Randomized trial predicts disc cutter wear in SPB-TBM excavation using optimization algorithms, indicating improved tunneling efficiency.
Key Points
This study aims to enhance the prediction of disc cutter wear in shield tunneling through optimization algorithms applied to a Backpropagation Neural Network.
Utilized 2,128 field data samples from the Shantou Bay Subsea Tunnel.
Employed Genetic Algorithm, Particle Swarm Optimization, and Sparrow Search Algorithm to optimize BPNN.
Conducted correlation analysis and evaluated models based on performance metrics like RMSE and R^2.
SSA-BP model achieved the highest accuracy in predicting disc cutter wear with R^2 = 0.9604.
RMSE for the model was 0.0434 mm and MAPE was 3.11%.
SSA significantly improved predictive capability compared to GA and PSO.