Abstract: There are two key problems affecting the intelligent operation of a longwall mining face,which are the mining equipment self-adaptive cutting in a coal seam,and the straightness of equipment in the process of continuous mining. In order to solve these two problems,the shearer accurate position information in longwall mining face must be obtained in real time. By analyzing the technical principle and hardware framework of shearer positioning system,it is found that the development of an error-reduction algorithm is the key to guarantee the shearer long time positioning accuracy in underground coal mine without GPS. According to shearer positioning principle,the positioning error is mainly caused by installation deviation and random drift error of inertial navigation system. The installation deviation is a deterministic error,and the calibration algorithm based on two-point method can reduce the positioning error by 99.12%. Aiming at the non holonomic constraints of shearer motion state,the closing path optimal estimation algorithm and dynamic zero velocity update algorithm based on the shearer kinematics model reduce the positioning error by 50% and 30%,respectively. The information filter model is used to integrate the closing path optimal estimation algorithm and dynamic zero velocity update algorithm,and the drift error of heading angle is suppressed. The automatic migration approach of UWB base station group realizes the shearer positioning at the end of the coal mine working face. The VB-UKF algorithm is implemented to smooth the time-varying measurement noise in the positioning process of the shearer,ameliorating the smoothness of the motion trajectory,which makes the localization trajectory of IMU/UWB tight fusion more accurate and provides the calibration benchmark for the inertial navigation system. The trajectory detection method of scraper conveyor based on shearer positioning trajectory realizes the online shape monitoring of scraper conveyor,and provides the theoretical basis and experiment data for automation curvature detection and face alignment in longwall mining face.