Research On Generalized Approximation Based Adaptive Fault Tolerant Control Algorithms For Autonomous Underwater Vehicles Subject To Thruster Faults

Autonomous underwater vehicles(AUVs)play an important role in the survey and exploration of marine resource.AUVs operate autonomously in marine environment without any umbilical cable.Safety and reliability are considered as AUVs'important features.Thrusters are the main source to provide the forces required to control an AUV,and thrusters are also one of the most likely sources of faults.After the occurrence of thruster fault,fault tolerant control is one of effective techniques to deal with thruster faults to maintain the safety of thrusters itself and the AUV.Therefore,research on thruster fault tolerant control will play an important role in improving AUVs' safety.This dissertation will investigate thruster fault tolerant control problem based on generalized approximation from four aspects,including cooperation mode between fault diagnosis and fault tolerant control,over-adjustment of control parameters under a great initial error,region tracking fault tolerant control,fault tolerant control without velocity measurement.The detailed research works are presented as follows.Investigate cooperation mode between fault diagnosis and fault tolerant control.The control performance of active fault tolerant control methods based on control reallocation technique seriously depends on the accuracy and timeliness of fault diagnosis system.Moreover,in marine environment,due to the effect of external disturbances,it is always difficult to extract fault features,identify the source and magnitude of thruster faults.In order to buy more time used for fault diagnosis and to maintain the reliability of control reallocation based fault tolerant control,this paper investigate adaptive fault tolerant control strategies independent with fault diagnosis.Adaptive terminal sliding mode based thruster fault tolerant control is proposed for AUV in time-varying ocean currents.In the proposed controller,adaptive strategy is incorporated into terminal sliding mode to estimate on-line the upper bounds of the lumped uncertainties,including effect of ocean current,and the change of thruster distribution gain caused by thruster fault,respectively.In addition,a continuous switching term based on fractional power is developed to replace the discontinuous switching term in sliding mode controller.Finally,simulations and pool experiments are performed on AUVs to verify the effectiveness of the proposed method.Investigate over-adjustment of control parameters for AUVs under a great initial error.In single closed-loop based fault tolerant control strategies,AUV's initial tracking error is always considered as error caused by AUV itself insufficient operation ability.This would mislead adaptive fault tolerant controller to over-adjust the estimated value of the upper bound of the lumped uncertainties,resulting in quick change in terms of control output.Therefore,virtual closed-loop system based adaptive fault tolerant control strategy is proposed for AUVs subject great initial deviation.In the proposed method,a virtual closed-loop system is constructed firstly,apart from the real closed-loop system.Then the original desired trajectory is input the virtual closed-loop system and the effect of AUV's initial deviation on real control performance is separated by a fixed-parameters controller in the virtual closed-loop system.After that,the virtual closed-loop system provides a "virtual desired trajectory" without initial deviation for the AUV real system.Then,by incorporating the control output in the virtual closed-loop system and neural network based compensator,the uniformly ultimate boundedness is achieved in presence of great initial deviation.Finally,simulations are performed on AUVs to verify the effectiveness of the proposed method.Investigate adaptive region tracking fault tolerant control problem for AUVs.In some special applications,such as underwater searching within wide ranges,long-distance pipeline tracking,users pay more attention to energy consumption.In other words,they expect longer missions as long as tracking error is within the desired region,rather than high tracking precision.As for these special applications,an adaptive region tracking controller was developed for AUVs based on the concept of region potential function.However,when adopting the region tracking controller for AUV subject to ocean current or thruster fault,it is difficult for the tracking errors to be kept within the desired region.Therefore,a novel region tracking fault tolerant control method is proposed based on backstepping technique for AUVs.At first,a type of piecewise and differential Lyapunov function is constructed.Then,the paper analyses and acquires the bound structures of the lumped uncertainties and the variation of thruster distribution matrix caused by unknown thruster faults,respectively,to achieve region tracking control.In addition,an adaptive adjustment scheme for the desired trajectory is developed to deal with thruster dynamics constraints.Finally,simulations and pool experiments are performed on AUVs to verify the effectiveness of the proposed method.Investigate adaptive fault tolerant control without velocity measurement for AUVs.When faults occur in velocity sensors or the measurement results of velocity sensors are not accurate,observers are generally used to estimate AUV's velocity information.High-gain observer is a typical one.Asymptotical convergence of the estimation error,rather than finite-time convergence can be obtained from a high-gain observer,and then based on the estimated velocities,the tacking error of the fault tolerant controller is not satisfactory.Therefore,adaptive sliding mode observer based fault tolerant control is proposed for AUVs without velocity measurement.In order to guarantee the finite-time convergence of the estimation error,linear transformation is introduced to AUV's state variables firstly,and then a nonlinear feedback is constructed in the developed finite-time convergent observer by combining a linear function,signature function,integral function and fraction-order function of the position estimation error.After obtaining the estimated velocity information,adaptive fault tolerant controller is derived based on an improved integral sliding mode surface and neural network.In addition,an adaptive compensation method is developed to counteract the approximation error of neural network,so as to avoid a discontinuity term in the developed control law.Finally,simulations and pool experiments are performed on AUVs to verify the effectiveness of the proposed method.