Particle Swarm Optimization–Based Machine Learning Algorithms for Developing the Modified Proctor Compaction Parameter Prediction Software

Abstract

Maximum dry density (MDD) and optimum moisture content (OMC) are two significant compaction criteria, especially for quality control and design engineers. Estimating laboratory proctor compaction test is rigorous, time-consuming, and expensive, hindering projects with limited budgets and tight schedules. This study proposed the novel application of hybrid particle swarm optimization (PSO) optimized Gaussian process regression (GPR), K-nearest neighbor (KNN), random forest (RF), and extreme gradient boosting (XGB) algorithms for predicting the soil compaction parameters. Analyzing 2148 in situ soil samples from various geological locations established the maximum proficiency of the XGB algorithm followed by KNN, GPR, and RF in MDD, whereas XGB, KNN, RF, and GPR in OMC. Furthermore, the level 1 and level 2 validation results ascertain the robustness of models in predicting MDD and OMC on different geological location datasets. Eventually, the AI-based computer software developed through this study offers reliable and efficient predictions for civil engineers. © The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2023.