Geotechnical Engineering

This study presents a hybrid AI model combining CNN and SVM to classify soil types based on image analysis, enhancing geotechnical site characterization. Using a dataset of 252 soil images, the model supports real-time decision-making via IoT-based portable devices. The hybrid AlexNet-SVM approach achieved the highest accuracy (96%), outperforming standalone CNN models. The integration of soil color recognition with geospatial data improves geological mapping and modeling, offering a valuable tool for engineers in the field. The results demonstrate that the proposed framework provides effective, accurate soil classification for geotechnical applications.

This review explores how machine learning (ML) enhances geotechnical engineering in seismic hazard assessment. ML techniques improve predictions of soil-structure interactions and foundation settlement during earthquakes, aiding real-time monitoring and risk management. The study highlights methods like aRVM, Random Forest, PSO-ANN, SVM, and hybrid models, emphasizing their role in accurate ground characterization and resilient infrastructure design. Integrating ML with traditional practices supports better decision-making and safer structures, particularly for tall buildings and fault-prone areas. The review underscores the value of hybrid algorithms and innovative foundation systems to improve predictive accuracy and address complex geotechnical challenges in seismic regions.

Integrating nanotechnology into civil engineering advances construction by improving material performance, structural monitoring, and lifecycle efficiency. This review analyzes applications like nano-enhanced concrete, self-healing systems, and embedded nanosensors, showing benefits such as up to 30% carbon emission reduction and 25% energy performance gains. It also addresses economic, workforce, and ethical considerations, alongside challenges like cost, large-scale implementation, and long-term safety. Emphasizing the need for interdisciplinary collaboration, the study highlights the intersection of engineering innovation, environmental impact, and policy to support responsible nanotechnology adoption in future civil infrastructure.

This study introduces an innovative geophysical method using non-invasive CSAMT for 2D and 3D Rock Quality Designation (RQD) estimation, overcoming limitations of traditional, costly, and time-consuming drilling. Applied to deep underground engineering, the approach provides faster, cheaper, and more accurate volumetric geomechanical data up to 1 km depth in complex rock settings. It enhances the evaluation of rock stability, supporting safer and more reliable construction, exemplified by the neutrino detector installation 700 m underground. This method improves objectivity in assessing rock integrity, reduces inconsistencies in geotechnical models, and advances the design of deep underground structures.

Examines a subway foundation pit project in Guangzhou using mechanical shaft sinking technology combined with intelligent cloud monitoring. The method improved construction efficiency by 22%, reduced costs by 20%, minimized environmental impact, and enhanced worker safety. Monitoring confirmed the method’s safety and feasibility, with stable stress–strain patterns during excavation. Recommendations include precise segment control and using advanced GPS and total station technology to improve accuracy. The approach reduced project investment by over RMB 41 million, setting a benchmark for subway excavation projects in South China and offering practical guidance for sustainable, efficient urban underground construction.