USING RANDOM FOREST FOR THE DEVELOPMENT OF PREDICTIVE MODEL FOR MILD STEEL IN EKPOMA

Abstract

In Nigeria buried steel infrastructure suffers premature failure due to soil-induced corrosion, yet existing degradation models rely on idealized laboratory simulations that ignore real-world soil heterogeneity and welding parameter interactions. This study addresses this critical gap by investigating mechanical property decay  behavior of mild steel weldments through longitudinal field exposure in Ekpoma  soil environment   with the aim of developing robust, field-validated predictive models to enhance infrastructure durability.

Field exposure tests were conducted over twelve months on 20 weldment specimens per site, with tensile strength, impact energy, and hardness evaluated post-exposure Random Forest (RF) were trained on welding parameters (current, voltage, gas flow) using 16 samples, with rigorous validation on 4 unseen field runs.

Random Forest demonstrated exceptional predictive capability across all sites and responses, achieving test-set R² values of  ( tensile), 0.821 (impact), and 0.882 ( hardness).  These results establish RF as a viable tool for field-based corrosion forecasting, proving that integration of real-world exposure data with ensemble learning is essential for accurate durability prediction and proactive infrastructure management in heterogeneous soil environments.