<p>This study investigates the remaining bearing capacity of high-strength steel angle members with random pitting corrosion under axial compression. A parameter matrix was established to characterise the stochastic initiation and development of pitting corrosion, facilitating the construction and verification of the random pitting steel angle finite element model. A comprehensive parametric study was conducted to evaluate the influence of pit distribution patterns, corrosion rates, and pit depths on the remaining bearing capacity. Additionally, a BP neural network model was proposed to predict the remaining bearing capacity. The results indicate that under identical corrosion rates, random corrosion leads to a more significant reduction in bearing capacity compared to uniform corrosion, primarily due to localised stress concentration. Pitting corrosion can trigger an abrupt loss of bearing capacity and alter failure modes, particularly in members with width-to-thickness ratios that do not exceed the limit. The proposed BP neural network demonstrates high reliability and accuracy in predicting remaining bearing capacity, providing a practical approach for structural evaluation.</p>