Excessive vibration can be detrimental to a wide variety of structures and machines. Vibration isolation is an effective means to suppress vibration. This study focuses on development of some new techniques to improve performance of vibration isolators. The thesis consists of two parts. In the first part, a novel vibration isolator is developed. In the second part, an adaptive fuzzy- neural network (FNN) controller is applied to active isolators.

The developed isolator possesses the characteristics of high-static-low-dynamic stiffness (HSLDS) and can act passively or semi-actively. The HSLDS property of the proposed isolator is obtained by connecting a mechanical spring, in parallel with a negative spring which is produced by a pair of electromagnets (EM) and a permanent magnet (PM) in attracting configuration. The isolator spring is composed of three parts:​ mechanical spring, the PM spring, and the EM spring. The characterization study intends to determine the stiffness of each of these springs. Due to the highly nonlinearity of the system, an experimental approach is taken. The mechanical spring used is a steel beam that demonstrates a nonlinear hardening effect. Both the PM spring and EM spring possess negative stiffness. The PM stiffness can be varied by adjusting the gap distance between the PM and the EMS while the EM stiffness can be varied by tuning the current to the EMS. With the determined stiffness relationships, the study explores how a combination of the mechanical spring and the PM spring can reduce the hardening effect, thus the natural frequency of the isolator. The study demonstrates that adding both the PM and EM springs to the isolator can result in a quasi-zero-stiffness (QZS). A computer simulation is conducted to find the displacement transmissibility of the isolator and the jump frequencies for different settings.

A commercial software package, Comsol Multiphysis, is employed to characterize the isolator as well. First, single physics simulations are conducted to determine the stiffness for each spring. Then, multi-physics simulations are conducted to determine the stiffness for the combined system. Finally, a base excitation is simulated to determine the displacement transmissibility of the isolator.

Experimental studies are conducted to investigate the effectiveness of the isolator. First the natural frequencies of the isolator are determined for different configurations. Then the relationship of the displacement transmissibility vs. the exciting frequency is measured against different setups. Finally, the on-line tuning capability of the isolator is investigated.

The proposed electromagnetic isolator can be used as an active isolator by properly choosing the polarities of the two electromagnets. To control such a system that is highly nonlinear and has unknown parameters, an adaptive fuzzy-neural network (FNN) control algorithm is used. First some background information is reviewed. Then the adaptive FNN controller design procedure is presented. Finally, numerical simulations are employed to compare the adaptive FNN controller with the three other controllers:​ Proportional-Plus-Derivative, Backstepping, and Adaptive Backstepping.