Adaptive discrete feedforward controller for tracking error compensation of servo‐hydraulic actuators in real‐time hybrid simulation

Abstract

Summary Servo‐hydraulic actuators are typically employed in real‐time hybrid simulation (RTHS) to impose dynamic loads on a specimen. The success and reliability of an RTHS depend on the accurate tracking of the command displacements by the actuators. This paper introduces a new adaptive compensation approach, adaptive discrete feedforward (ADF) controller, based on the principles of model‐based feedforward and feedback (MFF) controller. While the MFF controller relies on an accurate model of the controlled system obtained through system identification, the ADF controller updates its parameters to adapt to the changing dynamics of the controlled system during the test. This paper presents an experimental parametric investigation that determines the parameters required for implementing the ADF controller, such as the buffer size and the controller updating frequency. The performance of the ADF controller was compared with two other compensation methods: MFF controller and adaptive time series compensator in a series of predefined displacement tracking tests and RTHSs considering both linear and nonlinear specimens. The results are discussed in detail, where the ADF controller was shown to considerably improve the capability of the servo‐hydraulic actuator in tracking the command displacements.

Multi‐rate real‐time hybrid simulation with adaptive discrete feedforward controller‐based compensation strategy

“…In this study, the time-varying polynomial interpolator is extracted from AMRI for its superior interpolation accuracy while being paired with other more advanced outer-loop controllers for enhanced performance. Specifically, the authors' previously proposed ADF controller 21 is selected as the outer-loop controller to mitigate both amplitude and delay errors in mrRTHS in this research. The time-varying polynomial interpolator in the AMRI is built upon a set of orthonormal bases with Chebyshev polynomials of the first kind.…”

“…20 However, neither of them considers the variation of the communication delay and the effects of computational delay. This study applies the ADF controller 21 previously developed by the authors to compensate for all the actuator tracking errors, computational delay, and communication delay in mrRTHS. Unlike the conventional compensation methods designed with predefined parameters, the ADF controller adapts to the controlled plant by updating its control parameters during the mrRTHS.…”

“…More details on the formulation and implementation of the ADF controller can be found in. 21 Previous analyses revealed that the communication delay is caused by the buffer that transfers the interpolated displacement from the computational loop to the experimental loop, while the computational delay is attributed to the asynchronous reading and writing during the implementation of mrRTHS. This study has grouped the effects of the interpolator, buffer, writing, and reading in mrRTHS with the actuator dynamics and treated them as an equivalent plant.…”

“…The detailed design guidelines for both the ADF controller and ATS compensator can be found in Ref. 21 Three indices are used to quantify the time lag, global error, and local error between the predefined displacement r and measured displacement y in the predefined displacement tracking tests 46 :…”

Robust adaptive model‐based compensator for the real‐time hybrid simulation benchmark

“…Other approaches adjust the control parameters using a frequency domain analysis such as Tao and Mercan 21 or Xu et al 22 Some techniques consider discrete models of the control plant 23–25 . Similarly, Tao and Mercan 26 utilize model‐based compensation with adaptation in discrete form using a least‐squares approach. Moreover, nonlinear control combined with adaptive control such as sliding mode control, 22 self‐tunning regulator, 27 and backstepping 28 have been proposed.…”