Hybrid extreme learning machine for real-time rate of penetration prediction
Abstract This study presents a comparative analysis of hybrid Extreme Learning Machine (ELM) models optimized with metaheuristic algorithms Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Whale Optimization Algorithm (WOA), and Grey Wolf Optimizer (GWO) for real-time Rate of Penetration (...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2025-08-01
|
| Series: | Journal of Petroleum Exploration and Production Technology |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s13202-025-02048-x |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1849238083295248384 |
|---|---|
| author | Abdelhamid Kenioua Omar Djebili Ammar Touati Brahim |
| author_facet | Abdelhamid Kenioua Omar Djebili Ammar Touati Brahim |
| author_sort | Abdelhamid Kenioua |
| collection | DOAJ |
| description | Abstract This study presents a comparative analysis of hybrid Extreme Learning Machine (ELM) models optimized with metaheuristic algorithms Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Whale Optimization Algorithm (WOA), and Grey Wolf Optimizer (GWO) for real-time Rate of Penetration (ROP) prediction in drilling operations. The study aims to improve ROP prediction accuracy and computational efficiency while addressing the challenges of real-time implementation in dynamic drilling environments. The methodology involves a formation-specific modelling approach, where separate ELM models are trained for each formation using surface drilling parameters such as weight on bit (WOB), rotary speed (RPM), and flow rate. Metaheuristic algorithms optimize the ELM’s weights and biases to improve predictive performance. The models were trained and validated using a dataset of 13,262 data points from an Algerian field, with different statistical metrics used for evaluation. Sensitivity analysis using SHapley Additive exPlanations (SHAP) identified drilling torque and standpipe pressure as key ROP influencers. Results indicate that all hybrid models outperform standalone ELM, with ELM-GWO achieving the highest accuracy and fastest convergence. The real-time modelling framework, incorporating incremental learning and formation-based model recalibration, which ensures robust and adaptive ROP predictions and minimizing reliance on difficult-to-measure parameters such as unconfined compressive strength (UCS). This work contributes to the existing body of literature by introducing a formation-specific and real-time hybrid ELM modelling approach, demonstrating its potential for ROP optimization and cost reduction in petroleum drilling. |
| format | Article |
| id | doaj-art-4b744d648d70434190f3cc71816bd1f5 |
| institution | Kabale University |
| issn | 2190-0558 2190-0566 |
| language | English |
| publishDate | 2025-08-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | Journal of Petroleum Exploration and Production Technology |
| spelling | doaj-art-4b744d648d70434190f3cc71816bd1f52025-08-20T04:01:46ZengSpringerOpenJournal of Petroleum Exploration and Production Technology2190-05582190-05662025-08-0115912110.1007/s13202-025-02048-xHybrid extreme learning machine for real-time rate of penetration predictionAbdelhamid Kenioua0Omar Djebili1Ammar Touati Brahim2Laboratory of Energy, Mechanics and Engineering (LEMI), Faculty of Technology, University M’hamed Bougara of BoumerdesLaboratory of Energy, Mechanics and Engineering (LEMI), Faculty of Technology, University M’hamed Bougara of BoumerdesAmmar Touati Brahim, University of El OuedAbstract This study presents a comparative analysis of hybrid Extreme Learning Machine (ELM) models optimized with metaheuristic algorithms Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Whale Optimization Algorithm (WOA), and Grey Wolf Optimizer (GWO) for real-time Rate of Penetration (ROP) prediction in drilling operations. The study aims to improve ROP prediction accuracy and computational efficiency while addressing the challenges of real-time implementation in dynamic drilling environments. The methodology involves a formation-specific modelling approach, where separate ELM models are trained for each formation using surface drilling parameters such as weight on bit (WOB), rotary speed (RPM), and flow rate. Metaheuristic algorithms optimize the ELM’s weights and biases to improve predictive performance. The models were trained and validated using a dataset of 13,262 data points from an Algerian field, with different statistical metrics used for evaluation. Sensitivity analysis using SHapley Additive exPlanations (SHAP) identified drilling torque and standpipe pressure as key ROP influencers. Results indicate that all hybrid models outperform standalone ELM, with ELM-GWO achieving the highest accuracy and fastest convergence. The real-time modelling framework, incorporating incremental learning and formation-based model recalibration, which ensures robust and adaptive ROP predictions and minimizing reliance on difficult-to-measure parameters such as unconfined compressive strength (UCS). This work contributes to the existing body of literature by introducing a formation-specific and real-time hybrid ELM modelling approach, demonstrating its potential for ROP optimization and cost reduction in petroleum drilling.https://doi.org/10.1007/s13202-025-02048-xRate of penetrationDrilling optimizationExtreme learning machineMetaheuristic algorithmReal time |
| spellingShingle | Abdelhamid Kenioua Omar Djebili Ammar Touati Brahim Hybrid extreme learning machine for real-time rate of penetration prediction Journal of Petroleum Exploration and Production Technology Rate of penetration Drilling optimization Extreme learning machine Metaheuristic algorithm Real time |
| title | Hybrid extreme learning machine for real-time rate of penetration prediction |
| title_full | Hybrid extreme learning machine for real-time rate of penetration prediction |
| title_fullStr | Hybrid extreme learning machine for real-time rate of penetration prediction |
| title_full_unstemmed | Hybrid extreme learning machine for real-time rate of penetration prediction |
| title_short | Hybrid extreme learning machine for real-time rate of penetration prediction |
| title_sort | hybrid extreme learning machine for real time rate of penetration prediction |
| topic | Rate of penetration Drilling optimization Extreme learning machine Metaheuristic algorithm Real time |
| url | https://doi.org/10.1007/s13202-025-02048-x |
| work_keys_str_mv | AT abdelhamidkenioua hybridextremelearningmachineforrealtimerateofpenetrationprediction AT omardjebili hybridextremelearningmachineforrealtimerateofpenetrationprediction AT ammartouatibrahim hybridextremelearningmachineforrealtimerateofpenetrationprediction |