Dispersion quality of graphene nanoplatelets in cementitious materials with applications in geotechnical ground improvement, a machine learning based prediction model

Purpose This study aims to develop interpretable machine learning models using evolutionary polynomial regression (EPR) to predict UV–Vis absorbance of graphene nanoplatelet (GNP) dispersions in cementitious materials, enabling quality control and optimisation of nano-enhanced materials for geotechnical ground improvement applications. Design/methodology/approach EPR combines genetic algorithms with least squares regression to construct explicit polynomial models from experimental data. Six key parameters affecting GNP dispersion were investigated: loading, water content, superplasticizer dosage, sonication time, concentration and holding time. UV–Vis absorbance measurements provided training and validation data sets. Multi-objective optimisation balanced model accuracy against complexity, generating transparent mathematical expressions that reveal parameter significance and interactions governing dispersion quality. Findings EPR models achieved coefficient of determination values from 89% to 99% across parameters. GNP loading and concentration emerged as dominant factors with high accuracy, while superplasticizer showed minimal absorbance correlation despite rheological importance. The transparent five-term polynomial expressions successfully captured non-monotonic relationships for sonication time and complex ageing mechanisms for holding time. Models demonstrated strong predictive capability on both training and validation data sets, enabling systematic optimisation of dispersion protocols. Originality/value This study presents a novel application of EPR methodology to nanomaterial dispersion characterisation, overcoming black-box limitations of traditional machine learning approaches. The transparent mathematical expressions reveal underlying physical mechanisms while enabling practical quality control implementation. This work addresses a critical gap in predictive modelling for GNP-enhanced cementitious materials used in geotechnical engineering, supporting sustainable infrastructure development through optimised nano-enhanced ground improvement technologies.