Abstract：In mixed ground, one of the most difficult problem of shield machine tunneling is severe wear and eccentric wear of cutters. This is fundamentally critical to ensure the shield machine to advance distantly in mix-face ground. The wear quantity of shield cutters used in shield tunnel section of Shenzhen metro line 9 was monitored. By analyzing the monitoring data of different types of cutters, the fitting expression of cutters wear to advancing distance was established and the longest advancement distance of the shield machine was calculated. Based on genetic algorithms optimizing BP neural network, a modeling prediction method of cutter wear was established and the predicted results are close to the measured results, which may provide theoretical basis for estimating the wear of cutters and longest advancement distance in similar stratum. The research conclusion is helpful for shield machine selection and engineering construction with similar geotechnical conditions.
