基于局部异常因子算法的智能盾构掘进与沉降相关性分析研究OA

To achieve"intelligent"and"precise"risk management during shield tunnel construction,this study investigated the correlation between excavation soil pressure and ground surface settlement.An abnormal soil-pressure analysis model for shield tunneling based on the Local Outlier Factor(LOF)algorithm was established to perform in-depth mining of soil-pressure monitoring data during shield tunneling operations.By training the LOF model to learn from a normal soil-pressure density dataset,the model effectively identified periods with abnormal soil pressure and abnormal density,and the analysis showed a significant correlation between abnormal-density soil pressure and ground surface settlement.The effectiveness of the model was validated using a case study of the Zhengzheng section of Beijing Metro Line 22.The results showed that:(1)the LOF algorithm detected anomalies by comparing the influence of data points within their local neighborhoods,and it effectively handled complex datasets containing noise and outliers,thereby verifying the feasibility of applying this algorithm to abnormal soil-pressure identification in shield tunneling operations.(2)Based on the abnormal soil-pressure analysis model for shield tunneling,accurate detection of abnormal soil-pressure parameters was achieved through research involving soil-pressure data preprocessing,LOF-based case application,model training,testing,and evaluation.Using construction data from the section between Zhengwuzhongxin Station and Zhengwuzhongxindong Station(abbreviated as"Zhengzheng section")of Line 22,the LOF algorithm identified one ring corresponding to a settlement of more than 30 mm and five rings corresponding to a settlement of 10～20 mm during periods of abnormal soil pressure.Moreover,the mean and extreme values of soil pressure,as well as the cumulative settlement,during abnormal periods were significantly lower than those in the training dataset,which confirmed the effectiveness of the model.