Abstract: The straightening process of the scraper conveyor in the fully mechanized mining face is affected by the detection error of sensors and the displacement error of hydraulic supports, resulting in poor straightening effect. In practical environments, the sensor surfaces used for trajectory detection on hydraulic supports and scraper conveyors face significant non Gaussian and heavy tail noise interference with unknown statistical characteristics under complex coal dust, light disturbance, and mechanical impact. Traditional filtering algorithms have low accuracy in predicting the trajectory of scraper conveyors, leading to poor straightening effect. To address these issues, a scraper conveyor straightening method based on hierarchical Bayesian modeling and variational Bayesian adaptive Kalman filtering algorithm (VBAKF) is proposed. The proposed method models process noise and observation noise as a Student-t distribution, and achieves online estimation of noise covariance through scale variables, constructing a scraper conveyor trajectory prediction model suitable for complex non Gaussian environments. On this basis, a calculation process for the sliding compensation based on predicted trajectories is further constructed to achieve the purpose of straightening, and the effectiveness of the proposed straightening method is verified by setting different distribution characteristic errors in the case of unknown statistical characteristics of noise. The results show that under the conditions of unknown noise statistical characteristics and significant heavy tail interference, the proposed method can effectively reduce the impact of detection error and displacement error on the straightness of scraper conveyor compared to existing UKF algorithm, MCKF algorithm, and AKF algorithm. Moreover, when the covariance matrix of displacement error and detection error is unknown, the mean square error value between the predicted trajectory and the real trajectory of scraper conveyor obtained by this algorithm is controlled at around 1.6mm, which is 21%, 26%, and 40% lower than the methods based on UKF algorithm, MCKF algorithm, and AKF algorithm, respectively. It can effectively improve the straightening accuracy of scraper conveyor in comprehensive mining face.