Marine engineering materials are prone to serious corrosion damage, which affects the efficiency and reliability of marine equipment. The diversity of corrosion morphology makes it difficult to achieve the quantification and standardization of the microscopic local information on the corroded surface, which is of great significance to reveal the multi-scale corrosion mechanism. In this paper, an image intelligent recognition method for the corrosion damage of Q420 steel in seawater is established, which is based on the gray level co-occurrence matrix, binary image method and fractal model. Through the feature extraction of corrosion morphology, the quantitative analysis of corrosion morphology and the microscopic evaluation of corrosion characteristics are achieved. The image recognition data are consistent with the electrochemical result for most cases, which confirms the validity of this image intelligent recognition method. The average gray value and energy value of corrosion morphology reduces with the Cl⁻ concentration, indicating that the corrosion damage aggravates gradually. The increasing standard deviation and entropy reflects that the randomness of the pit distribution increases. The pitting ratio increases from 20.19% to 51.64% as the Cl⁻ concentration increases from 50% to 200% of the standard solution. However, there exists a discrepancy for high Cl⁻ concentration because of the irregular corrosion morphology and various pit depth. The fractal dimension increases with the complexity of the corroded surface at low Cl⁻ concentration, but the fractal dimension decreases at high Cl⁻ concentration because the corrosion complexity is interfered by the interconnection of corrosion holes due to the accelerated pit evolution.

The inlet elbow pipe of a double-suction pump in solid–liquid two-phase flow can easily lead to a variety of unstable flow conditions, such as flow asymmetry, turbulence intensity, and vortex formation. Additionally, the inertia and aggregation effects of particle movement complicate this phenomenon further, significantly affecting energy performance and causing erosion damage. This study explores the relationship among elbow curvature, solid–liquid two-phase flow characteristics, entropy production distribution, and wear patterns in a double-suction centrifugal pump. The interaction between inlet vortex evolution and particle movements is elucidated, followed by an exploration of the spatial energy performance and erosion distribution within the double-suction pump. The results indicate that a rise in elbow curvature has a significant effect on the flow field, intensifying flow asymmetry, enhancing turbulence and vortex formation, which, in turn, leads to greater energy loss and entropy production rate, while diminishing pump efficiency. The influence of elbow curvature on energy loss is determined through the entropy production theory. In solid–liquid two-phase flow, the inertia and aggregation effects of the particles intensify wall friction and turbulent dissipation, resulting in a greater localized wear intensity. Under conditions with greater curvature, the effects of particle aggregation are enhanced, resulting in a more pronounced erosion intensity. This study offers a comprehensive insight into the pump performance in terms of the interplay of inlet vortex characteristics and the particle migration process.