Self-tuning electromagnetic vibration systems

This thesis presents the efforts made toward making industrial mechanical vibration systems smarter. This objective is accomplished in two steps. The first step is realization of mechanical vibration actuators that can mimic the behavior of mechanical dampers and springs with variable and controllable damping and stiffness. The second step includes the design and implementation of algorithms that can find the optimum damping and stiffness in different operating conditions. First, electromagnetic actuators are selected for force generation. It is shown that creating a parallel RL circuit with variable parameters in the shunt of an electromagnetic actuator results in variable damping and stiffness behavior by the actuator. It is shown that this circuit configuration can be realized using a power electronics converter connected to a power source. Next, automatic control methods are developed for adding a self-tuning loop to the system including an electromagnetic actuator. To this end, the sliding mode extremum seeking controller was utilized to make the system self-tuning in a model-free control architecture. The concept is applied to two major problems in vibration systems: vibration energy harvesting and vibration absorption, which is also known as tuned mass damping. In the former application, single variable and multi-variable sliding mode extremum seeking controllers are used for controlling the damping and stiffness of the actuator to maximize the harvested power. In the latter case, the same controller is used with the objective of minimizing the unwanted oscillations in a host structure. Analytical methods, computer simulations, and experimental results are provided to support the proposed concept and verify the theoretical findings. The results show that it is possible to achieve efficient, variable, and controllable damping and stiffness with an electromagnetic actuator comprised of a brushless DC motor and a mechanical motion conversion mechanism. It was also shown that the proposed extremum seeking controllers successfully tune the variables toward the optimum points.