In order to solve the problems of high safety risk and low training efficiency in underwater operation of deep submersible vehicle. Taking China's first manned deep submarine vehicle “jiaolong” as the simulation prototype, an underwater training platform is developed, which is not limited by time and place. Using three-dimensional modeling technology to build the three-dimensional model of “jiaolong” A mathematical model of the motion of a manipulator is established to simulate the associated motion between the joints of the manipulator. The collision detection technique is used to determine whether there is interaction between the manipulator and the operated object, between the object and the sampling basket, and between the equipment and the scene. The system is based on 3D development engine, which includes cobalt-rich crust mining area, polymetallic sulphide area, shallow sea area and cold spring area and so on. With this system the operator can train the underwater operation process. The system uses virtual reality helmet as the final visual display mode and the line of sight collision detection based on helmet instead of mouse click to simulate the whole process of underwater operation of deep submersible.

We perform a virtual simulation of deep-sea hydrothermal fluids by using a combination of physical-based particles and grids. We use Vortex-In-Cell method to simulate vorticity and generate induced velocity fields, and the semi-Lagrangian method to simulate the advection and diffusion of temperature and density fields. Finally, we show the rich details of hydrothermal fluids by comparing simulations with real fluids.