Abstract:The paper studies and creates a fast iterative inversion algorithm for multi-array induction log data in horizontal layered medium,which fast and effectively reconstructs the original resistivity of formation,resistivity in flushed zone,invasion radius and depth of interface.First,using perturbation principle and forward modeling technique of semi-analytic solution,we build up fast algorithm of Fréchet derivative matrix,which includes partial derivatives of log response with respect to all model parameters and determine general relation between relative perturbation in log data and relative change of model vectors through normalization processing,overeoming the effects of physical-dimensional difference of model parameters on inversion algorithm.Then,the fast iterative inversion algorithm has been created on that basis,that is full use of the difference between vertical resolution of the coil system and surveying depth,we divided the inversion procedure into two parts:first,1-D iterative inversion has been done for apparent conductivity of short coil spacing and long coil spacing respectively in order to get the model vector more closing to the time model;then,2-D iterative inversion has been done for whole multi-array induction log data,fulfilling optimum fitting between theoretical modeling data and total log data.The effectness and anti-interference ability of the algorithm have been proved by numerical inversion results.Finally,the paper showed the effectiveness and the anti-interference ability of the method through numeric inversion results.