The technology associated with active magnetic bearings has been widely used in the last years and can be considered as being one of the most promising solutions for several applications in rotating machinery. Lubricants are not necessary, and high rotation speeds are reached without any relevant heating. Active magnetic bearings are classified as mechatronic systems because they are composed of mechanical and electronic parts that are controlled by using dedicated software. In this context, the present work is devoted to the design of robust controllers applied to supercritical rotors supported by active magnetic bearings. For this aim, numerical and experimental tests were carried out. Different from previous studies reported in the literature, the present contribution proposes a novel design procedure to robustify the neuro-fuzzy controller of a rotor supported by active magnetic bearings based on optimal robust design. This optimal design procedure tunes the robust neuro-fuzzy controller taking into account the optimal balance between vibration attenuation performance and robustness, that is the increase in vibration attenuation implies the reduction in the robustness. The first stage of the controller synthesis is dedicated to the specification of all design requirements. Then, the adaptative neuro-fuzzy controller was obtained, starting from the determination of the plant dominant poles and finally performing the model-based analysis of the system stability and performance. Finally, the vibration control performance and robustness are optimally balanced by using a robust optimization procedure. The behavior of the controller was evaluated by investigating the unbalance response of the rotating system. The obtained results demonstrated the effectiveness of the conveyed approach.

与主动磁轴承相关的技术在最近几年已被广泛使用，并且可以认为是旋转机械中多种应用中最有希望的解决方案之一。不需要润滑剂，并且无需任何相关加热即可达到高转速。有源电磁轴承被归类为机电系统，因为它们由机械和电子零件组成，这些零件通过使用专用软件进行控制。在这种情况下，本工作致力于设计适用于由主动磁轴承支撑的超临界转子的鲁棒控制器。为此，进行了数值和实验测试。与文献报道的先前研究不同，本贡献提出了一种新颖的设计程序，以基于最佳鲁棒性设计来使由主动磁轴承支撑的转子的神经模糊控制器鲁棒。考虑到振动衰减性能和鲁棒性之间的最佳平衡，该最佳设计过程会调整鲁棒神经模糊控制器，也就是说，振动衰减的增加意味着鲁棒性的降低。控制器综合的第一阶段专用于所有设计要求的规范。然后，从确定植物优势极点开始，最后对系统的稳定性和性能进行基于模型的分析，从而获得了自适应神经模糊控制器。最后，振动控制性能和鲁棒性通过使用鲁棒的优化过程达到了最佳平衡。通过研究旋转系统的不平衡响应来评估控制器的行为。获得的结果证明了这种方法的有效性。