Estimating Penetration Rate of Excavation Machine Using Geotechnical Parameters and Neural Networks in Tabriz Metro

In this study, the penetration rate of the excavation machine in tabriz metro line 2 using geotechnical parameters and neural networks is estimated. for this purpose, through comprehensive analysis, including borehole drilling, field and laboratory tests, and consideration of similar projects, the geotechnical parameters for soil and rock layers have been determined. preprocessing data techniques, such as normalization, have been applied to address challenges such as noise and bias in raw data. also, neural networks with varying architectures were evaluated using mean square error and correlation coefficient as evaluation metrics. the architecture (1-12-8) of this research demonstrates superior performance with a mean square error of 1.630 and a correlation coefficient of 0.932. this shows a strong relationship between predicted and actual penetration rate values. the findings of this research highlight the effectiveness of neural networks in estimating the penetration rate. accurate estimations of the non-linear penetration rate were achieved by employing a single-layer neural network with multiple neurons using appropriate transfer functions. overall, this research contributes to the understanding of geotechnical considerations for urban train routes and demonstrates the accuracy of neural networks for penetration rate estimation. these insights have implications for the design and engineering of similar projects.

estimating penetration rate of excavation machine using geotechnical parameters and neural networks in tabriz metro

in this study, the penetration rate of the excavation machine in tabriz metro line 2 using geotechnical parameters and neural networks is estimated. for this purpose, through comprehensive analysis, including borehole drilling, field and laboratory tests, and consideration of similar projects, the geotechnical parameters for soil and rock layers have been determined. preprocessing data techniques, such as normalization, have been applied to address challenges such as noise and bias in raw data. also, neural networks with varying architectures were evaluated using mean square error and correlation coefficient as evaluation metrics. the architecture (1-12-8) of this research demonstrates superior performance with a mean square error of 1.630 and a correlation coefficient of 0.932. this shows a strong relationship between predicted and actual penetration rate values. the findings of this research highlight the effectiveness of neural networks in estimating the penetration rate. accurate estimations of the non-linear penetration rate were achieved by employing a single-layer neural network with multiple neurons using appropriate transfer functions. overall, this research contributes to the understanding of geotechnical considerations for urban train routes and demonstrates the accuracy of neural networks for penetration rate estimation. these insights have implications for the design and engineering of similar projects.