Evaluation of Lateral Displacement Induced by Jet Grouting using Improved Random Forest

This study presents an improved random forest (IRF) model to evaluate the ground displacement caused by jet grouting. The proposed IRF model integrates a new hybrid particle swarm optimization-simulated annealing (PSO-SA) algorithm into random forest. The performance of the PSO-SA optimizer is investigated using a set of 8 benchmark functions; whereas, three benchmark models are used to test the IRF model. The results show that PSO-SA has much better searching and convergence abilities than its counterparts (PSO and SA). When the IRF was applied to predict ground displacement, the IRF model outperforms the benchmarks models. Moreover, the out-of-bag variable importance analysis shows that the ground lateral displacement resulting from jet grouting operations can be controlled through two operating parameters, namely, the retract rate of monitor and injected fluid pressure. The high accuracy and flexibility of the proposed model make it suitable for other engineering applications regardless of their complexity.

Automated summary

This study proposes an improved random forest (IRF) model to evaluate ground displacement caused by jet grouting. By integrating a hybrid particle swarm optimization-simulated annealing algorithm (PSO-SA) into random forest, the IRF model shows better searching and convergence abilities compared to its counterparts. The results demonstrate that the IRF model outperforms benchmark models in predicting ground displacement. Additionally, the analysis of variable importance shows that ground lateral displacement can be controlled through two operating parameters.