The indoor environment has the characteristics of complexity and variability. In order to achieve accurate indoor positioning, it is necessary to model the indoor environment to estimate the target's positioning. Aiming at the problem of low positioning accuracy of the existing indoor positioning algorithms, this paper adopts the lognormal distribution model according to the characteristics of the indoor environment, and uses Taylor formula to improve the traditional least squares algorithm to achieve more accurate indoor target positioning. In this algorithm, the average Kalman filter (Kalman Filter) is first used to model the indoor environment to eliminate the interference of random noise. Then the estimated coordinates obtained by the least squares algorithm are expanded with Taylor's formula to construct a loop iteration to gradually approach the real target position. The experimental results show that the improved algorithm better improves the positioning accuracy, obtains a more accurate estimated coordinate position, and ensures the stability of the positioning deviation.