Study on the Pressurization of Liquid Nitrogen in a Robotic Water Jet Thruster

As the process of LN 2 ’s phase transition and heat exchange with water in pressure-tight tank is complicated, the tank’s inner gas pressure is tough to be calculated by mathematical model or CFD simulation. Therefore by conducting orthogonal experimental design and its variance analysis, the first two significant influence factors of average pressurization rate of the pressurization process were determined, namely the mass of injected liquid nitrogen and the volume of compressed gas inside the pressure-tight tank. Then a RBF artificial neural network model was established, trained and tested for pressurization prediction by the orthogonal experimental data. The average value and standard deviation of relative prediction error for testing data set were 8.52% and 1.82%, respectively. Finally, the pressurization and free jumping experiments of the water jet thruster were performed. By applying the RBF neural network model, the average value and standard deviation of the relative prediction error were 8.93% and 2.33%, respectively. Thus the RBF neural network model can be taken as the theoretical pressurization prediction model and provide guidance for the robotic water jet thruster’s actual jump.