Explainable AI-Driven Multi-Horizon ROP Forecasting for Risk-Aware Drilling Decision Support

This repository contains the reproducibility code associated with the manuscript “Explainable AI-Driven Multi-Horizon ROP Forecasting for Risk-Aware Drilling Decision Support.” The archive includes the data preprocessing pipeline, baseline implementations, RMC-Transformer training code, repeated-run experiments, statistical analysis, risk-parameter sensitivity analysis, and the external validation workflow on the public Volve/USROP benchmark. The public USROP dataset used for external validation is available at:github.com is described in:doi.org original Volve field data are available from Equinor at:equinor.com The Oman-6ROP dataset is not included in this repository. It is derived from proprietary industrial drilling data and cannot be publicly redistributed because of intellectual-property ownership, confidentiality, privacy, contractual, and data-use restrictions. Access, if considered, remains subject to permission from the industrial data owner and the applicable confidentiality and data-use agreements. The repository therefore provides the complete software workflow and configuration files, while the restricted Oman industrial data must be supplied separately by authorized users.