---

Developed novel adaptive controller is suitable for time-varying linear system subject to harmonic reference signal with variable average and amplitude. This controller is experienced successfully at high frequencies on voice coil actuator (VCA) fatigue testing machine which has variable time-varying dynamic. In this applicable approach, assuming having linear system around operation point and slow rate time-variation, tracking control of harmonic reference signal is replaced with regulation control of average and amplitude of harmonic reference signal. In the proposed method, a Single-Input Single-Output (SISO) system estimated by a fourth-order model, is considered as the simplest decoupled Multi-Input Multi-Output (MIMO). This causes reduction in the amount of computations and no need to complicated hardware. Consequently, the proposed method provides a real-time control for implementation of random harmonic loading with rapid changes in average and amplitude. To complete control objectives, primary control by the PI controller, dynamic saturation blocks and linearization blocks, are employed. Soft start of loading due to PI controller, provides enough time for system identification of adaptive controller and guarantee avoidance of impact on specimen and credibility of fatigue test. In addition, the using of linearization blocks for trajectory planning of command signal and using of dynamic saturation blocks, restrain the overshoot of loading and prevent excitation of unmodelled dynamics. Finally, different materials under different frequency loading are successfully tested.

---

A new method was provided to control linear time-varying systems in which the reference signal is a harmonic function with variable amplitude, mean or frequency, such as block loading. The results of practical tests on the voice coil actuator fatigue test machine indicate that this method is robust to noise and disturbance. Also proposed control method compensates unknown time-varying time-delay which leads to bandwidth increase in harmonic loading. On the other hand in this paper the central pattern generator (CPG) was used to design the trajectory of input signal to the main plant. Because of soft applying of changes by CPG, it prevents excitation of high frequency dynamics. To implement the proposed method main plant, which can be estimated with a fourth-order single-input single-output (SISO) model was considered as a two-input two-output (TITO) decoupled system that each relevant input-output is first-order. The level of control loop pairs to each other is investigated by calculation of dynamic relative gain array (DRGA) matrix. The test results also show that loading control was properly carried out in the presence of various uncertainties such as nonlinearities, unmodeled dynamics and time-varying parameters.