Identification And Control Of Turntable Servo System

Servo systems are widely utilized in various advanced applications.As the rapid developments of science and technology,the higher accuracy of servo systems is demanded in the industry application.However,the non-linearity due to friction,unknown parameters and external disturbances in servo systems will lead to tremendous decline in terms of performance.In this thesis,both model-based and model-free approaches for non-linearity compensation are studied,which includes the identification and control of the servo systems.The model-based strategies,in which the non-linearity is compensated through online estimating the unknown parameters of the model,are carefully studied.Additionally,the advantages and disadvantages of both approaches are evaluated through comparisons.For convenience,the whole research is based on turntable servo system.The major contributions of the research are illustrated as below:(1)A RBF neural network with generalized tracking error based controller is proposed in this chapter.A model-free approach for compensation is integrating non-linearity from each aspect as a single non-linear function,which can be approximated by an RBF neural network.Then the generalized tracking error based controller could be accordingly designed,which includes three parts: generalized tracking error feedback term,robust term and RBF neural network approximates.The controller given has achieved the tracking control of the turntable servo system.The simulation results verify the effectiveness of the RBF neural network with generalized tracking error based control.(2)A multi-innovation least square algorithm with forgetting factor for parameter estimation is studied in this chapter.Higher accuracy has been achieved through corrections on estimations of parameters at the previous time stamp with multi-innovation,compared with the original least-square approach.The forgetting factor introduced to the multi-innovatio least-square approach eliminates the data saturation and enables the algorithm to estimate the time-varying parameters.Through simulation comparison,the validity of multi-innovation least-square method with forgetting factor is verified.(3)A filtering based adaptive parameter estimation with non-singular time-varying terminal sliding mode is studied in this chapter.By taking into account the error of parameter estimation through filtering technique to ensure that the estimations converge into their corresponding true values,then the accuracy of estimation has been increased.Moreover,the persistent excitation can be verified with the sign of the minimum eigenvalue by introducing the auxiliary matrix,which avoids adding disturbance and improves the efficiency of the system.Taking advantages of both time-varying sliding mode control and terminal sliding mode control,the non-singular time-invarying terminal sliding mode control is able to guaranteed convergence of the tracking errors within the limited time,eliminate the arrival stage,accelerate the convergence and avoid the singularities.The corresponding comparisons in simulations with least-square approaches have verified the advantages of the filtering based adaptive parameter estimation.(4)An optimal adaptive parameter estimation with time-varying gain based super-twisting sliding mode control is proposed in this chapter.By restricting the error of parameter estimation through a cost function with forgetting factors,an adaptive law with time-variant gain has been designed to minimize the error.Consequentially,the overshoot has been decreased and the accuracy of parameter estimation was enhanced.With forgetting factor taken into account,the weight of data at present increased in terms of the effectiveness,while that of the previous data shrank,thorough which the data saturation was eliminated and the estimations on time-variant parameter could be obtained.The design of super-twisting sliding mode control algorithm with time-variant gain is able to yield continuous control signals without jitter,which prevents system from chattering.Results of both simulations and the experimental turntable platform have verified the validity of the algorithm.