Abstract. The fabrication of microbeads at submillimter-scale has been focused on as the scientific research topic since they have been widely applied in various kinds of industries, such as pharmacology, cosmetics as well as food industry. The development of applied mathematics, material technology and information technology accelerated the submillimeter-scale analysis on the generation behaviors for the hydrogel microbeads and afterwards the manipulated its fabrication. In this work, the experiments were designed to manipulate hydrogel microbeads. The statistical analysis was carried out, the nonlinear modeling on the variation of microbeads was undertaken and its improved version was created. Submillimeter analysis-based machine learning technology was finally introduced to test and verified the neural network modelling stisfactorily.