To address the issue of data anomalies in the structure health monitoring system of a dock due to sensor failures,which subsequently affects the accuracy of structural safety assessment,a fault data identification and correction method is proposed based on an improved box plot method.By integrating the improved box plot method with scatter plot analysis technique,a statistical feature recognition model is constructed that includes four types of data:no fault,spike fault,drift fault,and bias fault,achieving precise classification of fault types.In the experimental verification phase,the method exhibits a 91.07% fault recognition accuracy rate among 56 sets of actual measurement data segments.For different fault types,the research designs differentiated correction strategies:using Lagrange interpolation for spike faults,applying moving window mean fitting to correct drift and bias faults.Finally,by comparing the autocorrelation of data before and after correction,the effectiveness of data correction is verified.The research results provide reliable data quality assurance method for the dock structure health monitoring system and have significant practical significance for improving the level of safety monitoring of major infrastructure.