Adaptive Smith predictor-based structural vibration control with time-varying delay: Adaptive model identification and hardware implementation

The presence of system time delay, model uncertainty and external excitation in structural vibration systems can lead to system instability. Therefore, a smith predictor with sliding mode controller based on extended state observer (SMCESO) is utilized to compensate for these uncertainties. However, a conventional predictor-based SMCESO (PSMCESO) fails to deal with time-varying delay. An adaptive smith predictor is designed to address the time-varying delay introduced by accelerometers, actuators, computation, and communicational network. In addition, knowing that an accurate nominal model is difficult to build in the smith predictor, an adaptive algorithm based on Lyapunov energy function, hence, is formulated to enable real-time system identification and correction. Finally, the identified parameters are incorporated into the nominal model of the adaptive smith predictor, achieving the whole integration of adaptive model identification and smith predictive control. The proposed adaptive smith predictor for time-varying delay compensation with a sliding mode controller based on an extended state observer(ASP-SMCESO) effectively ensures robust vibration suppression in systems subject to time-varying delay. An experimental platform was developed using the NI PCIe-6343 data acquisition card and the MATLAB/Simulink environment to comparatively validate the proposed ASP-SMCESO against SMCESO and PSMCESO. The vibration tests demonstrate that the control performance is enhanced in comparison with the two classical SMCESO-based approaches.