Abstract: Objectives:  The anomaly along-track stripes which often appear in the multi-beam sounding data have significant influence on shallow water. Aiming at the problem that the mechanism of traditional methods is not perfect and the elimination of anomalies is not complete, the causes are analyzed based on the multi-beam data processing principle. It is considered that the main reason is induced heave compensation residual caused by uncompensated induced heave or inaccurate measurement of position deviation of MRU (Motion Reference Unit).   Methods:  For this reason, a method of position deviation detection and elimination of abnormal stripes based on the sounding data of common coverage area is proposed in this paper. Firstly, homologous point pairs are found by automatic nearest plane point searching. Secondly, the relationship model between the water depth difference of the homologous point and the position deviation to be detected was constructed, and the position deviation was estimated by SVM (Support Vector Machines) regression parameter robust estimation method. Finally, a modified model is given to eliminate the influence of abnormal stripes in the sounding data.   Results:  Experimental results show that:(1) After the calibration of the proposed method, the standard deviation of the difference between the common coverage area of the round-trip lines decreased from 8.5cm to 2.6cm. (2) The mutual difference between the detection results of each section is about 1cm. (3) The proposed method performs worse on adjacent lines with 50% and 15% overlap area, with standard deviations decreasing from 6.9cm to 3.9cm and from 12.9cm to 7.6cm, respectively. (4) Using the detected relative position deviation to correct the entire measurement area, the standard deviation of the discrepancy value in the common coverage area was reduced from 0.07m to 0.03m.   Conclusions:  The proposed method can completely eliminate the influence of abnormal sounding stripes and provide a new method for detecting MRU installation deviation. However, the results are affected by the overlap area of the adjacent lines, so it is recommended to be performed on the round-trip lines whenever possible.