Numerical feasibility study for transverse vibration control of rotating shaft with a neural network-based tracking algorithm

Rotary elements have been applied to a variety of mechanical systems such as pumped-storage hydroelectricity and nuclear power plant. Due to their vibration problems occurred by misalignment, bent, and unbalance, a sharp decline efficiency of system and malfunction can be caused and furthermore, the rotor may be damaged. In order to control the rotor vibration actively, active vibration control using the magnetic bearing and piezo actuator is being vigorously studied to improve operating conditions of rotary devices. This research accomplished numerical simulations of active vibration control for an unbalanced rotor system using the active bearing system applying piezo actuators. Overall rotor system is modeled using energy method and an active bearing model with two actuators placed in both x- and y-direction is developed using lumped parameter method. For implementing active control scheme through the active bearing system, a signal tracking algorithm based on neural network is developed and utilized to the rotor system. The active bearing system shows good performance on transverse vibration reduction for rotating systems.