Corrosion resistance of bearing steel for marine operating robots by Ultrasonic Surface Rolling Process and construction of BP neural network prediction model

This study established a corrosion rate prediction model for ultrasonic surface rolling processed (USRP) bearing steel specimens using a BP neural network, along with a detailed flowchart of the prediction model. The relative error between the predicted and actual values was less than 3.68%, demonstrating high accuracy and effectively reducing experimental workload. To address the failure mechanism of bearing steel in marine robotic transmission components under prolonged corrosion, GCr15 bearing steel was selected as the substrate and strengthened via USRP. The corrosion behavior of the treated material was systematically investigated in a 3.5% NaCl solution. Comparative experiments were conducted by immersing untreated and USRP-treated GCr15 specimens in NaCl solution for 15, 30, and 45 days. The microstructural evolution and corrosion characteristics were analyzed using metallographic microscopy, scanning electron microscopy (SEM), and white light interferometry. Results demonstrated that USRP significantly improved the surface integrity of GCr15 steel, including reduced surface roughness and enhanced microhardness, with a hardened layer thickness reaching 120 µm. In corrosive environments, USRP-treated specimens exhibited fewer and smaller corrosion pits compared to untreated ones, effectively delaying crack initiation and propagation. Moreover, the corrosion rates of USRP-treated samples were consistently lower than those of untreated counterparts. This study confirms that USRP effectively enhances the corrosion resistance of bearing steel in marine robotic transmission systems, providing theoretical foundations and technical support for extending the service life of marine operational robots.