Water jet nozzle for LCD has been used as a wet cleaning process in many industries. It is necessary for the nozzle to consider cleaning effect and flux. In this paper, we applied the bubble dynamic theory(Rayleight-Plesset equation) to improve the cleaning efficiency. Generally, Rayleigh-Plesset equations for cavitation bubbles are used in analyzing computer simulation for caviting flows. Burst of bubbles causes potential energies and we can use these energies to remove organic and inorganic compounds on the LCD. Therefore, it is necessary to analyze the bubble generations and axiomatic design by computational fluid dynamics(CFD). By comparing the weight matrix of neural networks to the design matrix of axiomatic design, we propose methods to verify designs objectively. The optimal solution could be deduced by the regression analysis using the design parameters.