Research On Typical Methods Of Parameter Self-optimizing For Motion Control Of A DSRV Model

The deep submergence rescue vehicle (DSRV) is currently the most widely used andmost effective means of submarine rescue. Many countries that have submarines pay moreand more attention to the research and development of DSRV. The motion control system ofDSRV is the basis for the success of rescue operation. The control effect largely reflects thepros and cons of control algorithm. S surface control method is easy and applicable to thecontrol of underwater vehicles. But the control results are unsatisfactory in time-varyingconditions since the control parameters are fixed or adjusted by manual modification. Thispaper aims to research the typical methods of the parameter self-optimizing of S surfacecontroller for the motion control of DSRV.First, according to the structure, movement characteristics and actuator configuration ofthe DSRV experimental carrier, the mathematical model is established based on Newton Lawand Euler equations. It provides a simulation platform for the debugging of parameterself-optimizing of S surface controller.Secondly, the experience knowledge of manual adjustment is summarized afteranalyzing parameters of S surface controller. And the rule-based parameter self-optimizingmethod is discussed with fuzzy control theory. The simulation experiments and the tankexperiments show that the parameter optimizing method can enhance the partial adjustmentfunction of S surface controller. The control results improve significantly.Thirdly, the predictive model, feedback correction and rolling optimizing are introducedto the S surface controller by drawing on the basic idea of model predictive control. Thenstudy the model-based parameter self-optimizing method and its effectiveness is verified bysimulation tests.Finally, the two methods of parameter self-optimizing are combined reasonably afteranalyzing their advantages and disadvantages. Based on predictive model, fuzzy parameterself-optimizing method is put forward and this method utilizes the predictive model toprovide advanced system status information for fuzzy reasoning. It can not only solve thedelay problem of fuzzy reasoning input, but also avoid the large computational complexity ofrolling optimizing. The simulation results show that the control effect of this method is better than that of fuzzy S surface control.