ABSTRACT Investigation of constitutive models is crucial for understanding the creep behaviour of soft rock and mitigating large deformation disasters. However, the presence of bedding layers in rock masses complicates the determination of creep parameters, increasing testing costs. Therefore, this study aims to conduct parameter optimisation analysis using the particle swarm method to refine creep parameters for soft rock with varying dip angles and analyse their effects. The framework and methodology of the model are presented, where the classical Burgers model was chosen as the constitutive model and the creep parameters to be optimised were determined. The dataset records the creep behaviours of soft rock with different dip angles prepared through laboratory experiments on 3D printing specimens. The proposed model is applied to this dataset to determine optimal creep parameters. Results indicate that transient and steady-state creep parameters increase proportionally with the dip angle, while the viscosity coefficient decreases. The optimised model demonstrates improved descriptive capabilities, contributing to understanding the creep properties of challenging geological formations and guiding on-site reinforcement measures.