The inverse system control for underwater vehicle hovering based on support vector machine

The application of direct inverse system in underwater vehicle's hovering system is discussed in this paper. The dynamics model of hovering system is established based on it's peculiarity. The accurate inverse system is hard to be established, because of the uncertainties of hydrodynamic coefficients and the coupling affects of each degree of freedom. For these problems, this paper use Support Vector Machines (SVM) to identify α th-order nonlinear offline static inverse model of underwater vehicle's hovering system. The inverse system is a feed-forward controller, it is cascaded before the original system to compose a right-inverse system with the original system. And then the complex system is decoupled into two relatively independent Single Input and Single Output (SISO) pseudo-linear subsystem. Then additional feedback linear controller is designed by using developed linear system theory. The simulation results show that this method does not depend on the accurate mathematical model, high tracking accuracy, and successfully realize the depth and pitch control.