Ultrafine hexanitrostilbene （HNS） is widely used in explosion foil initiators and related applications due to its outstanding thermal stability and excellent high-voltage short-pulse performance. However， its high surface energy during service process leads to solid-phase ripening. Previous studies have explored the effects of temperature， residual solvents， and time on the solid-phase ripening of ultrafine HNS， but these investigations primarily focused on isolated or narrowly factors. Currently， no multivariate predictive model has been established. In this study， a predictive model was developed based on previously obtained small angle X-ray scattering （SAXS） data， including specific surface area （SSA） and relative specific surface area （RSSA）， obtained under varying temperatures and residual dimethylformamide （DMF） contents. The model was constructed using machine learning algorithms and optimized empirical models. It comprehensively accounts for time， temperature， and residual DMF content in its predictions. The results show that on the training dataset， the random forest （RF） model achieved an R² of 0.9989 in predictions， while the polynomial regression （PR） model and optimized empirical model attained R² values of 0.9091 and 0.9129， respectively. By comparing the prediction performance of these three models， the most suitable model for predicting the solid-phase ripening process of ultrafine HNS was identified. Furthermore， purity tests and scanning electron microscopy （SEM） characterization revealed that particle characteristic variations exert significantly influence on the extent of solid-phase ripening in ultrafine HNS. A predictive method was established for the solid-phase ripening process of ultrafine HNS， laying a foundation for investigating its aging mechanisms and optimizing storage stability.