Zhong, Wan (2015-08). Multiple Objective Optimization Design of the Magnetic Bearing Supported Rotordynamic System. Doctoral Dissertation.
Magnetic bearings have been widely used in turbomachinery field due to their advantages from the non-contact mechanism and the absence of a lubrication requirement. This study focuses on modeling and predicting the rotordynamic and thermal performances of the nonlinear active magnetic bearing supported rotordynamic system with flexible rotor and flexible foundation effects, and then optimizing the design of the magnetic bearing actuator and the control law to achieve the multiple goals simultaneously. Nonlinearities, including the nonlinear magnetic bearing force with respect to the rotor displacement and the control current, the flux saturation, and the power amplifier current and voltage saturation, are analyzed to improve the prediction of the rotordynamic system. Two-dimensional finite element method is used to determine the temperature distribution on the actuator and predict the hot spot temperature during rotor steady state operations. A multiple-input multiple-output (MIMO) flexible support model effects on rotordynamic behavior of the system are addressed. Multiple system properties and performances, like the bearing actuator mass, the maximum vibration amplitude, the power loss, and the external static load, are set as goals to be optimized. Genetic algorithms, including Non-dominated Sorting Genetic Algorithm-II (NSGA-II) and Neighborhood Cultivation Genetic Algorithm (NCGA), are selected as the main optimization strategies due to their advantages in solving complicated optimization problems.