Abstract: Prompt Gamma Neutron Activation Analysis (PGNAA) is a nondestructive technique for elemental analysis of large-volume samples. However, the presence of neutron self-shielding effects during measurements can cause significant discrepancies between measured and true values. To address this issue, this study proposes a correction method based on an optimization algorithm to compensate for neutron self-shielding in PGNAA measurements. An experimental setup consisting of an AmBe neutron source and a high-purity germanium detector was employed to analyze boron in aqueous solutions. Standard samples were measured, and neutron self-shielding correction factors derived from MCNP simulations were used to establish a calibration curve relating boron concentration to characteristic gamma-ray counts. A neutron self-shielding correction curve for boron solutions was then computed via MCNP simulations and incorporated into a calibration model developed using the differential evolution algorithm to estimate the concentrations of unknown samples. Validation using leave-one-out cross-validation demonstrated an average relative error of 1.5%, confirming the feasibility of the proposed method for accurately determining boron concentrations in aqueous solutions of unknown composition.