Control Of An Electrohydrostatic Actuator With Tolerance To Internal Leakage

An electrohydrostatic actuator(EHA)is a self-contained and pump-controlled hydraulic system.Because of easy maintainability,simple structure,reliability,potential light weight,low oil pollution,low heat loss,and compactness,EHAs have a wide range of applications such as aircraft,vehicles,manipulators and active suspension.However,there are some issues in EHAs:(1)Uncertainties existing in the systems could degrade the performance.(2)A leaky piston seal causes internal leakage flowing across actuator chambers,that could degrade the performance.(3)With closed-loop specifications satisfied,low-bandwidth controllers are more attractive.Quantitative feedback theory(QFT),as one type of robust controller design,allows construction of linear controllers by maintaining a balance between performance specifications,controller bandwidth and plant uncertainties.In order to solve the above three problems,this thesis employs QFT to design fault-tolerant controllers.The procedure is implemented as follows:(1)This thesis presents design and experimental evaluation of a position controller for the class of EHA that is tolerant to actuator internal leakage fault and robust to various environmental stiffnesses,load masses,effective bulk modulus and viscous damping coefficients.Based on QFT,the controller is designed using physical laws to satisfy tracking and stability specifications.The controller is also augmented with a friction compensator to improve the tracking performance.The ability of the augmented controller is demonstrated in both simulations and experiments.The experimental results show that the responses are within tracking bounds despite an internal leakage flow rate as high as 8.5 L/min.In addition,the friction compensator reduces the steady-state error from 0.2 mm to 0.05 mm.(2)For designing a low-bandwidth controller,mathematical model must first be constructed either from physical laws or system identification.Physical laws may not fully define the system because of the existing uncertainties and/or difficulty to accurately model certain phenomenon.Therefore,the obtained model may be not accurate enough and the resulting controller will be too conservative(has high bandwidth).System identification techniques,on the other hand,obtain the model from measured data.Thus,system uncertainties are captured and the model will be more accurate.The bandwidth of resulting controller will be low.This thesis verifies that using system identification techniques to model EHA could result in a low-bandwidth controller.A set of off-line parametric linear identifications are implemented under different conditions,including various environmental stiffnesses and load masses.The family of identified models are used to design the QFT controller according to tracking,stability and sensitivity specifications.In addition,the performance of the controller is examined against another QFT controller,that is designed for the same system using physical laws.The simulation results show that both controllers satisfy the prescribed specifications.However,the bandwidth of the controller designed based on system identification is lower than that designed based on physical laws.The corresponding control signal is less oscillatory and less sensitive to effect of disturbances.(3)Apart from position control,there is an interest in applying EHAs to situations where actuating pressure/force control of hydraulic actuators is particularly required.Examples are automotive active suspension,deep-drawing press and molding machine.In the thesis,a low-bandwidth QFT actuating pressure controller is designed based on a system identification technique that is tolerant to actuator internal leakage.A set of off-line parametric linear identifications are implemented to obtain the mathematical model.The family of identified models are then used to design the QFT controller.The controller could satisfy tracking,stability and sensitivity specifications under different conditions,including various levels of actuator internal leakage,environmental stiffnesses and load masses.The ability of the controller is tested in simulations and experiments.The experimental results show that the system specifications are satisfied despite internal leakage up to 12 L/min.