Q learning based under-actuated AUV virtual anchoring three-dimensional stabilization control and implementation method

The invention discloses a Q learning based under-actuated AUV virtual anchoring three-dimensional stabilization control and implementation method, and belongs to the technical field of motion controlof unmanned underwater robots. The method includes: inputting a control command and initializing all the variables; establishing a Q-learning iterative equation, and inputting all the variables; judging the convergence of Q iterations, selecting a strategy selection action if so, otherwise using virtual mooring control outputs to form the action; acting the action on an AUV, acquiring required Parameters, and updating a Q value and a Q table; performing state exchange, determining whether the AUV has reached a target state, ending the system after the number of times is satisfied, otherwise continuing iteration. The virtual mooring control part: establishes a dynamic equation of the under-actuated AUV; establishes a kinematic equation of the under-actuated AUV; establishes a mathematical model of the virtual moori