The employment of large-diameter shield machines has increased the likelihood of encountering composite formations,posing engineering challenges associated with excessive surface settlement.To tackle this issue,this s...The employment of large-diameter shield machines has increased the likelihood of encountering composite formations,posing engineering challenges associated with excessive surface settlement.To tackle this issue,this study introduces a hybrid model which integrates the extreme learning machine(ELM)with the sparrow search algorithm(SSA)to predict longitudinal surface settlement.Based on on-site measurements.this study analyzed longitudinal surface settlement patterns across both homogeneous and composite formations.Tunneling parameters,geological parameters,and geometrical parameters were considered as input parameters.Furthermore,this study conducted a comparative analysis of the predictive performance among SSA-ELM,ELM,and SSA-back propagation(BP),with respect to coefficient of determination(R^(2)),mean absolute error(MAE),root mean square error(RMSE),and training time.Last,in anticipation of potential risks,a feasible optimization approach is provided.SSA-ELM outperforms both ELM and SSA-BP in terms of R^(2),MAE,and RMSE,with values of 0.8822,0.3357,and 0.4072,respectively.Regarding training time,SSA-ELM requires 0.2346 s,prior to SSA-BP with a value of 1.8427.Although it is not as fast as ELM,the discrepancy between SSA-ELM and ELM is only 0.1187 s.Overall,SSA-ELM demonstrates higher performance and serves as an effective tool to guide the construction process.展开更多
基金sponsored by the National Natural Science Foundation of China(Grant Nos.52178386 and 52378336)。
文摘The employment of large-diameter shield machines has increased the likelihood of encountering composite formations,posing engineering challenges associated with excessive surface settlement.To tackle this issue,this study introduces a hybrid model which integrates the extreme learning machine(ELM)with the sparrow search algorithm(SSA)to predict longitudinal surface settlement.Based on on-site measurements.this study analyzed longitudinal surface settlement patterns across both homogeneous and composite formations.Tunneling parameters,geological parameters,and geometrical parameters were considered as input parameters.Furthermore,this study conducted a comparative analysis of the predictive performance among SSA-ELM,ELM,and SSA-back propagation(BP),with respect to coefficient of determination(R^(2)),mean absolute error(MAE),root mean square error(RMSE),and training time.Last,in anticipation of potential risks,a feasible optimization approach is provided.SSA-ELM outperforms both ELM and SSA-BP in terms of R^(2),MAE,and RMSE,with values of 0.8822,0.3357,and 0.4072,respectively.Regarding training time,SSA-ELM requires 0.2346 s,prior to SSA-BP with a value of 1.8427.Although it is not as fast as ELM,the discrepancy between SSA-ELM and ELM is only 0.1187 s.Overall,SSA-ELM demonstrates higher performance and serves as an effective tool to guide the construction process.
