法初始电阻率分布估计值,并利用基于改进粒子群算法的平衡法对海森矩阵进行预处理,降低条件数。同时在新算法迭代过程中,当振荡次数超过设定值时,采取自动更新灵敏度矩阵的策略。仿真与实验结果表明,与线性反投影算法、兰德韦伯算法及简化牛顿–拉夫逊算法相比,新算法降低了图像重建误差,有效提高了图像重建质量;而与标准牛顿–拉夫逊算法相比,新算法在满足成像精度前提下,避免了每次迭代均需更新灵敏度矩阵的缺点,提高了算法收敛速度。

Improved Newton-Raphson Algorithm for Electrical Resistance Tomography Image Reconstruction

Aiming to make image reconstruction algorithm for electrical resistance tomography meet the acquirement of measurement for two-phase flow in terms of precision and real-time,an improved Newton-Raphson algorithm was proposed.To solve the problem of local convergence,the reconstruction result of linear back projection algorithm was utilized as the initial estimation of resistivity distribution in the new method;also,balanced method based on improved particle swarm optimization was applied to pretreat the Hessian matrix to lower condition number.Meanwhile,during the iteration of the new method,the sensitivity matrix was updated automatically once the oscillation frequency exceeded the set value.Both simulation and experimental results demonstrate that the novel algorithm is effective to reduce the imaging error and then improves the imaging quality,comparing to linear back projection,Landweber and simple Newton-Raphson algorithm.In addition,comparing to standard Newton-Raphson algorithm,the improved algorithm accelerates the convergence rate by avoiding updating sensitivity matrix in each iteration without reducing precision.