Robust Control for Five–Degree–of-Freedom Active Magnetic Bearing Systems

• Main Authors: GIAP, VAN-NAM, 甲文南
• Other Authors: SU, TE-JEN
• Format: Others
• Language: en_US
• Published: 2017
• Online Access: http://ndltd.ncl.edu.tw/handle/g9fa8d

碩士 === 國立高雄應用科技大學 === 電子工程系 === 105 === In recent decades, the advantages of the active magnetic bearing systems (AMBs) has been successfully, being researched and developed in many control laboratories and research institutes. This master thesis concentrates on design control methodology to reach desired goals. We have two models, one axial such levity active magnet devices, and five axials with many practical advanced devices. With five-degree-of-freedom (Five-DOF) have two cases. Case 1: Sensor are installed at locations between the center of gravity of system and left/right radial active magnetic bearings (RAMBs), and case 2: sensors are installed at locations between tips of the system and bearings. A complete state-space modeling of the Five-DOF AMBs with various disturbances has been presented in this study as a challenge benchmark problem. We starts with the suspension of the active magnetic bearing (sAMB) systems. The first step we propose proportional-integral-derivative-surface sliding mode control (PID-s SMC) to construct control system. Chattering occurs by every time periods, then this study use saturation function to replaced signum function for more steady-state, furthermore we study employing PID controller and a Fuzzy fractional-order controller to reduce chattering phenomenon. Then controller is so-called Fuzzy-based PID-s SMC (FPIDSMC) method. Continuous with case 1 of Five-DOF AMBs still using PID controller and PID-s SMC. This system is high unstable model, with alternative from channels. For more steady-state, we propose neural network to observe disturbances and approximate uncertainties and reduce chattering. Herein we will see Grossberg network, then the controller is so-called robust observer-based neural network PID and SMC (RONNPIDSMC). With case 2 of Five-DOF AMBs we propose another method, there is proportional plus integral observer based on the linear quadratic digital tracker. All of the control methods perform well track.