Research On Predictive Control Strategy Of Autonomous Underwater Vehicle Near Surface

The research for underwater vehicles has been paid more and more attention with the development of marine resource and increasing need for military applications.Underwater vehicle can survey,explore and collect various marine resources,such as marine hydrological information,geographic information,marine life,seabed mineral and energy etc.So there are a variety of underwater vehicles which have been widely used in different occasions.Among them,AUV is one of the focuses of current research.For underactuated AUVs,although the cost of production and the size of AUV are reduced,the reduction of drive device also brings a series of underactuated system problem.At the same time,AUV has strong nonlinear dynamic characteristics under the constraint condition and it is susceptible to wave disturbance while sailing near surface.This is depend on the control system performance.Therefore,it is very important to study the predictive control strategy of the nonlinear system with disturbance.On this basis,it is also practical to study observer-based predictive control strategy.The main contents of this paper are as follows.Firstly,the kinematics and dynamics equations are described for AUV with the six degrees of freedom and the motion models in vertical plane and horizontal plane are given as the basis for subsequent chaters.Then by the analysis of wave force near surface,we can model the wave disturbance,and give the simulation results while AUV sailing near surface.Then,considering the AUV linearization model,an observer-based predictive control algorithm is proposed,and the discrete-time full order state observer and reduced order state observer are designed.In order to reduce the conservativeness of the design process,two improved algorithms are proposed to improve the performance of state estimation.For the controller design,a predictive control algorithm based on Laguerre function is proposed,and the state estimation information is applied to the optimization process of predictive control.Next,the research on the observer-based predictive control is extended to the cases of discrete-time nonlinear systems.By using alfa-exponential stability and H infinity performance index in FOSO design process,and using regional pole constraint and H infinity performance index in ROSO design process,the transient performance is improved and the disturbance is suppressed.Furthermore,an ROSO-based model predictive control based on model online linearization(MPC-MOL)is proposed,and the validity of proposed method is verified by simulation.Considering the energy consumption of AUV attitute control in vertical plane,a energy optimal method is proposed to improve the traditional generalized predictive control(GPC)strategy.Hildreth's quadratic programming(QP)method is used to handle the contraints and get the optimal control action.In particular,the methods for longitudinal velocity and acceleration fluctuations suppression are proposed to improve the existing method.Finally,the AUV-propeller-based electric propulsion system is studied.The double closedloop control system of brushless DC motor(BLDCM)is constructed.Considering the characteristics of the propeller dynamic load,the GPC is designed based on the speed-current and current-voltage CARIMA model.And the traditional GPC is extended with simple constraints in short control horizon.Then a finite control set based model predictive control(FCS-MPC)is proposed to suppress the commutation torque ripple.This method is a kind of current control strategy and simplifies the optimization process of predictive control and effectively reduces the commutation torque ripple of BLDCM.