Search

Aiming to online suppress the mechanical resonance existed generally in electromechanical servo system, this paper proposed an effective method based on extremum seeking control (ESC). Firstly, the theoretical analysis of 2-mass model with elastic load is carried out to obtain the mathematical expression and physical phenomenon of mechanical resonance. Then, a single variable ESC-based scheme is employed to identify the resonance frequency by introducing a band-pass filter with variable center frequency. Furthermore, based on the identified resonance frequency, the multivariable ESC is applied to design a parameter self-adjusting notch filter to suppress the resonance quickly online. In addition, an adaptive adjusting law of disturbance signal amplitude of ESC is designed to accelerate the convergence as well as eliminate the unexpected steady state oscillation existing in those parameters to be optimized. Finally, the effectiveness and superiority of this method are verified sufficiently by simulations.

To increase the suppression rate of surplus torque of load simulator, this paper proposes a high-performance model-based extended state observer(ESO) to estimate the total disturbance and compensate the system as the nominal model of actual system. First the third-order model-based ESO is designed for a class of second-order systems in which the model information is used in ESO structure to improve performance. And a parameter tuning method with better noise suppression ability and observation performance, instead of the classic pole-placement method, is used to obtain accurate disturbance observation, which can effectively enhance the surplus torque suppression ability of the load simulator. Furthermore, experiments on actual load simulator are conducted to verify the effectiveness and superiority of this method.

The load change of flight simulation turntable often leads to the deviation of its control performance due to the poor adaptability of the traditional fixed-parameter controller for the varying model parameters. Aiming to identify the model parameters of flight simulation turntable precisely and rapidly and lay a solid foundation for the realization of adaptive controller with constantly high dynamic performance regardless of load alternation, this paper proposes an effective method based on dual extended Kalman filter (DEKF) to identify model parameters of turntable, in which, two low-order EKFs instead of one high-order EKF are used to reduce the computational complexity and improve the identification accuracy. And a parameter weighing approach is designed to overcome the enlarged error caused by differences in magnitude order among the parameters. Furthermore, simulations and experiments are conducted to verify the effectiveness and superiority of this method.

Aiming to increase the convergence speed of perturbation based extremum seeking control, this paper proposed a novel gradient-based perturbation extremum seeking control (GPESC) scheme by combining sine perturbation signal and input-output gradient. In this scheme, the demodulated output signal in classic extremum seeking control (CESC) is replaced with filtered input-output gradient signal to decrease the fluctuation of convergence, which can speed up the movement toward extremum dramatically. Strict stability analysis of GPESC is presented by using averaging method, and parameter selection principles are discussed. Moreover the effectiveness and superiority of the proposed scheme are verified with simulations.

Aiming to improve the frequency bandwidth of angular vibration table, this paper proposes a novel bandwidth extension scheme based on extremum seeking control (ESC). In this scheme, two perturbation extremum seeking controllers are employed to adjust the phase and amplitude of the sine input signal based on the extracted envelope of tracking error as the cost function of ESC to be minimized, for the purpose of compensating phase lag and amplitude decay of the system output. Furthermore, an adaptive amplitude adjusting method is designed with nonlinear transformation to accelerate the convergence as well as eliminate the unexpected steady-state oscillation existing in system output, which is caused by the perturbation signal of ESC. Consequently, the frequency bandwidth of the closed-loop system is extended greatly with the proposed scheme, which is verified sufficiently with simulations.