Industrial environments, especially in the oil and gas sector, presentunique challenges for system identification techniques. Despite advancements,there still exists a gap in our understanding of integrating black box algorithms,grey box approaches, and machine learning for simulating real-world data.With the aim of optimizing understanding and prediction in complex industrialenvironments, real-world data simulation in oil production and drilling speedanalysis was explored. This study proposes an analysis of the integration ofblack box algorithms, grey box approaches, and machine learning in simulatingreal-world data, with an emphasis on oil production and the study of the drill-rock interaction in the oil well drilling process. In this work, machine learningtechniques such as neural networks and classical system identification methods,such as linear models like ARX (AutoRegressive with eXogenous inputs) andnonlinear ones like NARX (Nonlinear AutoRegressive with eXogenous inputs),were employed to capture the dynamic behaviors of the processes understudy. Additionally, real data from oil production and drilling were utilized,considering the specific characteristics and operational challenges of theseenvironments. Based on the results obtained, the techniques used demonstratedapplicability and yielded satisfactory outcomes. Specifically, the use of hybridmodels, combining physical knowledge with multiple model approaches formedby system identification algorithms and machine learning, showed potentialfor enhancing simulation. These findings underscore the effectiveness of thesemethods, suggesting that future research could focus on implementing thistechnique in identifying complex systems.