<p>In mining operations, accurate evaluation and prediction of blasting fragmentation play a pivotal role in optimizing blasting parameters, enhancing transportation efficiency, and maximizing economic returns. To this end, a dataset comprising 185 samples from the Sarcheshmeh mine was collected, and a 4-level grading evaluation system was established through fuzzy theory and multiple weighting methods. A comprehensive blasting evaluation model was developed by integrating traditional subjective and objective evaluation methods with a genetic algorithm–backpropagation (GA–BP) neural network-based weight optimization, enabling more reasonable fragmentation scoring. The model demonstrated strong predictive performance, achieving an overall prediction accuracy of 89.19% and a test set accuracy of 94.64%, indicating excellent generalization capability. Moreover, the cloud model theory was employed to characterize fragmentation tendencies, while a linear regression approach was used to construct a fragmentation prediction equation based on the comprehensive evaluation score, resulting in a coefficient of determination (R²) of 0.893. To facilitate practical application, the evaluation and prediction models were integrated into dedicated software with a simplified and user-friendly interface. This tool is intended to support pre-blasting assessment and optimization, thereby enhancing decision-making processes in mine management and mineral processing plant operations.</p>