基于遗传算法的组合ERT图像重建算法研究

Research on a hybrid ERT image reconstruction algorithm based on GA

针对目前电阻层析成像图像重建算法存在成像精度较低的问题,以及为了满足应用于多相流领域的精度要求,提出一种基于遗传算法的组合算法,将线性反投影算法、修正的牛顿-拉夫逊类算法与区间剖分引入遗传算法种群初始化操作中,同时为了改善单纯遗传算法局部搜索能力差与未成熟收敛的问题,将粒子群算法引入遗传算法变异操作中。实验结果表明组合算法效果明显优于线性反投影算法,修正的牛顿-拉夫逊类算法,有效克服了遗传算法早熟收敛现象,提高了成像精度。

In order to resolve the low image reconstruction quality problem of electrical resistance tomography and fulfill the measurement precision requirement for multi-phase flow,a hybrid algorithm based on genetic algorithm was proposed.The hybrid algorithm introduces linear back projection,modified Newton-Raphson method and divided interval into the population initialization of genetic algorithm,and applies particle swarm optimization to the mutation operation in order to overcome the poor local searching ability and premature convergence of simple genetic algorithm.Experimental results show that the novel algorithm is superior to linear back projection and modified Newton-Raphson method,and can overcome the premature phenomena effectively and improve the reconstruction quality.