Generalized predictive control design and experiment of model-switched remotely operated vehicle

Underwater vehicles have the characteristics of complex operation environment, highly nonlinear motion and time-varying parameters. In the complex control system, the generalized predictive control has the ability of nonlinear representation as usual. In order to improve the performance of roll control of ROV, a method based on generalized predictive control was proposed to restrain the overshoot and fluctuation caused by changing too fast of roll angle in this paper. When applying linear generalized prediction algorithm, the dynamic model is firstly used to describe the nonlinear dynamic process of MC-ROV and then it is equivalent to a time-varying linear system. In order to avoid the inverse matrix of generalized predictive control optimization, a direct generalized predictive algorithm for constrained input and output is used to decrease the amount of calculation with only one constraint. At the same time, due to the poor response and robustness of generalized prediction at the initial time, the PID algorithm is introduced. The experiment results on this ROV system demonstrate the effectiveness of the proposed approach and the satisfactory performance is achieved.